WO2010123043A1 - 癌の評価方法 - Google Patents
癌の評価方法 Download PDFInfo
- Publication number
- WO2010123043A1 WO2010123043A1 PCT/JP2010/057102 JP2010057102W WO2010123043A1 WO 2010123043 A1 WO2010123043 A1 WO 2010123043A1 JP 2010057102 W JP2010057102 W JP 2010057102W WO 2010123043 A1 WO2010123043 A1 WO 2010123043A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- feature
- cancer
- mirna
- cancer marker
- Prior art date
Links
Images
Classifications
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
- C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
- C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
- C12Q1/6876—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes
- C12Q1/6883—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material
- C12Q1/6886—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material for cancer
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
- C12Q2600/00—Oligonucleotides characterized by their use
- C12Q2600/112—Disease subtyping, staging or classification
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
- C12Q2600/00—Oligonucleotides characterized by their use
- C12Q2600/118—Prognosis of disease development
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
- C12Q2600/00—Oligonucleotides characterized by their use
- C12Q2600/158—Expression markers
-
- C—CHEMISTRY; METALLURGY
- C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
- C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
- C12Q2600/00—Oligonucleotides characterized by their use
- C12Q2600/178—Oligonucleotides characterized by their use miRNA, siRNA or ncRNA
Definitions
- the present invention relates to a method for evaluating the possibility of cancer by detecting a new cancer marker.
- the cancer marker is also referred to as a tumor marker.
- Specific examples of the cancer marker include, for example, PSA (prostate specific antigen: Prostate Specific Antigen), CEA (Carcinoembryonic Antigen), CA19-9 (Carcinoembryonic Antigen 19-9) 72-4).
- PSA prostate specific antigen: Prostate Specific Antigen
- CEA Carcinoembryonic Antigen
- CA19-9 Carcinoembryonic Antigen 19-9) 72-4.
- Non-Patent Documents 1 and 2 the expression of miRNAs such as has-mir-15, has-mir-16, miR-143, and miR-145 is down-regulated in lymphocytic leukemia and colon cancer.
- an object of the present invention is to provide an evaluation method using a novel cancer marker for evaluating cancer, and an evaluation reagent used therefor.
- the evaluation method of the present invention is an evaluation method for evaluating the possibility of cancer, A cancer marker detection step of detecting a cancer marker in a sample; And evaluating the possibility of cancer of the sample based on the expression level of the cancer marker detected in the cancer marker detection step,
- the sample is a cell or tissue;
- the cancer marker includes at least one miRNA of hsa-miR-92 and hsa-miR-494.
- the present inventors have found that the expression levels of hsa-miR-92 and hsa-miR-494 are changed in cells or tissues as cancer develops, and the present invention has been reached. .
- the evaluation method of the present invention by detecting the expression level of the miRNA in a sample, for example, the presence or absence of cancer or the progression of cancer can be determined with excellent reliability.
- the expression of the miRNA has a significant difference between negative and positive for canceration, for example, according to the present invention, for example, for an early stage cancer that is difficult to detect by general palpation or the like. Can be easily detected.
- the evaluation method of the present invention correspond to, for example, conventional evaluation of cancer by HE staining or the like, it is possible to evaluate cancer with even higher reliability.
- FIG. 1B of this invention It is a block diagram which shows the cancer pathological image diagnosis assistance system of Embodiment 1B of this invention. It is a figure explaining the content of the dyeing
- FIG. 20 is a block diagram illustrating a cancer pathological image diagnosis support system according to Embodiment 1A of the present invention.
- FIG. 20 is a flowchart for demonstrating an example of operation
- each term means the following.
- Cancer generally refers to a malignant tumor.
- Carcinogenesis generally means the onset of cancer and includes the meaning of “malignant transformation”.
- Onset is referred to as onset when, for example, a specific disease is diagnosed by comprehensive judgment based on disease-specific clinical symptoms or test data.
- the “preclinical phase” generally refers to a pre-onset state before a disease-specific clinical symptom appears, and an early state in which a minute amount of malignant tumor cells are already present.
- Prognosis means, for example, a course after treatment of a disease such as after surgery.
- the cancer marker in the present invention can be referred to as “prognostic factor” because it can be used as a judgment material for predicting prognosis, making a prospect, and selecting an appropriate treatment method, for example.
- the “advanced stage of cancer” can be determined as appropriate depending on, for example, the type of cancerous tissue, etc., and generally, stage 0 and stage I can be classified as early stage cancer, stage II as early stage cancer, and stage III to IV as advanced cancer.
- “possibility of cancer” means, for example, the possibility of developing cancer, whether or not it has become cancerous, the degree of progression of cancer in the preclinical stage or clinical stage, or the prognostic state, etc. including.
- the cancer marker miRNA in the present invention is at least one miRNA of hsa-miR-92 and hsa-miR-494.
- the cancer marker is also referred to as a cancer marker miRNA.
- the cancer marker miRNA may be, for example, a single strand (monomer) or a double strand (dimer).
- the cancer marker miRNA may be, for example, an immature miRNA or a mature miRNA.
- the immature miRNA include primary transcription initial miRNA (pri-miRNA) and precursor miRNA (pre-miRNA).
- pri-miRNA primary transcription initial miRNA
- pre-miRNA precursor miRNA
- the pri-miRNA has a hairpin loop structure by intramolecular bonding.
- the pri-miRNA is cleaved with Drosha and converted into a short pre-miRNA having a stem-loop structure.
- the pre-miRNA is also referred to as a stem-loop miRNA.
- the pre-miRNA is cleaved with Dicer to generate shorter double-stranded RNA (miRNA-miRNA * ).
- This double-stranded RNA is unwound on the RISC, and two single-stranded RNAs are generated.
- the single-stranded RNA is a mature miRNA.
- one of mature miRNAs is referred to as functional miRNA, and the other is referred to as Minor miRNA * .
- the cancer marker miRNA is not particularly limited, but is preferably a stem-loop miRNA and a mature miRNA, and particularly preferably a mature miRNA.
- hsa-miR-92 examples include hsa-miR-92a and hsa-miR-92b.
- Hsa-miR-92a may be either immature miRNA or mature miRNA as described above.
- the former is also referred to as immature hsa-miR-92a
- the latter is also referred to as mature hsa-miR-92a.
- Examples of immature hsa-miR-92a include hsa-miR-92a-1 and hsa-miR-92a-2, and any of them may be used.
- hsa-miR-92a-1 and hsa-miR-92a-2 are stem-loop miRNAs.
- the former is also referred to as stem loop hsa-miR-92a-1
- the latter is also referred to as stem loop hsa-miR-92a-2.
- Both are transcripts from different genomic regions, but the mature sequences of each are identical.
- the stem loop hsa-miR-92a-1 for example, has the sequence of Accession No.
- the stem loop hsa-miR-92a-2 has a sequence of Accession No. Registered in MI00000094.
- the mature hsa-miR-92a includes, for example, a functional miRNA. Registered in MIMAT000098.
- the sequence of this functional hsa-miR-92a is shown in SEQ ID NO: 1.
- Functional hsa-miR-92a (SEQ ID NO: 1) 5'-uauugcacuuguccccgcccugu-3 '
- Minor miRNA * can be mentioned.
- Examples of the Minor miRNA * include hsa-miR-92a-1 * and hsa-miR-92a-2 * .
- hsa-miR-92a-1 * has a sequence of Accession No. It is registered in MIMAT0004507. This sequence is shown in SEQ ID NO: 6.
- the sequence of hsa-miR-92a-2 * is registered in MIMAT0004508. This sequence is shown in SEQ ID NO: 7.
- Hsa-miR-92b is a transcription product from a genomic region different from hsa-miR-92a, but its seed sequence is similar to hsa-miR-92a. Therefore, hsa-miR-92b can be used as a cancer marker in the same manner as hsa-miR-92a. As described above, hsa-miR-92b may be either an immature miRNA or a mature miRNA. Hereinafter, the former is also referred to as immature hsa-miR-92b, and the latter is also referred to as mature hsa-miR-92b.
- the stem loop miRNA is hereinafter also referred to as stem loop hsa-miR-92b.
- the stem loop hsa-miR-92b has, for example, a sequence whose accession Mo. Registered in MI0003560.
- the mature hsa-miR-92b includes, for example, a functional miRNA, and its sequence is indicated by the accession no. Registered in MIMAT0003218.
- the sequence of this functional hsa-miR-92b is shown in SEQ ID NO: 3.
- Functional hsa-miR-92b (SEQ ID NO: 3) 5'-uauugcacucucccccggccuccc-3 '
- examples of the mature hsa-miR-92b include Minor miRNA * .
- examples of the Minor miRNA * include hsa-miR-92b * .
- hsa-miR-92b * is a transcript from a different genomic region than hsa-miR-92a-1 * or hsa-miR-92a-2 * , but its seed sequence is hsa-miR-92a-1 * Or similar to hsa-miR-92a-2 * .
- hsa-miR-92b * can be used as a cancer marker in the same manner as hsa-miR-92a-1 * or hsa-miR-92a-2 * .
- hsa-miR-92b * has a sequence of Accession No. It is registered in MIMAT0004792. This sequence is shown in SEQ ID NO: 4.
- Minor hsa-miR-92b * (SEQ ID NO: 4) 5′-agggacggggacggcggugcagug-3 ′
- Hsa-miR-494 may be either an immature miRNA or a mature miRNA as described above. Hereinafter, the former is also referred to as immature hsa-miR-494, and the latter is also referred to as mature hsa-miR-494.
- Examples of immature hsa-miR-494 include stem-loop miRNA. Hereinafter, this is also referred to as stem loop hsa-miR-494.
- the stem loop hsa-miR-494 has, for example, the sequence of Accession No. Registered in MI0003134.
- the mature type hsa-miR-494 includes, for example, a functional miRNA. It is registered in MIMAT0002816. The sequence of this functional hsa-miR-494 is shown in SEQ ID NO: 2. Functional hsa-miR-494 (SEQ ID NO: 2) 5'-ugaaacauaacggggagaaccuc-3 '
- each of the miRNAs has several variations, for example, at the 5 ′ end and the 3 ′ end. Therefore, each miRNA in the present invention further includes a variant having a different number of bases by several from the mature sequence described above.
- Wu H. et al. 2007, PLoS ONE 2 (10): e1020 miRNA profiling of naive, effector and memory CD8 T cells.
- Pablo Landgraf et al. 2007, Cell, vol. 129, p. 1401-1414 A Mammalian microRNA Expression Atlas Based on Small RNA Library Sequencing.
- Neilson et al. 2007, Genes Dev, vol. 21, p. 578-589 Dynamic regulation of miRNA expression in order to stage of cellular development.
- the cancer marker miRNA includes, for example, a polynucleotide comprising a base sequence having homology with the base sequence described in each of the above SEQ ID NOs, or a polynucleotide comprising a base sequence complementary thereto.
- the “homology” is the degree of identity when sequences to be compared are appropriately aligned, and means the occurrence rate (%) of an exact match of nucleotides between the sequences.
- the phrase “having homology” in the nucleotide sequences of polynucleotides means that the polynucleotides are sufficiently similar to maintain the function as miRNA in the present invention.
- the alignment can be performed by using an arbitrary algorithm such as BLAST, for example.
- the base sequences Even if the base sequences have differences due to point mutations such as substitution, deletion, insertion or addition, if they do not affect the function of the miRNA, they can be said to be homologous.
- the number of different bases is, for example, 1 to 20, 1 to 15, 1 to 10, 1 to 5, 1 to 3, 2, or 1. Further, by comparing two polynucleotides, for example, the base sequence is 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99 If they show% or 100% identity, they are said to be homologous.
- both can be said to be homologous.
- the stringent conditions are not particularly limited. For example, in a solution containing 6 ⁇ SSC, 0.5% SDS, 5 ⁇ Denhardt, 0.01% denatured salmon sperm nucleic acid, “Tm (° C.) ⁇ 25 ° C. And the like, and the like.
- the evaluation method of the present invention includes a cancer marker detection step for detecting a cancer marker in a sample, and the possibility of cancer in the sample based on the expression level of the cancer marker detected in the cancer marker detection step.
- the sample is a cell or tissue
- the cancer marker comprises at least one miRNA of hsa-miR-92 and hsa-miR-494.
- the evaluation method of the present invention can be said to be an evaluation method for determining whether or not the patient is suffering from cancer, for example.
- the cancer marker in the present invention is as described above.
- the cancer marker to be detected may be, for example, either or both of hsa-miR-92 and hsa-miR-494.
- hsa-miR-92 When the cancer marker to be detected is hsa-miR-92, for example, either one of mature type and immature type may be used, or both may be used. As for hsa-miR-92, for example, the type of mature type and immature type is not limited, and any one type, two types, or all may be used.
- the cancer marker to be detected is mature type hsa-miR-92a, for example, any one of functional hsa-miR-92, Minor hsa-miR-92a-1 * and Minor hsa-miR-92a-2 * It may be any two or all of them.
- the cancer marker to be detected is immature type hsa-miR-92a, for example, one or both of stem loop hsa-miR-92a-1 and stem loop hsa-miR-92a-2 may be used. May be.
- the cancer marker to be detected is mature type hsa-miR-92b, for example, either one or both of functional hsa-miR-92b and Minor hsa-miR-92b * may be used.
- hsa-miR-494 When the cancer marker to be detected is hsa-miR-494, for example, either one of mature type and immature type may be used, or both may be used.
- the type of hsa-miR-494 is not limited, and the type of mature type and immature type may be any one type or two types.
- functional hsa-miR-494 can be mentioned.
- the present invention is characterized by detecting the expression level of the cancer marker.
- the detection method of the said cancer marker is not restrict
- the detection method is preferably, for example, a method of visualizing the cancer marker.
- the visualization method is not particularly limited, but is preferably performed by, for example, color development, fluorescence, autoradiography, or the like.
- the cancer marker can be detected by visual observation, absorbance measurement, image processing, or the like.
- the cancer marker when the cancer marker is visualized by fluorescence, the cancer marker can be detected by visual observation, fluorescence intensity measurement, image processing, or the like.
- cancer marker miRNA staining visualization of cancer marker miRNA by color development or fluorescence is also referred to as miRNA staining.
- the cancer marker miRNA can be detected by, for example, visual observation of an autoradiographic image, image processing of the image, or the like.
- the cancer to be evaluated is not particularly limited.
- the cancer include colon cancer, rectal cancer, gallbladder cancer, gastric cancer, breast cancer, leukemia, pancreatic cancer, liver cancer, brain tumor, osteosarcoma and the like as described above.
- the sample is a biological sample and may be any cell or tissue, and is not particularly limited. Specific examples include, for example, the large intestine, rectum, gallbladder, stomach, breast, blood cells, liver, brain, bone, bone and surrounding tissues such as cells, blood cells such as leukocytes, and the like.
- the subject from which the sample is derived is not particularly limited, and examples thereof include humans.
- Other examples include non-human mammals including non-human primates, rodents, dogs, cats and the like.
- the detection of the cancer marker miRNA in the sample may be performed directly on the sample, for example, or indirectly on RNA recovered from the sample.
- the detection of the cancer marker miRNA in the sample may be performed directly on the sample, for example, or indirectly on RNA recovered from the sample.
- the present invention for example, since the expression region of the cancer marker in the sample can be specified, it is preferable to directly detect the cancer marker for the sample.
- the method for recovering RNA from the sample is not particularly limited, and a known method can be employed. Specific examples include guanidine-cesium chloride ultracentrifugation, AGPC (Acid Guanidinium-Phenol-Chloroform), and commercially available reagents and kits.
- the cancer marker miRNA can be detected indirectly by detecting the cDNA of the cancer marker miRNA for cDNA synthesized using the recovered RNA as a template.
- the cancer marker miRNA can be detected using, for example, a nucleic acid amplification method.
- the nucleic acid amplification method is not particularly limited, and examples thereof include a polymerase chain reaction (PCR) method, a reverse transcription PCR (RT-PCR) method, a real-time PCR method, and a real-time RT-PCR method.
- PCR polymerase chain reaction
- RT-PCR reverse transcription PCR
- real-time PCR method a real-time RT-PCR method.
- a real-time RT-PCR method is preferred.
- RNA is extracted from a sample, and cDNA is synthesized using the total RNA as a template and a random primer.
- an amplification reaction is performed using a primer capable of amplifying the cDNA of the target cancer marker miRNA, and the amplification product is detected.
- the expression level of the cancer marker miRNA in the sample that is, the presence or amount of expression of the cancer marker miRNA can be detected.
- the random primer used in the cDNA synthesis reaction is not particularly limited, and for example, a commercially available random primer can be used. Further, the primer used in the amplification reaction is not limited at all.
- the primer hybridizes to the cDNA of the cancer marker miRNA or a sequence complementary thereto, or the cDNA in the peripheral region of the cancer marker miRNA or a sequence complementary thereto. Possible primers are listed.
- the primer can be appropriately designed based on, for example, the base sequence of the cancer marker miRNA and common technical knowledge. Specific examples of the primer include a primer comprising a cDNA of the target cancer marker miRNA or a sequence complementary thereto, or a cDNA in the peripheral region of the cancer marker miRNA or a sequence complementary thereto.
- the sequence of the primer is, for example, about 70% or more complementary to the cDNA of the target cancer marker miRNA or a sequence complementary thereto, or the cDNA in the peripheral region of the cancer marker miRNA or a sequence complementary thereto.
- it is 80% or more, more preferably 90% or more, still more preferably 95% or more, and particularly preferably 100% complementary.
- the structural unit of the primer is not particularly limited, and a known structural unit can be adopted. Specific examples include nucleotides such as deoxyribonucleotides and ribonucleotides, and may include PNA (Peptide Nucleic Acid), LNA (Locked Nucleic Acid), and the like.
- the base in the structural unit is not particularly limited, and may include natural bases (non-artificial bases) such as A, C, G, T, and U, and may include non-natural bases (artificial bases).
- the length of the primer is not particularly limited and can be a general length.
- the detection method of the amplification product is not particularly limited, and a known method can be adopted.
- a fluorescent reagent coexists in the reaction solution of the amplification reaction.
- the fluorescent reagent include fluorescent substances that specifically bind to double-stranded nucleic acids, fluorescent substances that intercalate with double-stranded nucleic acids, and the like.
- the coexisting fluorescent substance binds to or intercalates with the double-stranded nucleic acid.
- the presence or absence of the amplification product can be confirmed, and the presence or absence of the target cancer marker miRNA can be indirectly confirmed.
- the amplification product can be quantified by measuring the fluorescence intensity of the fluorescent substance, and the target cancer marker miRNA can be quantified indirectly.
- the fluorescent reagent include SYBR (trademark) Green. Detection using the fluorescent reagent can be performed by, for example, a known method.
- a commercially available reagent such as SYBR (trademark) Green PCR master mix (trade name, Perkin-Elmer Applied Biosystems), ABI Prism 7900 Sequence Detection System (trade name, Perkin- A commercially available detection device such as Elmer Applied Biosystems) can be used, and can be carried out according to their manual.
- SYBR trademark
- Green PCR master mix trade name, Perkin-Elmer Applied Biosystems
- ABI Prism 7900 Sequence Detection System trade name, Perkin- A commercially available detection device such as Elmer Applied Biosystems
- the labeled probe examples include a probe having a fluorescent substance and a quencher, and specific examples include a TaqMan (trademark) probe and a cycling probe using RNase in combination.
- a TaqMan (trademark) probe examples include a TaqMan (trademark) probe and a cycling probe using RNase in combination.
- a method using such a labeled probe can be performed, for example, according to a known method.
- the cancer marker miRNA cDNA can be detected using a probe.
- the probe include a primer capable of hybridizing to the cDNA of the cancer marker miRNA or a sequence complementary thereto.
- the cancer marker cDNA and the probe are hybridized, and the probe hybridized to the cancer marker cDNA is detected.
- the presence or absence or amount of the probe hybridized to cDNA of the cancer marker corresponds to the presence or amount of the cancer marker miRNA in the RNA recovered from the sample. Therefore, the presence or amount of the cancer marker miRNA in the sample can be indirectly detected by detecting the probe.
- the probe can be detected by a known method, for example, the same method as described later.
- the cancer marker miRNA may be directly detected from the recovered RNA.
- a hybridization method using a probe can be mentioned.
- the probe for example, a probe that can specifically hybridize to the cancer marker miRNA can be used.
- the cancer marker miRNA and the probe are hybridized, and the probe hybridized to the cancer marker miRNA is detected.
- the presence / absence or amount of the probe hybridized to the cancer marker corresponds to the presence / absence or amount of the cancer marker miRNA in RNA recovered from the sample. Therefore, the presence or amount of the cancer marker miRNA in the sample can be indirectly detected by detecting the probe.
- the method for detecting the probe is not particularly limited.
- a labeled probe labeled with a labeling substance can be used as the probe, and the labeling substance can be detected.
- the labeling substance for example, a substance that can be detected by itself can be used, and examples thereof include a coloring substance, a fluorescent substance that emits fluorescence, and a radioactive substance.
- the presence and amount of the coloring material can be determined by, for example, the presence or absence of coloring and the strength of coloring.
- the color-developing substance may be, for example, a substance that exhibits color development, or may be a substance that liberates a substance that exhibits color development by an enzyme reaction or the like, or changes to a substance that exhibits color development by an enzyme reaction or an electron transfer reaction. It may be a substance.
- the presence and amount of the fluorescent substance can be determined based on the presence or absence of fluorescence and the intensity of fluorescence.
- the fluorescent substance may be, for example, a substance that emits fluorescence itself, a substance that liberates a substance that emits fluorescence by an enzyme reaction or the like, or a substance that emits fluorescence by an enzyme reaction or an electron transfer reaction. It may be a substance.
- the presence and amount of the labeling substance can be determined by measuring the radioactivity level with a scintillation counter or the presence or absence of an image and the density of the image by an autoradiography method, for example. Examples of the hybridization method using the probe labeled with the radioactive substance include Northern blotting and microarray analysis.
- the labeling substance may be a labeling substance that can be detected by other reagents, for example.
- a labeling substance include enzymes such as alkaline phosphatase (AP) and horseradish peroxidase (HRP).
- AP alkaline phosphatase
- HRP horseradish peroxidase
- the labeling substance is an enzyme, for example, the presence or absence of color development or fluorescence due to the reaction with the enzyme by adding, as the other reagent, a substrate that emits color or fluorescence by the enzyme reaction or electron transfer accompanying it, etc. Absorbance or fluorescence intensity may be detected.
- the substrate is not particularly limited and can be appropriately set depending on the type of enzyme.
- BCIP bromochloroindolyl phosphate
- NBT nilotu blue tetrazolium
- DAB Benzidine tetrahydrochloride
- the labeling substance that can be detected by the other reagents include biotin and avidin, and preferably biotin.
- biotin and avidin When the probe is labeled with biotin, it is preferable to add, for example, the enzyme, the coloring material, the fluorescent material, or the radioactive material, to which avidin is bound, as the other reagent. Since biotin, which is a labeling substance for the probe, binds to avidin, the enzyme or the like bound to avidin may be detected by the method described above.
- the other reagent may be, for example, a complex of avidin, biotin, the enzyme, or the like, so-called avidin-biotin complex. This method is a so-called ABC (avidin-biotin complex) method.
- biotin of one complex and biotin of another complex can be bound, and the number of molecules such as an enzyme that binds to one molecule of probe can be increased. For this reason, detection with higher sensitivity is possible.
- the biotin may be a biotin derivative, for example, and the avidin may be an avidin derivative such as streptavidin, for example.
- examples of the method for detecting the probe hybridized to the cancer marker miRNA include a method using an antigen-antibody reaction using a labeled probe labeled with an antigen as the probe.
- a labeled primary antibody that can specifically bind to the antigen and is labeled with a labeling substance.
- other methods include a primary antibody that can specifically bind to the antigen and a labeled secondary antibody that can specifically bind to the primary antibody and is labeled with a labeling substance.
- the cancer marker miRNA and the antigen-labeled probe are hybridized. Subsequently, the labeled primary antibody is bound to the antigen-labeled probe bound to the cancer marker miRNA via the antigen. Then, the labeling substance of the labeled primary antibody bound to the probe is detected. Thereby, the probe hybridized to the cancer marker miRNA can be detected, and as a result, the cancer marker miRNA can be indirectly detected.
- the type of antigen for labeling the probe is not particularly limited, and examples thereof include digoxigenin (DIG).
- the primary antibody is not particularly limited, and can be appropriately set according to, for example, the type of the antigen. When the antigen is DIG, for example, an anti-DIG antibody can be used.
- the labeling substance of the labeled primary antibody is not particularly limited and is the same as described above.
- the latter method using a labeled secondary antibody is a so-called sandwich method.
- the cancer marker miRNA and the antigen-labeled probe are first hybridized.
- the primary antibody is bound to the antigen-labeled probe bound to the cancer marker miRNA via the antigen.
- the labeled secondary antibody is bound to the primary antibody bound to the probe.
- the labeled secondary antibody binds to the cancer marker miRNA via the primary antibody.
- the labeling substance of the labeled secondary antibody is detected.
- the kind of antigen for labeling the probe and the primary antibody are not particularly limited, and are the same as described above.
- the labeling substance of the secondary antibody is not particularly limited and is the same as described above.
- the detection of the cancer marker miRNA is preferably performed directly on the sample because, for example, the expression region of the cancer marker miRNA in the sample can be identified.
- the sample is preferably immobilized.
- the method for detecting the cancer marker miRNA is preferably, for example, a hybridization method using a probe.
- the cancer marker is preferably detected by an in situ hybridization method for an immobilized sample, and in particular, immunohistochemistry (IHC) using an antigen-antibody reaction is preferably applied.
- IHC immunohistochemistry
- the cytoplasm or nucleus can be stained with the cancer marker miRNA.
- a labeled probe is preferable as the probe, and the same probes as described above can be used.
- the in situ hybridization method can be performed, for example, according to a known method. Moreover, it can also carry out according to the instruction manual using a commercially available kit etc. Examples of the kit include a Ventana brand name RiboMap in situ hybridization kit.
- in situ hybridization method for example, preparation of a section slide of the sample, pretreatment of the section slide, hybridization of the labeled probe, and detection of a hybridization signal are performed. Specific examples of the in situ hybridization method are shown below, but the present invention is not limited thereto.
- a section slide is prepared for the cell or tissue as the sample.
- the section slide can be prepared by fixing the sample using a fixing solution, embedding, cutting the sample to a desired thickness, and placing the sample on the slide.
- the fixing solution include cross-linking agents such as formaldehyde and paraformaldehyde; PLP (periodate-lysine-paraformaldehyde); Zamboni solution; glutaraldehyde; coagulating precipitates such as ethanol and acetone.
- the immobilization method may be any production method such as an unfixed frozen section, a fixed frozen section, and a fixed paraffin embedded section.
- the immobilization conditions are not particularly limited, but for example, it is preferably performed at room temperature for 1 hour or more, more preferably 6 hours or more.
- the section slide is pretreated.
- the pretreatment include deparaffinization treatment, rehydration, re-immobilization, acid treatment, and proteinase K treatment for improving the permeability of the labeled probe.
- a neutralization treatment of a positive charge may be performed.
- the labeled probe is added to the pretreated slice slide, hybridization is performed, and a hybridization signal corresponding to the labeled substance of the labeled probe is detected.
- the amount of the labeled probe to be added is not particularly limited, and can be appropriately set depending on, for example, the type of the labeling substance, the labeling ratio in the entire probe used, and the like.
- a specific example is, for example, 1 to 1000 ng per general slide used in the in situ hybridization method, but is not limited thereto.
- Hybridization conditions are not particularly limited.
- the heat denaturation treatment before hybridization is preferably a treatment temperature of 50 to 100 ° C. and a treatment time of 1 to 60 minutes, more preferably a treatment temperature of 60 to 95 ° C. and a treatment time of 5 to 10 minutes.
- the hybridization is preferably performed at a treatment temperature of 40 to 80 ° C. and a treatment time of 1 to 36 hours, more preferably at a treatment temperature of 45 to 70 ° C. and a treatment time of 4 to 24 hours.
- the signal detection method is not particularly limited, and can be appropriately determined according to, for example, the type of the labeled substance of the labeled probe, the labeled primary antibody, or the labeled secondary antibody, as described above.
- color development or fluorescence for example, the presence or absence or amount of the cancer marker miRNA in the sample can be detected from the presence or absence of color development or fluorescence, or the intensity of color development or intensity of fluorescence.
- the color development or fluorescence may be determined visually, for example, or may be determined by image processing.
- the presence or amount of the cancer marker miRNA in the sample can be detected from the presence or absence of an image in an autoradiography image or the density of the image using an autoradiography method.
- the presence / absence of the image or the density of the image may be determined visually, for example, or may be determined by image processing.
- the image processing is not particularly limited, and can be performed using a known system or software.
- the sequence of the probe is not particularly limited.
- the probe include probes that can specifically bind to any of the aforementioned cancer marker miRNAs.
- a commercially available product may be used, or it may be prepared in-house.
- the sequence of the probe can be appropriately designed based on, for example, the base sequence of the aforementioned cancer marker miRNA and common technical knowledge.
- Specific examples include a probe composed of a sequence complementary to a cancer marker miRNA for detection or a probe containing the complementary sequence.
- the probe sequence is preferably about 70% or more complementary, more preferably 90% or more, and particularly preferably 100% complementary to the target cancer marker miRNA. .
- the structural unit of the probe is not particularly limited, and for example, a known structural unit can be adopted. Specific examples include nucleotides such as deoxyribonucleotides and ribonucleotides, and structural units such as PNA and LNA. Examples of the LNA include BNA (Bridged Nucleic Acid) such as 2 ′, 4′-Bridged Nucleic Acid.
- the base in the structural unit is not particularly limited, and may be a natural base such as adenine, guanine, cytosine, thymine, uracil, or an unnatural base.
- the length of the probe is not particularly limited. For example, the length is 10 to 100 bases, and preferably 15 to 40 bases.
- the probe include a polynucleotide comprising a complementary sequence of the base sequence represented by SEQ ID NO: 5.
- hsa-miR-92a detection probe SEQ ID NO: 5'-acagggccgggacaagtgcaata-3 '
- the evaluation method of the present invention evaluates the possibility of cancer of the sample based on the expression level of the cancer marker miRNA detected as described above in the evaluation step.
- the expression level of the cancer marker examples include the presence / absence or expression level of the cancer marker miRNA in the sample.
- the expression level may be, for example, the actual miRNA amount or a value correlated with the actual miRNA amount. Examples of the latter include a signal value obtained upon detection of the cancer marker miRNA.
- the signal value can be appropriately determined according to, for example, the miRNA detection method or the type of signal value detection device.
- the detection method uses a nucleic acid amplification method such as a PCR method such as a real-time RT-PCR method
- the detection method can also be represented by a copy number per 1 ⁇ L (copies / ⁇ L) or the like. Further, as will be described later, when an miRNA-stained image obtained by staining miRNA is used, for example, color or fluorescence brightness or saturation corresponds to the signal value.
- a section slide visualizing the cancer marker miRNA and HE (hematoxylin and eosin) staining are performed. It is preferable to prepare a section slide and collate them.
- HE staining and visualization of cancer marker miRNA are performed on section slides of the same sample, and both are collated. Then, by comparing the tumor area determined from HE staining with the visualized positive area of the cancer marker miRNA, the cancer can be determined with higher reliability. Positive means, for example, that the cancer marker is present, and negative means, for example, that the cancer marker is not present or is below the detection limit.
- the method for judging cancer by comparing a section slide visualizing the cancer marker miRNA and a section slide stained with HE is not particularly limited, but can be performed, for example, as follows. First, as described later, the HE staining and miRNA visualization are performed on adjacent section slides. On the other hand, for the HE-stained section slide, the tumor region is determined by, for example, microscopic observation. Then, the HE-stained section slide is compared with the section slide visualizing the cancer marker. As a result, for example, when the tumor region in the HE-stained section slide and the cancer marker positive region in the section slide visualizing the cancer marker overlap, the tumor region and the cancer marker positive region can be determined as cancer.
- the section slide for detecting the cancer marker miRNA and the section slide for HE staining are fixed and embedded. It is preferable that the slices are adjacent slices cut out from the sample. Specifically, continuous slices are preferred. As a result, the two images can be collated more accurately.
- the verification of the visualized cancer marker miRNA and the HE staining is preferably performed by, for example, image verification. That is, it is preferable to prepare a cancer marker visualization image and an HE-stained image for the immobilized sample and collate them.
- the image can be converted into a digital image of the immobilized sample visualized by the cancer marker miRNA and the immobilized sample visualized by the HE staining using a CCD (Charge Coupled Device Image Sensor) or a scanner, for example.
- CCD Charge Coupled Device Image Sensor
- the method for judging cancer by comparing the visualized image of the cancer marker miRNA with the HE-stained image is not particularly limited, but can be performed as follows, for example. First, as described above, the HE staining and visualization of the cancer marker miRNA are performed on adjacent section slides, and an HE staining image and a cancer marker visualization image are prepared. Then, a tumor region is determined for the HE-stained image, and the HE-stained image and the cancer marker visualized image are collated. As a result, when the tumor region in the HE-stained image and the cancer marker miRNA positive region in the cancer marker visualized image overlap, the tumor region and the cancer marker positive region can be determined as cancer.
- region is not confirmed in the said HE dyeing
- the expression levels of the cancer marker miRNAs hsa-miR-92a and hsa-miR-494 increase with canceration, for example, in cells and tissues.
- the cancer marker miRNA for example, significantly increases the expression level after the onset of cancer compared to before the onset of cancer, the expression level of the preclinical stage is significantly higher than before the preclinical stage. Increased, significantly increased expression level in clinical phase compared to pre-clinical phase, significantly increased early level expression level compared to early phase, expressed after initial phase compared to early phase It is understood that the level increases significantly.
- the method may include determining the likelihood of cancer.
- the term “normal person” means, for example, a person who is not determined to have developed the cancer to be evaluated, or a person who has not been determined to have the possibility of developing the cancer. To do.
- “patient” means, for example, a person who is determined to have developed the cancer to be evaluated.
- the subject Determines that the cancer is likely.
- the expression level of the cancer marker miRNA in the sample of the subject is compared with the expression level of the cancer marker miRNA in the sample of the normal subject.
- the examiner determines that the cancer is relatively advanced.
- the expression level of the cancer marker miRNA in the sample of the subject is compared with the expression level of the cancer marker miRNA in the sample of each patient according to the advanced stage, and the subject has a similar expression level. Determined to be in the same progression as the patient shown.
- the expression level of the normal cancer marker miRNA in the above (1) and (2) can be determined using, for example, a sample collected from the normal person.
- the expression level of the cancer marker miRNA of each patient in (3) can be determined, for example, by classifying patients for each advanced stage and using samples collected from patients in each advanced stage.
- the expression levels of normal subjects and patients in the above (1) to (3) may be determined in advance, for example, and need not be determined every time evaluation is performed.
- the normal and patient samples in the above (1) to (3) are preferably, for example, the same type of sample as the subject sample.
- the normal and patient samples are preferably samples prepared by the same method and conditions as the sample of the subject, for example.
- the expression level of the cancer marker miRNA of the normal person or patient may be, for example, the expression level of one normal person or one patient, or the expression level of multiple normal persons or multiple patients Therefore, the expression level calculated by a statistical method may be used.
- the expression level of the subject is higher than the expression level of the normal subject, it can be determined that the subject has a high possibility of cancer.
- the expression level of the subject is equal to or lower than the expression level of the normal subject, it can be determined that the possibility of cancer is low for the subject.
- the possibility of cancer can be said to be, for example, the possibility of becoming cancerous or the possibility of suffering from cancer.
- the subject develops the cancer relatively. Can be judged. Even if the expression level of the subject is higher than the expression level of the normal subject, it can be determined that the smaller the difference is, the less advanced the cancer.
- the expression level is determined for each patient in each advanced stage.
- the significant difference is, for example, T-test (t test), F test, chi-square It can be judged by statistical methods such as tests.
- the evaluation method for example, it can be determined that the possibility of cancer is high with high reliability even for a subject in the preclinical stage of cancer that has been difficult to determine.
- the advanced stage of cancer can be determined with high reliability. Therefore, in the prevention and treatment of cancer, for example, important information for determining a policy such as medication or surgery can be obtained with high reliability.
- the evaluation method of the present invention can evaluate cancer by, for example, calculating the content of staining positive cells.
- the staining positive cell means, for example, a cell having the cancer marker and stained by staining the cancer marker.
- the evaluation method of the present invention acquires a cancer marker stained image in which the cancer marker is stained for the immobilized sample.
- a HE-stained image acquisition step of acquiring a HE-stained image An information acquisition step of acquiring information of a tumor region in the HE-stained image; A matching step of calculating a matching position between the HE stained image acquired in the HE image acquiring step and the cancer marker stained image acquired in the cancer marker detecting step; A specifying step of identifying a tumor region in the cancer marker-stained image based on information on the tumor region in the HE-stained image acquired in the information acquisition step and information on the matching position calculated in the matching step; and It is preferable to include a staining positive cell detection step of detecting staining positive cells in the tumor region based on information on the tumor region in the cancer marker staining image identified in the identification step.
- the staining positive cell detection step may be, for example, a calculation step of calculating a staining positive cell content rate in the tumor region based on information on the tumor region in the cancer marker staining image identified in the identification step.
- the cancer marker-stained image may be, for example, an image obtained by visualizing the cancer marker miRNA by color development or an image visualized by light emission.
- the evaluation of cancer to be compared with the HE-stained image can be performed, for example, by a cancer pathological image diagnosis support method described later.
- This method can be realized, for example, by executing a cancer pathological image diagnosis support system, a cancer pathological image diagnosis support program, or a cancer pathological image diagnosis support device, which will be described later. Details of these will be described later.
- the evaluation reagent of the present invention is an evaluation reagent for use in the evaluation method of the present invention, and is a cancer marker detection reagent of the present invention, that is, hsa-miR-92 and hsa-miR-494.
- a miRNA detection reagent for detecting at least one of the miRNAs is included. According to such an evaluation reagent, the evaluation method of the present invention can be easily carried out.
- the present invention is characterized by detecting at least one of hsa-miR-92 and hsa-miR-494 as a cancer marker miRNA, and the method for detecting these miRNAs is not limited at all.
- the miRNA detection reagent contained in the evaluation reagent of the present invention only needs to be able to detect any of these cancer marker miRNAs.
- the type and composition of the reagent are not limited. Those skilled in the art can set detection reagents for these cancer marker miRNAs based on common general technical knowledge.
- the miRNA detection reagent is not particularly limited, and examples thereof include probes that can hybridize to any of the cancer marker miRNAs as described above.
- the probe may be a labeled probe as described above. Further, depending on the miRNA detection method and the type of labeling substance of the labeling probe, other reagents may be included.
- the evaluation reagent of the present invention may further contain, for example, an enzyme, a buffer solution, a washing solution, a lysis solution, a dispersion solution, a dilution solution, etc. depending on the miRNA detection method.
- the form of the evaluation reagent of the present invention is not particularly limited, and may be, for example, a liquid wet system or a dry dry system.
- the evaluation kit of the present invention is an evaluation kit for use in the evaluation method of the present invention, and is a miRNA detection reagent for detecting at least one miRNA of hsa-miR-92 and hsa-miR-494. It is characterized by including.
- Examples of the miRNA detection reagent include the evaluation reagent of the present invention, and are as described above. According to such an evaluation kit, the evaluation method of the present invention can be easily performed.
- the form of the evaluation kit of the present invention is not particularly limited, and may be, for example, a liquid wet system or a dry system.
- the various reagents are, for example, separate and may be used together at the time of use, or may be mixed before use.
- the evaluation kit of the present invention includes, for example, instructions for use.
- the present invention is a system, program, method and apparatus for supporting diagnosis based on a pathological image of cancer, and includes the following first and second embodiments.
- a first aspect of the present invention is a cancer pathological image diagnosis support system (hereinafter referred to as an image diagnosis support system) that supports diagnosis based on a pathological image of cancer.
- an image acquisition means for acquiring a HE stained image and a cancer marker stained image
- Information acquisition means for acquiring information of a tumor region in the HE-stained image
- a matching unit for calculating a matching position between the HE-stained image and the cancer marker-stained image acquired by the image acquiring unit
- Identification means for identifying a tumor region in the cancer marker-stained image based on information on the tumor region in the HE-stained image acquired by the information acquisition unit and information on the matching position calculated by the matching unit;
- staining positive cell detection means for detecting staining positive cells in the tumor region based on information on the tumor region in the cancer marker stained image identified by the identification unit.
- the staining positive cell detection means may be, for example, a calculation means for calculating a staining positive cell content rate in the tumor region based on information on the tumor region in the cancer marker stained image specified by the specifying means. Good (hereinafter the same).
- the calculating means calculates a staining intensity in addition to the staining positive cell content rate.
- the staining intensity include staining intensity in a tumor region in the cancer marker stained image.
- the image diagnosis support system of the present invention further includes: Input receiving means for receiving input of the specified information of the pathological image to be diagnosed and the specified information of the examination type; A stained image database storing the HE stained image and the cancer marker stained image; The image acquisition means preferably acquires the HE stained image and the cancer marker stained image from the stained image database based on the designation information.
- the pathological image designation information is an image identifier of the HE-stained image
- the image acquisition means acquires the HE stained image having the image identifier and the cancer marker stained image adjacent to the HE stained image from the stained image database.
- the designation information of the pathological image is an image identifier of the cancer marker stained image
- the image acquisition means acquires the cancer marker stained image having the image identifier and the HE stained image adjacent to the cancer marker stained image from the stained image database.
- the specified information of the pathological image is a subject identifier of a subject to be diagnosed
- the image acquisition means preferably acquires the HE stained image having the subject identifier and the cancer marker stained image from the stained image database.
- the information acquisition means preferably acquires information on a tumor region in the HE-stained image from the stained image database.
- the image diagnosis support system of the present invention further includes a tumor region calculation unit that calculates a tumor region in the HE-stained image acquired by the image acquisition unit, It is preferable that the information acquisition unit acquires information on a tumor region in the HE-stained image calculated by the tumor region calculation unit.
- the image diagnosis support system of the present invention further includes: An input receiving means for receiving an input of a slide identifier of a slide to be diagnosed and designation information of an examination type; A slide database in which the slides are stored; A slide acquisition means for acquiring the slide having the slide identifier from the slide database; The image acquisition unit preferably acquires the HE-stained image and the cancer marker-stained image by photographing the slide acquired by the slide acquisition unit.
- the present invention is an image diagnosis support system for supporting diagnosis based on a pathological image of cancer, A terminal and a server, The terminal and the server can be connected via a communication network outside the system,
- the terminal Terminal-side transmission means for transmitting information in the terminal to the server via the communication network;
- Terminal-side receiving means for receiving information transmitted from the server via the communication network,
- the server Server-side transmission means for transmitting information in the server to the terminal via the communication network;
- Server-side receiving means for receiving information transmitted from the terminal via the communication network;
- As the pathological image to be diagnosed an image acquisition means for acquiring a HE stained image and the cancer marker stained image;
- Information acquisition means for acquiring information of a tumor region in the HE-stained image;
- a matching unit for calculating a matching position between the HE-stained image and the cancer marker-stained image acquired by the image acquiring unit;
- Identification means for identifying a tumor region in the cancer marker-stained image based on information on the tumor region in the HE-
- the staining positive cell detection means for example, based on the information on the tumor area in the cancer marker stained image specified by the specifying means, the staining positive cell content rate in the tumor area It may be a calculation means for calculating.
- information on the staining positive cell content rate calculated by the calculation unit of the server is transmitted from the server side transmission unit to the terminal side reception unit.
- the present invention is a server used in the image diagnosis support system of the present invention,
- the server Server-side transmission means for transmitting information in the server to a terminal via the communication network;
- Server-side receiving means for receiving information transmitted from the terminal via the communication network;
- an image acquisition means for acquiring a HE stained image and a cancer marker stained image;
- Information acquisition means for acquiring information of a tumor region in the HE-stained image;
- a matching unit for calculating a matching position between the HE-stained image and the cancer marker-stained image acquired by the image acquiring unit;
- Identification means for identifying a tumor region in the cancer marker-stained image based on information on the tumor region in the HE-stained image acquired by the information acquisition unit and information on the matching position calculated by the matching unit;
- staining positive cell detection means for detecting staining positive cells in the tumor region based on the information on the tumor region in the cancer marker stained image identified by the identifying unit.
- the staining positive cell detection unit calculates, for example, a staining positive cell content rate in the tumor region based on information on the tumor region in the cancer marker staining image identified by the identification unit. It may be a calculation means.
- the present invention is a terminal used in the image diagnosis support system of the present invention, Terminal-side transmission means for transmitting information in the terminal to the server via the communication network; Terminal-side receiving means for receiving information transmitted from the server via the communication network, Information on the pathological image is transmitted from the terminal side transmission unit to the server side reception unit, and information on staining positive cells detected by the staining positive cell detection unit of the server is transmitted from the server side transmission unit. It is transmitted to the terminal side receiving means.
- information on the staining positive cell content rate calculated by the calculation unit of the server may be transmitted from the server side transmission unit to the terminal side reception unit.
- the present invention is a cancer pathological image diagnosis support method (hereinafter referred to as an image diagnosis support method) for supporting diagnosis based on a pathological image of cancer,
- an image acquisition step of acquiring a HE stained image and a cancer marker stained image;
- An information acquisition step of acquiring information of a tumor region in the HE-stained image;
- a matching step of calculating a matching position between the HE-stained image and the cancer marker-stained image acquired in the image acquiring step;
- a specifying step for specifying the tumor region in the cancer marker-stained image;
- a staining positive cell detection step of detecting staining positive cells in the tumor region based on information on the tumor region in the cancer marker staining image identified in the identifying step.
- the staining positive cell detection step includes, for example, a staining positive cell content rate in the tumor region based on information on the tumor region in the cancer marker staining image identified in the identification step. It may be a calculation step of calculating (hereinafter the same).
- the calculation step calculates a staining intensity in addition to the staining positive cell content rate.
- the staining intensity include staining intensity in a tumor region in the cancer marker stained image.
- the image acquisition step acquires the HE-stained image and the cancer marker-stained image based on, for example, designation information of the pathological image to be diagnosed.
- the HE-stained image and the cancer marker-stained image are preferably acquired from, for example, a stained image database in which the HE-stained image and the cancer marker-stained image are stored.
- the pathological image designation information is preferably an image identifier of the HE-stained image.
- the image acquisition step preferably acquires, for example, the HE stained image having the image identifier and the cancer marker stained image adjacent to the HE stained image from the stained image database.
- the pathological image designation information is an image identifier of the cancer marker stained image.
- the image acquisition step preferably acquires, for example, the cancer marker stained image having the image identifier and the HE stained image adjacent to the cancer marker stained image from the stained image database.
- the pathological image designation information is a subject identifier of a person to be diagnosed.
- the said image acquisition process acquires the said HE stained image and the said cancer marker stained image with the said subject identifier from the said stained image database, for example.
- the diagnostic imaging support method of the present invention it is preferable that information on a tumor region in the HE stained image is also stored in the stained image database. And it is preferable that the said information acquisition process acquires the information of the tumor area
- the image diagnosis support method of the present invention preferably further includes a tumor region calculation step of calculating a tumor region in the HE-stained image acquired in the image acquisition step.
- the information acquisition step preferably acquires information on a tumor region in the HE-stained image calculated in the tumor region calculation step.
- the present invention is also a cancer pathological image diagnosis support program (hereinafter referred to as an image diagnosis support program) that supports diagnosis based on a cancer pathological image, and the image diagnosis method of the present invention can be executed by a computer. It is characterized by that.
- the image diagnosis support program of the present invention includes, for example, an image acquisition step of acquiring a HE stained image and a cancer marker stained image as the pathological image to be diagnosed, An information acquisition step of acquiring information of a tumor region in the HE-stained image; A matching step of calculating a matching position between the HE-stained image and the cancer marker-stained image acquired in the image acquiring step; Based on the information on the tumor region in the HE-stained image acquired in the information acquisition step and the information on the matching position calculated in the matching step, a specifying step for specifying the tumor region in the cancer marker-stained image; The computer is caused to execute a staining positive cell detection step of detecting staining positive cells in the tumor region based on information on the tumor region in the cancer marker staining image identified in the identifying step.
- the staining positive cell detection step includes, for example, a staining positive cell content rate in the tumor region based on information on the tumor region in the cancer marker stained image specified in the specifying step. It may be a calculation step for calculating.
- the staining positive cell content rate can be finally obtained as a quantitative value.
- the first embodiment of the present invention will be specifically described by taking Embodiments 1A to 1D as examples.
- the cancer marker stained image is also referred to as “miRNA stained image”. Note that the present invention is not limited to these embodiments.
- FIG. 19 is a block diagram showing an example of the configuration of an image diagnosis support apparatus including the image diagnosis support system of the present invention.
- the diagnostic imaging support apparatus 190 includes a processing unit 191 and a storage unit 192.
- the image diagnosis support apparatus 190 is connected to the CCD 194, the scanner 195, and the display 196 connected to the microscope 193 in the processing unit 191.
- the diagnostic imaging support apparatus 190 includes, for example, a CPU (Central Processing Unit), a RAM (Random Access Memory), an input unit, a drive, an input / output interface (I / F), a communication bus, and the like.
- the CPU is responsible for overall control of the image diagnosis support apparatus.
- each means in the image diagnosis support apparatus 190 can be constructed, and the image diagnosis support apparatus 190 can be realized.
- the image diagnosis support apparatus 190 implements its operation by mounting circuit parts made of hardware parts such as LSI (Large Scale Integration) incorporating a computer program for realizing the functions of the respective means. You can also.
- a computer program may be in the form of a recording medium storing the computer program. Examples of the recording medium include HDD, FD, CD-ROM (CD-R, CD-RW), MO, DVD, memory card, and the like.
- the storage unit 192 is, for example, a ROM, HDD, HD, or the like.
- the HDD controls reading and writing of data with respect to the HD under the control of the CPU, and the HD stores the written data under the control of the HDD, for example.
- the display 196 displays various information such as images, data, and documents.
- Examples of the input unit include a keyboard and a mouse.
- the scanner 195 for example, scans the aforementioned section slide and optically converts the image into an electrical signal.
- the CCD 194 converts, for example, a microscope image of the section slide into an electric signal.
- the image diagnosis support apparatus 190 may be accessible to, for example, an external stained image database that stores information related to stained images.
- the image diagnosis support apparatus 190 may be connected to the stained image database via a communication line, for example.
- FIG. 20 is a block diagram showing an outline of the configuration of the image diagnosis support system in the present embodiment.
- the present invention is not limited to this embodiment.
- the image diagnosis support system of the present embodiment includes an image acquisition unit 2001 that acquires a HE-stained image and a miRNA-stained image, an information acquisition unit 2002 that acquires information on a tumor region in the HE-stained image, and an image acquisition Matching means 2003 for calculating the matching position between the HE-stained image acquired by the means and the miRNA-stained image, information on the tumor region in the HE-stained image acquired by the information acquiring means, and information on the matching position calculated by the matching means Based on the identification means 2004 for identifying the tumor region in the miRNA-stained image and the information on the tumor region in the miRNA-stained image identified by the identification means, the calculation for calculating the staining positive cell content in the tumor region Means 2005 are provided.
- the image diagnosis support system may further include staining level determination means for determining the staining level of miRNA for the miRNA-stained image.
- the staining level determination means determines, for example, the staining level in the specified tumor region of the miRNA stained image.
- each component means should just be a functional block implement
- the image diagnosis support system of the present embodiment is the same as, for example, Embodiments 1B to 1E described later.
- FIG. 21 is a flowchart showing the flow of the process.
- This process is an example of the image diagnosis method of the present invention, and can be executed by, for example, the image diagnosis support system, the image diagnosis support program, or the like of the present invention.
- an HE-stained image and an miRNA-stained image are acquired (step S2101).
- the image can be acquired as an electric signal converted by an image sensor such as a scanner or a CCD.
- tumor region information is acquired for the HE-stained image (S2102).
- the information on the tumor region in the HE-stained image may be information determined by a doctor or the like, or may be information calculated by a known method.
- the HE-stained image and the miRNA-stained image are superposed and matched to calculate a matching position (S2103).
- the tumor region in the miRNA-stained image is calculated from the tumor region information of the HE-stained image and the obtained matching position information. That is, in the miRNA-stained image, a region corresponding to the tumor region in the HE-stained image is calculated and specified as a tumor region (S2104).
- the tumor region in the miRNA-stained image specified based on the information of the HE-stained image is also referred to as “tumor region based on the HE-stained image”.
- the miRNA staining level in the tumor region based on the HE staining image of the miRNA staining image is determined (S2105).
- the miRNA staining level for example, in each slide, since the degree of staining varies depending on the staining process, temperature, type of probe, type of color developing substance or fluorescent substance, etc., it is preferable to standardize the image. .
- the miRNA staining level may be similarly determined for regions other than the tumor region based on the HE-stained image.
- the information on the stained image obtained in this step is accumulated in the stained image database as described above (S2106).
- a tumor region is detected again from the miRNA-stained image (S2107). That is, it is determined whether the staining level of the miRNA-stained image is a level that indicates tumor cells or a level that indicates less than the level that indicates tumor cells. Then, the region indicating the former staining level is specified as a tumor region based on the miRNA staining level. As a result, when the tumor region based on the HE-stained image and the tumor region based on the miRNA staining level overlap, the tumor region based on the miRNA staining level is set as a detection target region.
- the tumor region based on the miRNA staining level is set as a detection target region.
- the threshold between staining at a level meaning a tumor and staining below a level meaning a tumor can be determined, for example, by detecting the intensity of staining at a plurality of levels for the miRNA-stained image. Thereby, for example, even when a non-tumor cell is weakly stained, it can be determined that it does not correspond to staining of a tumor cell. Further, for example, the threshold data is also stored in the stained image database as stained image information.
- the detection target area determined as described above is output as the determined cancer area.
- the staining positive cell content is calculated for the detection target region, and the result is output.
- the tumor region in the miRNA-stained image can be determined based on the database without matching with the HE-stained image.
- FIG. 1 is a block diagram showing an example of an image diagnosis support system of the present invention.
- This system is a system that supports diagnosis based on a pathological image of cancer.
- the system is acquired by an image acquisition unit that acquires a HE-stained image and a miRNA-stained image as a pathological image to be diagnosed, an information acquisition unit that acquires information on a tumor region in the HE-stained image, and an image acquisition unit.
- miRNA staining A specifying unit that specifies a tumor region in the image
- a calculating unit that calculates a staining positive cell content in the tumor region based on information on the tumor region in the miRNA-stained image specified by the specifying unit.
- the system of the present embodiment includes an input device 111, an output device 112, a stained image database 113, a processing device 120, and a storage device 130.
- the stained image database 113 includes one or more HE-stained images, a miRNA-stained image that is a sample of a continuous section adjacent to the HE-stained image specimen (section slide), the HE-stained image, and the miRNA-stained image.
- the adjacent information of the specimen with the image and the tumor region information calculated from the HE-stained image or determined by a doctor or the like are accumulated.
- the relationship information with the subject is associated with the subject identifier.
- a subject identifier 201 for uniquely identifying a subject for example, as shown in FIG. 2, an image identifier 202, staining information 203, image data 204, specimen adjacent information 205, And tumor region information 206 in the HE-stained image is included.
- the image identifier 202 is an identifier for identifying a plurality of pathological images that exist for each subject.
- Staining information 203, image data 204, and tumor region information 206 are distinguished from other images by an image identifier 202.
- the staining information 203 stores the staining information of the image, and examples of the staining information include HE staining information and cancer marker miRNA staining information.
- Image data is stored in the image data 204.
- the specimen adjacency information 205 a correspondence relationship is stored by the image identifier 202.
- tumor region information 206 of the HE-stained image tumor region information calculated from the HE-stained image or determined by a doctor or the like is stored.
- the HE-stained image tumor region information 206 may be stored separately in association with the image identifier 202.
- the input device 111 and the output device 112 for example, a normal input / output device provided in a computer can be used.
- the input device 111 is, for example, a keyboard or a mouse.
- the output device 112 is, for example, a display device or a printer.
- the input device 111 and the output device 112 may be an input file and / or an output file, or may be another computer or the like.
- the storage device 130 includes, for example, a main storage device and an auxiliary storage device provided in a computer, and is used to hold various programs and data executed by the processing device 120.
- the processing device 120 includes, for example, a CPU of a computer and operates under program control.
- the processing device 120 includes an input reception processing unit 121, a stained image and tumor region information acquisition unit 122, an image matching processing unit (matching unit) 123, a miRNA-stained image tumor region extraction unit (identification unit) 124, and a staining. And a positive cell content rate calculation unit (calculation means) 125.
- the stained image and tumor region information acquisition unit 122 has, for example, the functions of the above-described image acquisition unit and information acquisition unit.
- the input reception processing unit 121 receives pathological image designation information to be diagnosed and examination type designation information from the user or the like through the input device 111.
- Examples of the test type include the type of cancer marker miRNA to be tested.
- the input reception processing unit 121 stores the information in the diagnostic image information and the examination target storage unit 131 in the storage device 130, and moves the process to the stained image and tumor region information acquisition unit 122.
- the pathological image designation information is the image identifier 202.
- the image identifier 202 is a HE-stained image or a miRNA-stained image, and one or a plurality of images can be designated.
- the examination type designation information is an miRNA staining item, and any one miRNA or two or more miRNAs can be designated among the cancer marker miRNAs of the present invention.
- the stained image and tumor region information acquisition unit 122 acquires the HE stained image, the miRNA stained image, and the tumor region information in the HE stained image to be diagnosed from the stained image database 113, and each of the HE stained images in the storage device 130 is stained with HE.
- the data is stored in the image data storage unit 132, the miRNA-stained image data storage unit 134, and the HE-stained image tumor region information storage unit 133, and the process is transferred to the image matching processing unit 123.
- the image data 204 having the image identifier 202 is stored in the HE staining image data storage unit 132. . Further, referring to the diagnostic image information and the examination type and specimen adjacency information 205 stored in the examination object storage unit 131, the image data 204 of the miRNA-stained image of the continuous section specimen adjacent to the HE image specimen to be diagnosed is obtained. , And miRNA-stained image data storage unit 134. Further, information of the tumor region information 206 of the HE-stained image is stored in the tumor region information storage unit 133 of the HE-stained image.
- the staining information 203 having diagnostic image information and the image identifier stored in the examination target storage unit 131 is miRNA staining
- the image data 204 having the image identifier 202 is stored in the miRNA stained image data storage unit 134.
- the HE-stained image data storage unit 132 stores the HE-stained image data 204 of the continuous section specimen adjacent to the miRNA-stained image specimen to be diagnosed.
- information of the tumor region information 206 of the HE-stained image is stored in the tumor region information storage unit 133 of the HE-stained image.
- the image matching processing unit 123 reads the HE-stained image and the miRNA-stained image from the HE-stained image data storage unit 132 and the miRNA-stained image data storage unit 134, respectively, and calculates a matching position between the HE-stained image and the miRNA-stained image. Further, the image matching processing unit 123 stores the matching position information in the matching position information storage unit 135 and moves the process to the tumor region extraction unit 124 of the miRNA-stained image. Examples of the matching position information include a rotation angle, a horizontal / vertical deviation width, and the like.
- the HE-stained image and the miRNA-stained image are images similar to each other because they are images obtained by staining continuous sections.
- the HE staining and the miRNA staining may be, for example, the same hue or different hues. However, since the HE staining image and the miRNA staining image are matched, the miRNA staining has a hue different from the HE staining. It is preferred to stain the cells with In HE staining, cell nuclei are usually stained blue with hematoxylin, and cytoplasm is stained pink with eosin.
- the hue of staining by miRNA staining can be appropriately set depending on, for example, the coloring material or fluorescent material used. For image matching, each image can be binarized, and a phase-only correlation method, a residual sequential test method, a method using a singular point, or the like can be used.
- the tumor region extraction unit 124 of the miRNA-stained image includes the tumor region information of the HE-stained image, the miRNA-stained image from the tumor region information storage unit 133 of the HE-stained image, the miRNA-stained image data storage unit 134, and the matching position information storage unit 135, respectively. Data and matching position information are read, and a tumor area in the miRNA-stained image data is calculated. Further, the tumor region extraction unit 124 of the miRNA-stained image stores the information of the tumor region in the miRNA-stained image data in the tumor region information storage unit 136 of the miRNA-stained image, and moves the processing to the staining positive cell content rate calculation unit 125. .
- the staining positive cell content rate calculation unit 125 reads miRNA staining image data and tumor region information from the miRNA staining image data storage unit 134 and the tumor region information storage unit 136 of the miRNA staining image, respectively, and counts the number of staining positive cell nuclei in the tumor region. And the number of staining negative cell nuclei are counted, and the staining positive cell content is calculated and output from the output device 112.
- FIG. 1 An example of the operation of the system shown in FIG. 1 will be described as an embodiment 1B of the image diagnosis support method and image diagnosis support program of the present invention with reference to the flowcharts of FIGS.
- miRNA staining will be described on the assumption that staining of positive cell nuclei is blue and staining of negative cell nuclei is brown.
- the present invention is not limited to this.
- positive cell nuclei and negative cell nuclei can be counted by a general method for identifying stained cell nuclei.
- nuclei may be stained by a general staining method, and cell nuclei with positive miRNA staining and negative cell nuclei may be counted.
- This method is generally a method for supporting diagnosis based on a pathological image of cancer, and includes the following steps (a) to (e).
- the program of this embodiment is a program that supports diagnosis based on a pathological image of cancer, and causes the computer to execute the steps (a) to (e).
- A An image acquisition step of acquiring a HE-stained image and a miRNA-stained image as a pathological image to be diagnosed
- An information acquisition step of acquiring information on a tumor region in the HE-stained image (c) acquired in the image acquisition step Based on the information on the tumor region in the HE-stained image acquired in the matching step (d) information acquisition step and the information on the matching position calculated in the matching step, which calculates the matching position between the HE-stained image and the miRNA-stained image, Identifying step of identifying tumor region in miRNA-stained image (e) Calculation step of calculating staining positive cell content in the tumor region based on information of tumor region in miRNA-stained image identified in the identifying step
- the stained image database 113 includes a subject identifier 201, an image identifier 202, staining information 203, image data 204, and specimen adjacent data that are a series of data for the subject.
- Information 205 and tumor region information 206 in the HE-stained image are stored.
- the tumor region information 206 is information in which a tumor region has been calculated in advance from an HE-stained image, or tumor region information specified by a doctor.
- an image identifier of an HE-stained image or miRNA-stained image that designates a diagnostic image and an miRNA test item request that designates a cancer marker miRNA to be examined are input to the input reception processing unit 121 of the processing device 120 from the input device 111. Supplied.
- the input reception processing unit 121 passes the diagnostic image designation information and examination target designation information from the diagnostic image information and examination target storage unit 131 in the storage unit 130 to the stained image and tumor region information acquisition unit 122. Then, the process proceeds to the stained image and tumor region information acquisition unit 122 (step S301).
- the stained image and tumor region information acquisition unit 122 searches the stained image database 113 using the diagnostic image information and the image identifier in the examination target storage unit 131. If the staining information 203 having the designated image identifier is HE staining, the stained image and tumor region information acquisition unit 122 stores the image data 204 having the image identifier in the HE stained image data storage unit 132. . Further, the tumor region information 206 of the HE-stained image is stored in the tumor region information storage unit 133 of the HE-stained image.
- the diagnostic image information and the miRNA test item in the test target storage unit 131 are read out, the specimen adjacent information 205 in the stained image database 113 is referred to, and the miRNA stained image data 204 that is a continuous section specimen adjacent to the HE stained image is obtained. , And stored in the miRNA-stained image data storage unit 134.
- the staining information 203 having the designated image identifier is miRNA staining
- the image data 204 having the image identifier is stored in the miRNA stained image data storage unit 134.
- the HE stained image data 204 which is a continuous section specimen adjacent to the miRNA stained image, is stored in the HE stained image data storage unit 132.
- the tumor region information 206 of the HE-stained image is stored in the tumor region information storage unit 133 of the HE-stained image.
- the image matching processing unit 123 calculates a matching position between the HE stained image stored in the HE stained image data storage unit 132 and the miRNA stained image stored in the miRNA stained image data storage unit 134.
- the matching position is calculated using, for example, the phase-only correlation method after matching the color scales of the two images.
- the obtained matching position information is stored in the matching position information storage unit 135.
- the matching position information is, for example, a rotation angle and a horizontal / vertical deviation width.
- a process transfers to the tumor area
- the tumor region extraction unit 124 of the miRNA-stained image includes the tumor region information of the HE-stained image stored in the tumor region information storage unit 133 of the HE-stained image and the matching position information stored in the matching position information storage unit 135.
- the tumor region in the miRNA-stained image stored in the miRNA-stained image data storage unit 134 is calculated.
- the tumor region information of the obtained miRNA-stained image is stored in the tumor region information storage unit 136 of the miRNA-stained image.
- a process transfers to the staining positive cell content rate calculation part 125 (step S304).
- the staining positive cell rate calculation unit 125 receives the miRNA staining image data stored in the miRNA staining image data storage unit 134 and the tumor region information stored in the tumor region information storage unit 136 of the miRNA staining image. Then, the number of staining positive cell nuclei and the number of staining negative cell nuclei in the tumor region are counted, the staining positive cell content rate is calculated, and output from the output device 112 (step S305). For example, when staining positive cell nuclei are stained brown and staining negative cells are stained blue, the number of nuclei stained brown and the number of nuclei stained blue are counted. This step is performed, for example, according to the procedure shown in FIGS. 4, 5 and 6 as described below.
- the outside of the tumor region of the miRNA-stained image data is masked from the received miRNA-stained image data and tumor region information (step S401). Then, a brown region that is a brown-stained region and a blue region that is a blue-stained region in the tumor region are identified by discriminant analysis (step S402).
- step S501 image data is converted into an HSV color space (step S501), an unstained region is removed by S (saturation) and V (lightness) (step S502), and a value range of H (hue) is set. [0, 1] is converted to [0.3, 1.3] (step S503).
- step S504 it is checked whether the H (hue) value of all the pixels is included in either [0.3, 0.8] or [0.8, 1.3] (step S504). If all pixels are included on one side, [0.3, 0.8] is output as a blue region and [0.8, 1.3] is output as a brown region (step S507).
- the threshold value t is calculated by discriminant analysis (step S505), and [0.3, t] is output as a blue area and [t, 1.3] is output as a brown area (step S505). S506).
- nucleus extraction is performed in the brown region (step S403), and then nucleus extraction is performed in the blue region (step S404).
- steps S601 first, when a brown region or a blue region is input (step S601), an average and variance of V (lightness) values are taken into consideration, and a V ′ value in which the V value is emphasized by a sigmoid function is calculated. (Step S602).
- Step S603 Next, a Gaussian filter is applied to the binary image, and the position of the nucleus is calculated by comparing adjacent pixels (step S604).
- step S405 the number of nuclei detected in the brown region is counted (step S405), and the number of nuclei detected in the blue region is counted (step S406). Finally, the ratio of the number of brown nuclei to the total number of nuclei, that is, the number of brown nuclei / (the number of brown nuclei + the number of blue nuclei) is calculated (step S407).
- the HE-stained image and the miRNA-stained image are acquired by the image acquisition unit, and the information on the tumor region in the HE-stained image is acquired by the information acquisition unit. Thereafter, a matching position between the HE stained image and the miRNA stained image is calculated by the matching means. Subsequently, based on the information on the tumor region in the HE-stained image and the information on the matching position, the tumor region in the miRNA-stained image is specified by the specifying unit. Based on the information on the tumor region in the miRNA-stained image, the staining means calculates the staining positive cell content in the tumor region in the miRNA-stained image. Thereby, a staining positive cell content rate can be obtained as a quantitative value. As a result, for example, the doctor can make a diagnosis by miRNA staining based on the quantitative value.
- the work labor burden of a doctor or the like is reduced.
- the tumor region determined by the HE-stained image can be associated with the miRNA-stained image by matching the HE-stained image and the miRNA-stained image, which are images of the continuous section specimen. Further, by applying discriminant analysis to the H (hue) value, for example, a brown region and a blue region can be distinguished. Furthermore, the ratio of the number of brown nuclei to the total number of nuclei can be calculated by extracting nuclei in each of the brown and blue regions. Therefore, by presenting the staining positive cell content rate to a doctor or the like, information useful for diagnosis by the doctor can be provided and the diagnosis can be supported.
- FIG. 7 is a block diagram showing an example of an image diagnosis support system of the present invention.
- the system of this embodiment is different from the system of Embodiment 1B shown in FIG. 1 in that the staining positive cell content rate calculation unit 125 calculates the staining intensity in addition to the staining positive cell content rate.
- the other configurations and operations are the same as those in Embodiment 1B.
- the staining positive cell content rate and staining intensity calculation unit 725 reads miRNA staining image data and tumor region information from the miRNA staining image data storage unit 134 and the tumor region information storage unit 136 of the miRNA staining image, respectively. Then, the staining positive cell content rate and staining intensity calculation unit 725 counts the number of staining positive cell nuclei and the number of staining negative cell nuclei in the tumor region, calculates the staining positive cell content rate, and further calculates the staining intensity. And output from the output device 112.
- Embodiment 1C of the image diagnosis support method and image diagnosis support program of the present invention with reference to the flowcharts of FIGS.
- the processing of this embodiment is different from the processing of Embodiment 1B shown in FIG. 3 in that not only the staining positive cell content rate calculation but also the staining intensity calculation is performed, but other operations are the same as those in the above embodiment. The same as 1B.
- the staining positive cell content rate and staining intensity calculation unit 725 receives the miRNA staining image data stored in the miRNA staining image data storage unit 134 and the tumor region stored in the tumor region information storage unit 136 of the miRNA staining image. Then, the number of staining positive cell nuclei and the number of staining negative cell nuclei in the tumor region are counted to calculate the staining positive cell content, and the staining intensity (0: negative, 1: weakly positive, 2: moderately positive, 3: strong positive) are calculated and output from the output device 112 (step S805). This process is performed, for example, according to the procedure shown in FIGS. 9 and 10 as described below.
- step S908 the nuclear staining intensity in the brown region is calculated.
- step S1001 the brown area determined in step S402 in FIG. 9 is input (step S1001). Then, in consideration of the average and variance of the V (brightness) values, a V ′ value in which the V value is emphasized by a sigmoid function is calculated (step S1002). If the value is equal to or lower than the threshold value x, which is the V ′ value, the pixel is in the nucleus region and the number of pixels X is counted (step S1003).
- step S1004 the ratio of the number of pixels satisfying the condition V value ⁇ a is obtained with respect to the number of pixels in the nucleus region.
- step S1005 the staining intensity is output as “3: strong positive” (step S1005). If this is not the case, the ratio of the number of pixels satisfying the condition V value ⁇ b is obtained with respect to the number of pixels in the nucleus region.
- step S1006 it outputs as dyeing
- the ratio of the number of pixels satisfying the condition V value ⁇ c is obtained with respect to the number of pixels in the nucleus region. If the ratio is equal to or greater than a certain ratio (step S1008), the staining intensity is output as “1: weakly positive” (step S1009). If none of these applies, the staining intensity is output as “0: negative” (step S1010).
- Embodiment 1B only the staining positive cell content rate is presented to the doctor, but in Embodiment 1C, not only the staining positive cell content rate but also the staining intensity can be presented to the doctor. For this reason, more useful information can be provided for diagnosis by a doctor, and diagnosis can be supported.
- Other effects of the present embodiment are the same as those of Embodiment 1B.
- FIG. 11 is a block diagram showing an example of an image diagnosis support system according to the present invention.
- the system according to this embodiment is different from the system according to Embodiment 1B shown in FIG. 1 in that a tumor determination and tumor region calculation unit (tumor region calculation means) 1126 is provided.
- the stained image database 113 includes, for example, one or more HE-stained images, a miRNA-stained image that is a sample of a continuous section adjacent to the HE-stained image sample, and a sample of the HE-stained image and the miRNA-stained image. Adjacent information is accumulated.
- the tumor region information calculated from the miRNA-stained image or the tumor region information determined by a doctor or the like may or may not be present.
- the stained image and tumor region information acquisition unit 122 acquires the HE stained image 204, the miRNA stained image 204, and the tumor region information 206 of the HE stained image from the stained image database 113, and the HE in the storage device 130, respectively.
- the stained image data storage unit 132, the miRNA stained image data storage unit 134, and the HE stained image tumor region information storage unit 133 are stored.
- the processing is transferred to the image matching processing unit 123.
- the processing is transferred to the tumor determination and tumor region calculation unit 1126.
- the tumor determination and tumor region calculation unit 1126 reads the HE staining image data from the HE staining image data storage unit 132, determines the tumor, calculates the tumor region, and moves the processing to the image matching processing unit 123.
- a tumor determination method and a tumor region calculation method for example, the method described in Patent Document 1 can be used.
- the staining intensity may be calculated together with the staining positive cell content rate as in Embodiment 1C.
- FIG. 12 is a block diagram showing an example of an image diagnosis support system according to the present invention.
- the system of this embodiment includes a slide imaging unit (slide acquisition unit) 1222 instead of the stained image and tumor region information acquisition unit 122 illustrated in FIG. 1 and the like, and includes a slide database 1213 instead of the stained image database 113.
- diagnostic slide information and inspection object storage unit 1231 are provided.
- the system further includes a slide imaging device 1214 and a tumor determination and tumor region calculation unit 1126. Unless otherwise indicated, in this embodiment, the other configurations and operations are the same as those in the embodiment 1B.
- the slide database 1213 includes one or more HE-stained slides, a miRNA-stained slide that is a specimen of a continuous section adjacent to the HE-stained slide specimen, and a sample of the HE-stained slide and the miRNA-stained slide. Neighbor information is accumulated. Each slide is associated with information related to the subject by a subject identifier.
- the slide photographing device 1214 photographs a designated slide and converts it into digital data.
- the input reception processing unit 121 receives the designation information (slide identifier) of the slide to be diagnosed and the designation information of the examination type through the input device 111 from the user or the like. Then, these pieces of information are stored in the diagnostic slide information and inspection object storage unit 1231 of the storage device 130, and the process is transferred to the slide photographing unit 1222.
- the slide photographing unit 1222 acquires, from the slide database 1213, the HE stained slide and the miRNA stained slide, which are adjacent specimens to be diagnosed. Furthermore, the slide photographing unit 1222 photographs the slide obtained by the slide photographing device 1214 and converts it into digital data, thereby obtaining a HE-stained image and a miRNA image. These images are stored in the HE-stained image data storage unit 132 and the miRNA-stained image data storage unit 134 of the storage device 130, respectively, and the process proceeds to the tumor determination and tumor region calculation unit 1126. As described above, in the present embodiment, the slide photographing unit 1222 has both functions of a slide acquisition unit and an image acquisition unit.
- the staining intensity may be calculated together with the staining positive cell content rate as in Embodiment 1C.
- the present invention is not limited to the above embodiment, and various modifications are possible.
- the input reception processing unit 121 may receive not the image identifier but the subject identifier of the person to be diagnosed.
- the stained image and tumor region information acquisition unit 122 may search the stained image database 113 for an image having a subject identifier and tumor region information.
- a cancer pathological image diagnosis support apparatus (hereinafter referred to as an image diagnosis support apparatus) of the present invention comprises: A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; Diagnostic means based on the characteristics, The feature determination means calculates a feature of each learning pattern corresponding to each feature candidate, and sets a feature candidate having the maximum mutual information amount with the class information of
- the feature candidate having the maximum mutual information amount between the feature of each learning pattern corresponding to each feature candidate and the class information of each learning pattern is As the next feature of the set of
- the classification table creation means calculates each feature of each learning pattern using the feature set, creates the classification table in which each feature of each learning pattern and class information are arranged, and Classify the learning patterns by a classification table;
- the feature extraction means calculates each feature of the input pattern using the feature set,
- the diagnosis unit diagnoses the input pattern based on the calculation result and the classification table.
- the diagnostic imaging support apparatus of the present invention A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; Diagnostic means based on the characteristics, The feature determination means prepares a set of a predetermined number of the learning patterns to be transitioned according to the value of the feature, calculates the feature of each learning pattern corresponding to each feature candidate, and The feature candidate that maximizes the mutual information
- the classification table creation means calculates each feature of each learning pattern using the feature set, creates the classification table in which each feature of each learning pattern and class information are arranged, and Classify the learning patterns by a classification table;
- the feature extraction means calculates each feature of the input pattern using the feature set,
- the diagnosis means performs the transition of the input pattern based on each feature of the input pattern and a transition table in which a set to which the learning pattern belongs is sequentially recorded when each feature of the feature set is determined. And, as a result of the transition, the input pattern is diagnosed based on a set to which the input pattern belongs.
- the diagnostic imaging support apparatus of the present invention A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; Diagnostic means based on the characteristics, The feature determination means prepares a set of a predetermined number of the learning patterns to be transitioned according to the value of the feature, calculates the feature of each learning pattern corresponding to each feature candidate, and The feature candidate that maximizes the mutual information
- the classification table creating means calculates each feature of each learning pattern using the feature set, and uses the classification table formed by arranging each feature of each learning pattern and class information.
- the feature extraction means uses the feature set to calculate each feature of the input pattern that indicates the probability that the ranking feature will be a predetermined value;
- the diagnostic means predetermines the input pattern based on each feature of the input pattern and a transition table that sequentially records a set to which the learning pattern belongs when determining each feature of the feature set. The probability of having class information is calculated and diagnosed.
- the learning pattern input unit and the pattern input unit are pixels belonging to a color region to which a preset tumor cell nucleus belongs, based on RGB values of each pixel in the stained pathological image. And calculating the distance between the center of the distribution of the color area and each pixel belonging to the color area, giving a signal to each pixel according to the distance, and from the distribution of the signal in the pathological image It is preferable to detect a peak and input an image centered on the peak as the learning pattern.
- the feature candidates generated by the feature candidate generation unit include a feature candidate obtained from a feature extraction function.
- the feature candidates generated by the feature candidate generation unit include a feature candidate obtained from a feature extraction function obtained by standardizing a complex Gabor function.
- the feature candidates generated by the feature candidate generation unit include a feature candidate for identifying a tumor color.
- the feature determination unit compares a signal of each pixel included in the learning pattern calculated by the learning pattern input unit with a preset threshold value.
- the feature determination unit includes an average value of a signal of each pixel included in the learning pattern calculated by the learning pattern input unit and a signal of a pixel located in the vicinity of each pixel. Is preferably compared.
- the feature determination unit performs an operation using a noise parameter determined in advance for each of the feature candidates to each of the learning patterns.
- the feature determination means may calculate a probability that the feature of each learning pattern takes a predetermined value as the feature of each learning pattern corresponding to each feature candidate. preferable.
- the classification table creating means replaces the value of the feature with a corresponding position in the classification table, It is preferable to put redundant terms.
- each feature of the input pattern is a value of a probability that the feature of the ranking is a predetermined value
- the diagnosis unit uses the feature to generate the classification table. It is preferable to make a determination by calculating the probability that each feature pattern included in is a predetermined class information value.
- the image diagnosis support program of the present invention is characterized in that the image diagnosis method of the present invention can be executed by a computer.
- the image diagnosis support program of the present invention includes, for example, a learning pattern input unit that cuts out an image centered on a tumor from a pathological image used for learning and inputs it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; An image diagnosis support program of an image diagnosis support apparatus having a diagnosis means for diagnosing based on the characteristics, The feature
- the classification table creating means uses the feature set to calculate each feature of each learning pattern, and uses the classification table formed by arranging each feature of each learning pattern and class information.
- Execute the process of classification Causing the feature extraction means to execute a process of calculating each feature of the input pattern using the feature set;
- the diagnosis unit is configured to execute a process of diagnosing the input pattern based on the calculation result and the classification table.
- the program of the present invention is a program for causing various means of the diagnostic imaging support apparatus to execute the processing steps as described above.
- the image diagnosis support program of the present invention includes: A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; An image diagnosis support program of an image diagnosis support apparatus having a diagnosis means for diagnosing based on the characteristics, The feature determining means prepares a set of a predetermined number of learning patterns to be transitioned according to feature values, calculates the features of each learning pattern
- the classification table creating means calculates each feature of each learning pattern, and creates the classification table in which each feature of each learning pattern and class information are arranged.
- Causing the learning pattern to be classified by the classification table Causing the feature extraction means to execute a process of calculating each feature of the input pattern using the feature set; Based on each feature of the input pattern and a transition table that sequentially records a set to which the learning pattern belongs when determining each feature of the feature set, the diagnostic means And performing a process of diagnosing the input pattern based on a set to which the input pattern belongs as a result of the transition.
- the program of the present invention is a program for causing various means of the diagnostic imaging support apparatus to execute the processing steps as described above.
- the image diagnosis support program of the present invention includes: A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; An image diagnosis support program of an image diagnosis support apparatus having a diagnosis means for diagnosing based on the characteristics, The feature determination means prepares a predetermined number of sets to which the learning pattern should be changed according to the value of the feature, calculates the feature of each learning pattern
- the program of the present invention is a program for causing the various means of the diagnostic imaging support apparatus to execute the processing steps as described above (the same applies hereinafter).
- the image diagnosis support program of the present invention is, for example, In the learning pattern input means and the pattern input means, A process of selecting pixels belonging to a color region to which a preset tumor cell nucleus belongs, from the RGB values of each pixel in the stained pathological image; Processing for calculating the distance between the center of the distribution of the color area and each pixel belonging to the color area; A process of giving a signal to each pixel according to the distance; Processing for detecting a peak from the distribution of the signal in the pathological image; It is preferable to execute a process of inputting an image centered on the peak as the learning pattern.
- the feature candidates generated by the feature candidate generation unit include a feature candidate obtained from a feature extraction function.
- the feature candidates generated by the feature candidate generation means include a feature candidate obtained from a feature extraction function obtained by standardizing a complex Gabor function.
- the feature candidates generated by the feature candidate generation unit include a feature candidate for identifying a tumor color.
- the feature determination unit compares the signal of each pixel included in the learning pattern calculated by the learning pattern input unit with a preset threshold value. It is preferable to execute.
- the feature determination unit includes a signal of each pixel included in the learning pattern calculated by the learning pattern input unit and a signal of a pixel located in the vicinity of each pixel. It is preferable to execute a process of comparing the average value.
- the image diagnosis support program of the present invention causes the feature determination means to execute a process of applying an operation with a noise parameter predetermined for each feature candidate to each learning pattern.
- the image diagnosis support program calculates, for example, the probability that the feature of each learning pattern takes a predetermined value as the feature of each learning pattern corresponding to each feature candidate in the feature determination unit. It is preferable to execute the processing.
- the image diagnosis support program of the present invention replaces the feature value at the corresponding position in the classification table with the classification table creation unit. It is preferable to execute processing for placing redundant terms.
- each feature of the input pattern is a value of a probability that the feature of the ranking becomes a predetermined value
- the diagnosis table is used for the classification table by using the feature. It is preferable to execute a process of making a determination by calculating a probability that each feature pattern included in is a value of predetermined class information.
- the image diagnosis support method of the present invention comprises: A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; An image diagnosis support method using an image diagnosis support device having a diagnosis means for diagnosing based on the characteristics, The feature determination means calculates the feature of each learning pattern corresponding to each feature candidate, and sets the feature candidate that maximizes the mutual information amount with the class information of the learning
- the feature candidate that maximizes the mutual information amount between the feature of each learning pattern corresponding to each feature candidate and the class information of each learning pattern.
- the classification table creating means calculates each feature of each learning pattern using the feature set, and the pattern is generated by the classification table formed by arranging each feature of each learning pattern and class information.
- Perform the steps of classifying The feature extraction means performs a step of calculating each feature of the input pattern using the set of features,
- the diagnosis means performs a step of diagnosing the input pattern based on the calculation result and the classification table.
- the use of the diagnostic imaging support apparatus is not essential, and a method for performing the various steps as described above may be used.
- the image diagnosis support method of the present invention comprises: A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; An image diagnosis support method using an image diagnosis support device having a diagnosis means for diagnosing based on the characteristics, The feature determining means prepares a set of a predetermined number of the learning patterns to be transitioned according to a feature value, calculates the features of each
- the classification table creation means calculates each feature of each learning pattern using the feature set, creates the classification table formed by arranging each feature of each learning pattern and class information, Performing the step of classifying the learning pattern by the classification table;
- the feature extraction means performs a step of calculating each feature of the input pattern using the set of features,
- the diagnosis means performs the transition of the input pattern based on each feature of the input pattern and a transition table in which a set to which the learning pattern belongs is sequentially recorded when each feature of the feature set is determined.
- the method of the present invention does not necessarily require the use of the image diagnosis support apparatus, and may be a method for performing the various steps as described above.
- the image diagnosis support method of the present invention comprises: A learning pattern input means for cutting out an image centered on a tumor from a pathological image used for learning and inputting it as a learning pattern; Learning pattern storage means for storing and holding the learning pattern to which class information is attached; Feature candidate generation means for generating a plurality of feature candidates; Feature determining means for determining a set of features suitable for diagnosis from among the feature candidates generated by the feature candidate generating means; Feature storage means for storing and holding a set of features determined by the feature determination means; A classification table creation means for creating a classification table; A pattern input means for cutting out an image centered on a tumor candidate from a pathological image to be diagnosed and inputting it as an input pattern; Feature extraction means for extracting features from the input pattern; An image diagnosis support method using an image diagnosis support device having a diagnosis means for diagnosing based on the characteristics, The feature determination means prepares a set of a predetermined number of the learning patterns to be transitioned according to the value of the feature, calculates the feature of each
- the classification table creation means calculates each feature of each learning pattern using the feature set, creates the classification table formed by arranging each feature of each learning pattern and class information, Performing the step of classifying the learning pattern by the classification table;
- the feature extraction means performs a step of calculating each feature of the input pattern, using the feature set, indicating a probability that the feature of the rank is a predetermined value;
- the diagnostic means predetermines the input pattern based on each feature of the input pattern and a transition table that sequentially records a set to which the learning pattern belongs when determining each feature of the feature set. Calculating a probability of having the determined class information, and performing a diagnosis step.
- the learning pattern input unit and the pattern input unit include: From the RGB value of each pixel in the stained pathological image, the step of selecting a pixel belonging to a color region to which a preset tumor cell nucleus belongs, Calculating the distance between the center of the color area distribution and each pixel belonging to the color area; Applying a signal to each pixel according to the distance; Performing a step of detecting a peak from the distribution of the signal in the pathological image; It is preferable to perform a step of inputting an image centered on the peak as the learning pattern.
- the feature candidates generated by the feature candidate generation unit include a feature candidate obtained from a feature extraction function.
- the feature candidates generated by the feature candidate generation means include a feature candidate obtained from a feature extraction function obtained by standardizing a complex Gabor function.
- the feature candidates generated by the feature candidate generation unit include a feature candidate for identifying a tumor color.
- the feature determination unit performs a step of comparing a signal of each pixel included in the learning pattern calculated by the learning pattern input unit with a preset threshold value. It is preferable.
- the feature determination means includes an average value of a signal of each pixel included in the learning pattern calculated by the learning pattern input means and a signal of a pixel located in the vicinity of each pixel. Is preferably performed.
- the feature determination means performs a step of applying an operation with a noise parameter predetermined for each feature candidate to each of the learning patterns.
- the feature determination means calculates a probability that a feature of each learning pattern takes a predetermined value as a feature of each learning pattern corresponding to each feature candidate. It is preferable to implement.
- the classification table creation means replaces the feature value with a redundant value at a corresponding position in the classification table.
- the step of placing a term is performed.
- each feature of the input pattern is a value of a probability that the feature of the ranking becomes a predetermined value
- the diagnosis unit uses the feature to generate the classification table. It is preferable to perform a step of calculating and determining a probability that each feature pattern included in is a predetermined class information value.
- the diagnostic imaging support system of the present invention includes: An information processing terminal that holds pathological image data in which patient-specific information is added to a pathological image; and an image diagnosis server that diagnoses the pathological image data;
- the diagnostic imaging server The diagnostic imaging support apparatus of the present invention for diagnosing the pathological image included in the pathological image data; Diagnostic result storage means for storing a diagnostic result by the diagnostic imaging support apparatus together with information specific to the patient;
- the information processing terminal requests transmission of the diagnosis result with information specific to the patient,
- the diagnostic imaging server compares the patient-specific information received from the information processing terminal with the patient-specific information stored together with the diagnosis result, and the patient-specific information received from the information processing terminal When the patient-specific information stored together with the diagnosis result matches, the diagnosis result is transmitted to the information processing terminal.
- the diagnostic imaging support system further includes a billing server that holds usage fee data of the diagnostic imaging support apparatus and the information processing terminal.
- the billing server adds a usage fee for the diagnostic imaging support apparatus when the diagnostic result is stored in the diagnostic result storage means.
- the billing server adds a usage fee for the information processing terminal when the information processing terminal receives the diagnosis result.
- the server of the present invention is a server used in the image diagnosis support system of the present invention,
- the server Server-side transmission means for transmitting information in the server to a terminal via a communication network;
- Server-side receiving means for receiving information transmitted from the terminal via the communication network;
- the diagnostic imaging support apparatus of the present invention for diagnosing a subject using the pathological image data;
- Diagnostic result storage means for storing a diagnostic result by the diagnostic imaging support apparatus together with information specific to the patient;
- the patient-specific information received from the information processing terminal is compared with the patient-specific information stored together with the diagnosis result, and stored together with the patient-specific information received from the information processing terminal and the diagnosis result When the patient-specific information matches, the diagnosis result is transmitted to the information processing terminal.
- the terminal of the present invention is a terminal used in the image diagnosis support system of the present invention, and the terminal An information processing terminal for holding pathological image data in which patient-specific information is added to a pathological image, and terminal-side transmission means for transmitting information in the terminal to the server via the communication network; Terminal-side receiving means for receiving information transmitted from the server via the communication network, Requesting transmission of the diagnostic result accompanied by the patient-specific information and receiving the diagnostic result transmitted from the server.
- image diagnosis support apparatus image diagnosis support method, image diagnosis support program, and image diagnosis support system of the present invention, for example, to identify whether the nature of a tumor is benign or malignant, a cell nucleus, its surrounding tissue, etc.
- Sub-images centered on cell nuclei, stroma, etc. by extracting sub-image image data and storing the sub-images as learning patterns and input patterns, Based on the image, the presence or absence of the tumor and the benign / malignant tumor can be determined with high accuracy and in a short time.
- the pathological image is the aforementioned cancer marker-stained image.
- the sub-image is considered in consideration of changes in cell nuclei and surrounding tissues.
- the presence or absence of a tumor and the benign / malignant tumor can be determined based on the sub-image with high accuracy and in a short time.
- the present invention will be described by taking Embodiments 2A and 2B as examples.
- the cancer marker stained image is referred to as “miRNA stained image”. Note that the present invention is not limited to these embodiments.
- FIG. 13 is a block diagram illustrating a configuration of the diagnostic imaging support apparatus according to the present embodiment.
- the diagnostic imaging support apparatus according to the present embodiment includes a learning pattern input unit 1300, a learning pattern storage unit 1301, a feature candidate generation unit 1302, a feature determination unit 1303, a feature storage unit 1304, and a classification table creation unit. 1305 and a classification table 1306.
- Learning pattern input means 1300 extracts sub-images such as cell nuclei and cytoplasm from the miRNA-stained image, and stores the sub-images in learning pattern storage means 1301.
- Learning pattern storage means 1301 is means for storing and holding a desired number of sub-images used for learning.
- Feature candidate generation means 1302 is means for sequentially generating feature candidates from a predetermined number of feature parameter sets.
- the feature determination unit 1303 is a unit that determines a feature set that is optimal for pattern identification from among the feature candidates generated by the feature candidate generation unit 1302.
- the feature storage unit 1304 is a unit that stores and holds the feature set determined by the feature determination unit 1303.
- the classification table creation unit 1305 is a unit that creates a classification table 1306 for performing diagnosis using the feature set determined by the feature determination unit 1303.
- FIG. 14 is a flowchart for explaining the procedure of the feature determination process of the image diagnosis support apparatus according to this embodiment.
- the feature candidate generation unit 1302 sequentially generates feature candidates based on a large number of preset feature parameter sets (for example, N) (S1401).
- N preset feature parameter sets
- parameter sets 1 to N_1 are feature candidates related to texture
- parameter sets N_1 + 1 to N_1 + N_2 are feature candidates related to colors
- N_1 + N_2 + 1 to Let N parameter sets be feature candidates related to colors averaged by surrounding pixels.
- feature candidates related to texture, feature candidates related to color, and feature candidates related to color averaged in surrounding pixels are adopted as feature candidates.
- Elements necessary for determining the features of the pixels included in the pathological image can be generated as feature candidates.
- This feature is determined, for example, by one of procedures 1 to 3 shown below.
- the s-th feature parameter set (k_s, r 0 _s, ⁇ _s, th_s) is substituted into (k, r 0 , ⁇ , th), and is defined by the parameters k, r 0 , ⁇ .
- a complex Gabor function Gab and a Gaussian function G exemplified in the following formula (1) are generated.
- the feature c is calculated using a grayscale image obtained by performing grayscale conversion on a color sub-image stored in the learning pattern storage unit 1301 as a learning pattern.
- the feature candidate generating unit 1302 sends the complex Gabor function Gab and the Gaussian function G of the formula (1) to the feature determining unit 1303 together with the threshold parameter th and the feature candidate identification number s (step S1402).
- the data is sent to the determination unit 1303 (step S1403).
- the present invention is not limited to this.
- the method of the present invention can be applied to a case of three or more classes.
- the feature determination unit 1303 calculates the feature c according to the following equation (2) using the feature candidates for the sub-images sequentially received from the learning image storage unit 1301 (step S1404).
- the feature candidates are, for example, a complex Gabor function, a Gaussian function, and other parameters shown in the equation (1).
- the t-th learning pattern is I_t (r, i_rgb), and the calculation is repeated for all learning patterns (M).
- the denominator is a normalization (standardization) factor for suppressing the fluctuation of the value of a due to the size of the pattern (image brightness).
- the denominator equation can be replaced by another form of normalization factor. Further, depending on the pattern to be handled, such a normalization factor may be omitted.
- the s-th feature parameter set (x_s, y_s, color_index) and the feature candidate identification number s are sent to the feature determination unit 1303 (step S1402).
- (x_s, y_s) represents the position of the pixel that determines the feature c in the sub-image
- miRNA staining of cells with the cancer marker miRNA is an important element in characterizing a pathological image, it is preferable to stain the cell nucleus, cytoplasm, stroma, pores, and the like with different colors.
- color_index 1, 2, 3, 4 is assigned to the cell nucleus color, cytoplasm color, stroma color, and pore color, respectively.
- r is a pixel coordinate
- t 1 to M.
- the feature determination unit 1303 determines the color of the pixel located at the position (x_s, y_s) in the learning pattern for the learning pattern received from the learning pattern storage unit 1301 (step S1403) by the following method. If it matches the color specified by the parameter color_index, the value of the feature c is set to 1, and otherwise, the value of the feature c is set to 0 (step S1404).
- the t-th learning pattern is set to I_t (r, i_rgb), and the determination of the feature c is repeated for all learning patterns (M).
- a predetermined threshold value for example 0.25.
- the value of the hematoxylene signal is added only for a pixel in which the value of the hematoxylene signal is in the vicinity of the pixel of interest, for example, 0.25 or more, and the average value is calculated.
- a predetermined threshold value for example, 0.25
- the configuration may be determined.
- Colors of holes, cytoplasm, stroma, etc. other than cell nuclei are classified according to a predetermined color area.
- (x_s, y_s) represents the position of the pixel that determines the feature c in the sub-image
- th represents the threshold parameter
- the feature determination unit 1303 determines the feature c by the following method for the learning pattern received from the learning pattern storage unit 1301 (step S1403) (step S1404).
- the t-th learning pattern is set to I_t (r, i_rgb), and the determination of the feature c is repeated for all learning patterns (M).
- the feature determining unit 1303 designates the color of the pixel (x ′, y ′) within 2 pixels from the pixel in the vicinity of the sub-image position (x_s, y_s) to the x coordinate and the y coordinate by color_index. Check if it matches the color.
- the pixel (x ′, y ′) is, for example, a pixel within two pixels from the pixel of interest (x_s, y_s) to the x coordinate and y coordinate, and
- the number of neighboring pixels having a color that matches the color specified by color_index is added and divided by the total number of neighboring pixels to obtain an average value. If the average value exceeds the threshold parameter (th_s), the feature c is set to 1. In other cases, the feature c is set to 0.
- the feature determination unit 1303 can determine the feature c of the pixel of interest by, for example, the following method in addition to the above-described three procedures.
- the RGB value of each pixel of the sub-image including the pixel of interest is HSV converted, and the RGB value is converted to hue (H: 0 to 1), saturation (S: 0 to 1), brightness (V: 0 to 1). ) Value.
- the color of the pixel at the position (x, y) in the sub-image specified by (x, y, H ′, S ′, V ′) received from the feature candidate generation unit 102 as the feature parameter set is ( H ′, S ′, V ′), the feature c is 1; otherwise, the feature c is 0.
- the feature determination unit 1303 calculates the mutual information MI obtained from the s-th feature candidate, Calculation is performed according to the following equation (3), and this is stored together with the feature candidate identification number s (step S1405).
- M is the total number of sub-images.
- M (q) is the total number of sub-images belonging to class q
- M (c) is the total number of sub-images with feature c
- M (q, c) is the sub-feature with feature c and belonging to class q The total number of images.
- c]> C is an averaging operation regarding c, and is calculated by the following formula (4).
- the next (s + 1) th feature candidate is sent from the feature candidate generating means 1302, and the same processing is repeated (steps S1402 to S1405).
- the feature determination unit 1303 compares the mutual information MI obtained from each feature candidate. Then, the feature candidate that provides the maximum mutual information Max MI [Q; C] is determined as the first feature of the feature set to be determined (step S1406).
- the feature determination unit 1303 determines the second feature.
- the feature determination unit 1303 sequentially receives feature candidates from the feature candidate generation unit 1302 (step S1402), and calculates a feature c for each sub-image (steps S1403 and S1404). Note that the calculation result of the feature c in step S1404 at the time of determining the first feature described above is stored and held according to the usable storage capacity, and the feature determination unit 1303 stores the stored content (feature candidate). ) May be replaced with an operation of reading out.
- the feature determination means 1303 has a condition that the first feature c 1 that has already been determined is known.
- the mutual information MI 2 obtained from the sth feature candidate is calculated according to the following equation (5), and stored together with the feature candidate identification number s (step 1405).
- M (c 1 ) is the total number of sub-images whose first feature is c 1
- M (q, c 1 ) is a sub-image whose first feature is c 1 and belongs to class q
- M (c, c 1 ) is the total number of sub-images whose feature is c and the first feature is c 1
- M (q, c, c 1 ) is feature c and 1
- the th feature is c 1 and the total number of sub-images belonging to class q.
- the feature determination unit 1303 compares the conditional mutual information MI 2 obtained from each feature candidate, feature candidate of information is obtained and determined as the feature c 2 a second set of features to be determined (step S1406).
- the MI m + 1 represents the amount of information obtained from the feature c under the condition that the m-th feature (c 1 , c 2 ,..., C m ) is known. Even if a new feature is selected, such processing is continued until the obtained information amount (additional information amount) becomes smaller than a preset threshold MI_th. For example, when the threshold value MI_th is set to zero, the following procedure is performed to determine the next feature until the obtained information amount (additional information amount) becomes zero, that is, until the end condition is satisfied. Repeated.
- the feature determination process ends when this end condition is satisfied.
- Each parameter of the determined feature set is stored in the feature storage unit 1304 (step S1407).
- the following configuration can be considered in which a device for reducing the number of feature candidates generated by the feature candidate generation unit 1302 is added.
- the structure which fixes the value of threshold value MI_th to the intermediate value of these two in-class average values can be considered.
- the mutual information MI is calculated for each complex Gabor function by the above equation (3).
- a threshold value MI_th that gives MI is recorded.
- the complex Gabor function is used as the feature extraction function that constitutes the feature candidate.
- the feature candidate may be constituted only by the other feature extraction function as appropriate. May be.
- a modification in which a partial space is configured for each class and an index indicating the distance to the partial space is added to the feature candidate is also preferable. It is also possible to add a weighted average luminance near a certain point calculated using a Gaussian function to a feature candidate. Also, the weighted average brightness near a point calculated using a Gaussian function, normalized by the average brightness calculated using a Gaussian function with a larger spread, that is, the vicinity of a point is It is also possible to add an index indicating whether it is brighter or darker than the surrounding area to the feature candidate. In addition, standard features used in diagnosis can be added to feature candidates.
- a classification table 1306 (shown in FIG. 16) for use in pattern identification can be created.
- the classification table creating unit 1305 activated by a desired unit creates the classification table 1306 will be described.
- the classification table creating unit 1305 determines each sub-image from the learning pattern storage unit 1301 and each parameter of the feature set from the feature storage unit 1304 (hereinafter, a total of n features are determined). Receive) Then, each sub-image and each feature value (c 1 , c 2 ,..., C n ) for each sub-image are stored in the classification table 1306.
- a redundant term (don't care term) is used. For example, when a certain sub-image can be classified only with the feature values from the beginning to the i-th feature (c 1 , c 2 ,..., C i ), the value of the feature vector after the i + 1th is Replace with a symbol representing don't care and store.
- FIG. 15 is an example of a flowchart showing a procedure for creating the classification table 1306 according to this embodiment.
- the classification table creation unit 1305 uses the parameters of the feature set stored in the feature storage unit 1304 for the input sub-image, and uses the feature vectors (c 1 , c 2 ,. c n ) is calculated (steps S1501 and S1502).
- step S1503 It is determined whether or not a sub-image having a feature vector that matches the feature vector exists in the classification table 1306 (step S1503).
- the value of the corresponding feature is determined as matching regardless of the value.
- the first to i-th features (c 1 , c 2 ,. ., C i ) checks whether there is a sub-image that matches this sub-image (step S1505).
- step S1506 the process returns to step S1501, and the next sub-image is input.
- step S1501 the increment variable i is incremented by 1, and the process returns to step S1501. That is, the process of increasing i is continued until it can be identified by the value of the i-th feature of the input sub-image.
- sub-images belonging to different classes may have the same feature vector. In this case, for example, the number of sub-images belonging to each class is counted, and the larger class can be determined as the class represented by this feature vector.
- FIG. 16 is an example of the classification table 1306 employed by the present invention.
- FIG. 16 shows a table for storing the identification mark (q) of each sub-image class and the feature vectors (c 1 , c 2 ,..., C n ).
- the symbol “*” represents don't care.
- FIG. 17 is a block diagram showing an example of the processing flow of the diagnostic method of the present invention.
- FIG. 17 shows a pattern input unit 1701, a feature extraction unit 1702, and a diagnosis unit 1703.
- a feature storage unit 1304 for storing and holding a set of determined features used for feature extraction by the feature extraction unit 1702 and a created classification table 1306 used for diagnosis by the diagnosis unit 1703 are shown.
- the pattern input unit 1701 is a unit for inputting a sub image from a desired medium. In the present embodiment, it is means for inputting an image (sub-image) centered on the cell nucleus of the cells constituting the pathological tissue. In the present embodiment, an image centered on the cell nucleus is input, but the present invention is not limited to this. For example, an image used as a judgment material by a pathologist in diagnosis of a pathological tissue such as a cell nucleus, a void, a cytoplasm, and a stroma can be input as a sub-image.
- the feature extraction unit 1702 is a unit that extracts the feature from the sub-image sent from the pattern input unit 1701 using the determined feature set.
- Diagnostic unit 1703 is a unit for diagnosing information represented by the sub-image based on the feature obtained by the feature extraction unit 1702.
- a sub-image is taken and sent to the feature extraction means 1702.
- the feature extraction unit 1702 calculates a sub-image feature vector for the input sub-image, and sends the calculation result to the diagnosis unit 1703.
- the feature vector is calculated by using a feature set determined by the procedures 1, 2, and 3 stored in the feature storage unit 1304, for example, a feature set determined by the feature determination method described above.
- the diagnosis unit 1703 refers to the classification table 1306, searches for a match with the feature vector, reads the class indicator recorded therein, and outputs it as a diagnosis result. At this time, if a symbol representing don't care is recorded on the classification table 1306, the diagnosis unit 1703 determines that the part is the same regardless of the value of the corresponding feature. To do.
- ID3 and the like are also in common with the present invention in that the classification rule at each node of the decision tree is determined according to the information amount maximization criterion.
- a classification rule for example, a feature
- the different classification rules Is determined.
- the node depth is the same
- the same feature is determined as an arbitrary n-th feature. This is a major difference between the two.
- the learning patterns are completely classified.
- a big difference appears in the generalization performance that is, the discrimination performance with respect to a sub-image that has not been learned.
- both trees have the same depth (assumed to be n)
- 2n features are determined in ID3 or C4.5, whereas only n features are determined in the present invention. Therefore, the present invention has a simpler structure.
- the difference in the number of determined features grows exponentially as the problem becomes more difficult and deeper trees are needed.
- processing for extracting a sub-image from a pathological image in the learning pattern input unit 1300 and the pattern input unit 1701 will be described.
- sub-image extraction centering on cell nuclei will be described, but the present invention is not limited to this. Therefore, the morphological features that the pathologist pays attention to when observing the pathological image, such as holes, cytoplasm, and stroma, can be extracted as sub-images.
- the process of extracting the sub-image centered on the cell nucleus is to calculate the miRNA staining signal from the RGB value of each pixel in the pathological image and to detect the center position of the cell nucleus from the distribution of the miRNA staining signal of each pixel in the pathological image. It is divided roughly into the step to do. Actually, processing such as smoothing of the miRNA staining signal is also included.
- a pathological image in which cell nuclei are stained blue by miRNA staining will be described as an example.
- the learning pattern input unit 1300 and the pattern input unit 1701 first apply a miRNA staining signal to each pixel of the pathological image in which the cell nucleus is stained blue.
- the miRNA staining signal having a value of 0 is calculated for the area stained with color. This process is performed by examining the color distribution of cell nuclei in the RGB space and calculating the distance from the center of the distribution to the RGB value of each pixel. That is, if the RGB value of each pixel is examined and the value is located near the center of the color distribution of the cell nucleus in the RGB space, a large miRNA staining signal close to 1 is far from the center. A miRNA staining signal close to 0 is given. However, since the method of staining nuclei may vary from sample to sample due to differences in staining processing, the color distribution of cell nuclei is calculated by an adaptive method here.
- a pixel whose RGB value falls within the color region of the cell nucleus is selected from the pathological image, and this is used as a pixel representing the color of the cell nucleus.
- the color region of the cell nucleus is determined in advance by the following method. First, images of cell nuclei that vary in staining due to differences in staining processing are collected. Next, in each image, the RGB value of each pixel in the cell nucleus region is examined. At the same time, in each of these images, for example, the RGB values of the pixels in a region stained with a color characteristic of cytoplasm, stroma, and pores are also examined. Then, the pixel region of the cell nucleus, which is dyed with a characteristic color of the cytoplasm, stroma, and pores, is not included at all or hardly, and the color region of the cell nucleus composed of pixels of the cell nucleus region is determined.
- the learning pattern input unit 1300 and the pattern input unit 1701 give a miRNA staining signal to each pixel based on the following method.
- N pixels whose RGB values fall within the color region of the cell nucleus are selected from the pathological images input to the learning pattern input unit 1300 and the pattern input unit 1701 (Ste S1802).
- T is a symbol representing a vector transposition operation.
- ⁇ the distance L between each pixel (R, G, B) and the average value (R 0 , G 0 , B 0 ), and miRNA staining signal (Hema) according to the following equation (8) Is calculated (step S1804).
- the miRNA staining signal calculated for each pixel according to the equation (8) is defined as Hema ( ⁇ r).
- ⁇ r (x, y) is a position vector representing the position of the pixel in the pathological image.
- Hema ( ⁇ r) is smoothed by the following equation (9) using the smoothing mask M low (step S1805), and the peak is set as the center position of the cell nucleus (steps S1806 and S1807).
- the smoothing mask M low for example, a function represented by the following formula (10) is used.
- the normalization factor 1 / l of the equation (10) is determined by the following equation (11).
- S ex and S in are parameters, which are determined in advance.
- the value of S ex is the typical size (radius) degree of cell nuclei, and the value of S in is set to a value of 1.2 times that of S ex.
- miRNA staining signal (miR ′) is calculated for each pixel, the value of miR ′ at the point of interest is larger than a predetermined threshold (for example, 0.25) and in the vicinity of the point (for example, If the difference between the x, y and y coordinates is greater than the value of miR ′ at any point of any pixel within 3 pixels), that point is detected as a peak, and this is taken as the center of the cell nucleus (step S1807).
- a predetermined threshold for example, 0.25
- a predetermined threshold for example, 0.25
- the difference between the x, y and y coordinates is greater than the value of miR ′ at any point of any pixel within 3 pixels
- the miRNA staining signal smoothing process and the peak detection process among the above processes include a plurality of (for example, parameters S ex and S in ) having different sizes (for example, 3 types) Smoothing processing and peak detection processing are performed using a smoothing mask. And it is good also as a structure which makes the peak position detected by one of processes the center of a cell nucleus.
- Learning pattern input means 1300 and pattern input means 1701 first detect the cell nucleus center by the above processing on the input pathological image. Then, a large number of images (sub-images) of a predetermined size centered on the detected cell nucleus center are extracted from the pathological image (by the number of detected cell nucleus centers), and each of these sub-images is learned patterns or input. Extracted as a pattern (step S1808).
- the image diagnosis support system of the present invention may have means for evaluating the effectiveness of miRNA-stained images, for example.
- the accuracy of cancer evaluation can be further improved by evaluating the effectiveness of miRNA-stained images.
- staining image is synonymous with evaluation of the effectiveness of the slice slide which performed miRNA dyeing
- the third embodiment further includes correction means for correcting the staining state of the miRNA-stained image and non-tumor cells in the corrected miRNA-stained image in the diagnostic imaging support system of the first or second form.
- correction means for correcting the staining state of the miRNA-stained image and non-tumor cells in the corrected miRNA-stained image in the diagnostic imaging support system of the first or second form.
- Non-tumor cell detection means and determination means for determining the presence or absence of miRNA staining of the detected non-tumor cells.
- the processing by the system of this embodiment can be performed as follows, for example.
- First, the staining state of the acquired miRNA stained image is corrected.
- the correction is performed on the dye, intensity, and the like in consideration of the state of the section slide used, the state of the section slide similarly stained, the staining condition, the acquisition condition of the image data, and the like.
- non-tumor cells are detected from the miRNA-stained image after correction.
- the non-tumor cells can be determined based on information such as the shape and size of cells, the shape and size of cell nuclei, and the location of tissue. Further, this determination can be made by a module that has performed machine learning based on the above-described conditions, for example.
- the detection of the non-tumor cells can be performed, for example, by obtaining a stained image with a counter stain and matching the counter stained image with the miRNA stained image.
- the matching is the same as the matching between the miRNA stained image and the HE stained image, for example.
- the counter stain can be appropriately determined according to, for example, the type of sample that is the subject, and examples thereof include a Cologne Echrotroth.
- the type of non-tumor cells to be detected is not particularly limited, and can be appropriately determined according to the type of sample as a specimen, for example, lymphocytes, fibroblasts, vascular endothelial cells, etc. Whether a cell is a non-tumor cell can be determined by whether it is a defined cell size and / or shape.
- the miRNA stained image the presence or absence of miRNA staining of the detected non-tumor cell is determined.
- the miRNA-stained image is regarded as having no effectiveness, and the process is terminated without proceeding to further steps.
- this miRNA-stained image is considered valid and proceeds to further steps, for example, detection of tumor regions based on the miRNA staining described above.
- an LNA-modified probe (trade name: miRCURY-LNA detection probe, Exiqon) labeled with digoxigenin (DIG) was used.
- DIG digoxigenin
- SEQ ID NO: 5 The sequence of the probe for detecting hsa-miR-92a is shown in SEQ ID NO: 5
- SEQ ID NO: 6 The sequence of the probe for negative control is shown in SEQ ID NO: 6.
- the following negative control probe sequence is a scrambled sequence of the following hsa-miR-92a detection probe sequence.
- hsa-miR-92a detection probe (SEQ ID NO: 5) 5'-acagggccgggacaagtgcaata-3 '
- Negative control probe (SEQ ID NO: 6) 5'-gtgtacacgtctataccccca-3 '
- In situ hybridization is performed by using the RiboMap in situ hybridization kit (trade name, Ventana Medical Systems), using the Ventana Discovery in Ventilator product name. Unless otherwise indicated, the standard protocol provided by RiboMap application note (http://www.ventanamed.com) of Ventana Medical Systems was used.
- leukocytes were collected from the whole blood of each leukemia patient.
- the obtained leukocytes were fixed with paraformaldehyde fixing solution, embedded in paraffin, and prepared with a general method.
- the section was deparaffinized and then subjected to in situ hybridization.
- in situ hybridization initial fixation of a section after deparaffinization is performed by incubating the slide having the section with formalin-based RiboPrep (trade name, Ventana Medical Systems) at 37 ° C. for 30 minutes. It was.
- FIG. 1 The results are shown in FIG.
- the upper and middle panels are photographs showing the staining of leukocytes from AML patients (FAB classification M3), and the lower panels are photographs showing the staining of leukocytes from ALL patients.
- the left panel is a photograph showing counterstaining with Cologne Echrotroth
- the middle panel is a photograph showing staining with the hsa-miR-92a detection probe
- the right panel is the photograph shown above. It is a photograph which shows the dyeing
- the bar in each figure shows a length of 50 ⁇ m.
- FIG. 22 shows the results of one AML patient and one ALL patient, but similar results were obtained in other patients.
- hsa-miR-92a expression was not detected in normal leukocytes.
- hsa-miR-92a is strongly expressed in leukocytes of AML patients and ALL patients, it can be seen that detection of hsa-miR-92a in leukocytes can evaluate the possibility of cell cancer. It was.
- Example 2 A section was prepared in the same manner as in Example 1 except that the tissue collected from the breast was used, and hsa-miR-92a was detected by in situ hybridization. The result is shown in FIG. (A) to (D) of FIG. 23 are photographs showing the results of miRNA staining of breast tissue collected from different parts.
- Example 3 A section was prepared in the same manner as in Example 1 except that tissue collected from hepatocytes was used, and hsa-miR-92a was detected by in situ hybridization.
- HCC hepatocellular carcinoma
- LC liver cirrhosis
- hepatocytes of HCC patients were collected from patients with different ages, sexes, hepatitis virus types, clinical stages, and tumor differentiation levels.
- FIG. 24 shows the staining results of two hepatocytes (Case 1 and Case 2) as representatives of the 22 HCC patient-derived hepatocytes.
- FIG. 24 is a photograph showing staining of hepatocytes derived from an HCC patient.
- the upper and middle panels show the results of case 1 hepatocytes, and the lower panels show the results of case 2 hepatocytes.
- the left panel is a photograph showing counterstaining with Cologne Echrotroth and HE
- the middle panel is a photograph showing staining with the hsa-miR-92a detection probe
- the right panel is a photograph.
- FIG. 3 is a photograph showing staining using the negative control probe.
- FIG. The bar in each figure shows a length of 100 ⁇ m.
- the middle panel is an enlarged photo of the upper panel. In each panel, the dark part is the dyed part and the light part is the non-stained part.
- the expression level of RNU48 was also measured in the same manner.
- the ratio of the expression level of hsa-miR-92a to the expression level of RNU48 (hsa-miR-92a / RNU48) was calculated as the expression level of hsa-miR-92a.
- hsa-miR-92a is strongly expressed in the hepatocytes of HCC patients, it was found that the detection of hsa-miR-92a in hepatocytes can evaluate the possibility of cell cancer. .
- the present invention by detecting the expression level of the cancer marker of the present invention in a sample, for example, the presence / absence and progression of cancer can be determined with high reliability. Furthermore, by making the evaluation method of the present invention correspond to the evaluation of cancer by conventional HE staining or the like, it becomes possible to evaluate cancer with even higher reliability.
- Input Device 112 Output Device 113 Stained Image Database 120 Processing Device 121 Input Acceptance Processing Unit 122 Information Acquisition Unit 123 Image Matching Processing Unit 124 Tumor Region Extraction Unit 125 Staining Positive Cell Content Calculation Unit 130 Storage Devices 131, 132, 133, 134 135, 136, 1231 Storage unit 725 Staining positive cell content rate and staining degree calculation unit 1126 Tumor determination and tumor region calculation unit 1213 Slide database 1214 Slide device 1222 Slide photographing unit 1300 Learning pattern input unit 1301 Learning pattern storage unit 1302 Feature candidate Generation unit 1303 Feature determination unit 1304 Feature storage unit 1305 Classification table creation unit 1306 Classification table 1701 Pattern input unit 1702 Feature extraction unit 1703 Diagnosis unit 190 Diagnosis support apparatus 191 processing unit 192 storage unit 193 microscope 194 CCD 195 Scanner 196 Display 2001 Image acquisition means 2002 Information acquisition means 2003 Matching means 2004 Tumor region specifying means 2005 Calculation means
Landscapes
- Chemical & Material Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Organic Chemistry (AREA)
- Proteomics, Peptides & Aminoacids (AREA)
- Engineering & Computer Science (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Analytical Chemistry (AREA)
- Zoology (AREA)
- Genetics & Genomics (AREA)
- Wood Science & Technology (AREA)
- Physics & Mathematics (AREA)
- Biotechnology (AREA)
- Microbiology (AREA)
- Molecular Biology (AREA)
- Hospice & Palliative Care (AREA)
- Biophysics (AREA)
- Oncology (AREA)
- Biochemistry (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
Description
試料における癌マーカーを検出する癌マーカー検出工程と、
前記癌マーカー検出工程において検出した前記癌マーカーの発現レベルに基づいて、前記試料の癌の可能性を評価する工程とを含み、
前記試料が、細胞または組織であり、
前記癌マーカーが、hsa-miR-92およびhsa-miR-494の少なくとも一方のmiRNAを含むことを特徴とする。
本発明における癌マーカーmiRNAは、前述のように、hsa-miR-92およびhsa-miR-494の少なくとも一方のmiRNAである。以下、癌マーカーを、癌マーカーmiRNAともいう。
機能性hsa-miR-92a(配列番号1)
5’-uauugcacuugucccggccugu-3’
Minor hsa-miR-92a-1*(配列番号6)
5’-agguugggaucgguugcaaugcu-3’
Minor hsa-miR-92a-2*(配列番号7)
5’-ggguggggauuuguugcauuac-3’
機能性hsa-miR-92b(配列番号3)
5’-uauugcacucgucccggccucc-3’
Minor hsa-miR-92b*(配列番号4)
5’-agggacgggacgcggugcagug-3’
機能性hsa-miR-494(配列番号2)
5’-ugaaacauacacgggaaaccuc-3’
Wu H.et al., 2007年, PLoS ONE 2(10):e1020 miRNA profiling of naive,effector and memory CD8 T cells.
Pablo Landgraf et al., 2007年, Cell, vol.129, p.1401-1414 A Mammalian microRNA Expression Atlas Based on Small RNA Library Sequencing.
Neilson et al., 2007年, Genes Dev, vol.21, p.578-589 Dynamic regulation of miRNA expression in order to stage of cellular development.
Ruby et al., 2006年, Cell, vol.127, p.1193-1207 Large-scale sequencing reveals 21U-RNAs and additional microRNAs and endogeneous siRNAs in C.elegans.
Obernoster et al., RNA 2006年, vol.12, p.1161-1167 Post-transcriptional regulation of microRNA expression.
Lagos-Quintana et al., 2002年, Curr Biol,vol.12, p.735-739 Identification of tissue-specific microRNAs from mouse.
本発明の評価方法は、前述のように、試料における癌マーカーを検出する癌マーカー検出工程と、前記癌マーカー検出工程において検出した前記癌マーカーの発現レベルに基づいて、前記試料の癌の可能性を評価する工程とを含み、前記試料が、細胞または組織であり、前記癌マーカーが、hsa-miR-92およびhsa-miR-494の少なくとも一方のmiRNAを含むことを特徴とする。本発明によれば、例えば、被検者の試料について、前記癌マーカーを検出することによって、癌が発症する可能性、癌化しているか否か、前臨床期(初期段階)もしくは臨床期等の癌の進行度、または、予後の状態の評価等を行うことが可能となる。本発明の評価方法は、例えば、癌に罹患しているか否かの評価方法ともいえる。
hsa-miR-92a検出用プローブ(配列番号5)
5’-acaggccgggacaagtgcaata-3’
(1)被検者の試料における前記癌マーカーmiRNAの発現レベルを、正常者の試料における前記癌マーカーmiRNAの発現レベルと比較し、前記正常者の発現レベルよりも高い場合に、前記被検者は、前記癌の可能性が高いと決定する。
(2)被検者の試料における前記癌マーカーmiRNAの発現レベルを、正常者の試料における前記癌マーカーmiRNAの発現レベルと比較し、前記正常者の発現レベルよりも相対的に高い程、前記被検者は、前記癌が相対的に進行していると決定する。
(3)被検者の試料における前記癌マーカーmiRNAの発現レベルを、進行期別の各患者の試料における前記癌マーカーmiRNAの発現レベルと比較し、前記被検者は、同程度の発現レベルを示す患者と同じ進行期であると決定する。
前記HE染色画像における腫瘍領域の情報を取得する情報取得工程、
前記HE画像取得工程において取得された前記HE染色画像と、前記癌マーカー検出工程において取得された前記癌マーカー染色画像とのマッチング位置を算出するマッチング工程、
前記情報取得工程において取得された前記HE染色画像における腫瘍領域の情報および前記マッチング工程において算出された前記マッチング位置の情報に基づいて、前記癌マーカー染色画像における腫瘍領域を特定する特定工程、および、
前記特定工程において特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞を検出する染色陽性細胞検出工程を含むことが好ましい。
本発明の評価試薬は、前述のように、本発明の評価方法に使用するための評価試薬であって、本発明の癌マーカーの検出試薬、すなわち、hsa-miR-92およびhsa-miR-494の少なくとも一方のmiRNAを検出するためのmiRNA検出試薬を含むことを特徴とする。このような評価試薬によれば、本発明の評価方法を簡便に実施できる。
本発明の評価キットは、前述のように、本発明の評価方法に使用するための評価キットであって、hsa-miR-92およびhsa-miR-494の少なくとも一方のmiRNAを検出するmiRNA検出試薬を含むことを特徴とする。前記miRNA検出試薬は、例えば、前記本発明の評価試薬があげられ、前述の通りである。このような評価キットによれば、本発明の評価方法を簡便に行える。
本発明は、癌の病理画像に基づく診断を支援するためのシステム、プログラム、方法および装置であって、下記に示す第1の形態および第2の形態があげられる。
本発明は、癌の病理画像に基づく診断を支援する、癌病理画像診断支援システム(以下、画像診断支援システムという)であって、
診断対象となる前記病理画像として、HE染色画像および癌マーカー染色画像を取得する画像取得手段と、
前記HE染色画像における腫瘍領域の情報を取得する情報取得手段と、
前記画像取得手段により取得された、前記HE染色画像と前記癌マーカー染色画像とのマッチング位置を算出するマッチング手段と、
前記情報取得手段により取得された前記HE染色画像における腫瘍領域の情報および前記マッチング手段により算出された前記マッチング位置の情報に基づいて、前記癌マーカー染色画像における腫瘍領域を特定する特定手段と、
前記特定手段により特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞を検出する染色陽性細胞検出手段とを備えることを特徴とする。
診断対象となる前記病理画像の指定情報と検査種別の指定情報との入力を受け付ける入力受付手段と、
前記HE染色画像および前記癌マーカー染色画像が格納された染色画像データベースとを備え、
前記画像取得手段は、前記指定情報に基づいて、前記染色画像データベースから前記HE染色画像および前記癌マーカー染色画像を取得することが好ましい。
前記病理画像の指定情報は、前記HE染色画像の画像識別子であり、
前記画像取得手段は、前記染色画像データベースから、前記画像識別子を持つ前記HE染色画像と前記HE染色画像に隣接する前記癌マーカー染色画像とを取得することが好ましい。
前記病理画像の指定情報は、前記癌マーカー染色画像の画像識別子であり、
前記画像取得手段は、前記染色画像データベースから、前記画像識別子を持つ前記癌マーカー染色画像と前記癌マーカー染色画像に隣接する前記HE染色画像とを取得することが好ましい。
前記病理画像の指定情報は、診断対象者の被検者識別子であり、
前記画像取得手段は、前記染色画像データベースから、前記被検者識別子を持つ前記HE染色画像と前記癌マーカー染色画像とを取得することが好ましい。
前記染色画像データベースには、前記HE染色画像における腫瘍領域の情報も格納されており、
前記情報取得手段は、前記染色画像データベースから、前記HE染色画像における腫瘍領域の情報を取得することが好ましい。
前記情報取得手段は、前記腫瘍領域算出手段により算出された前記HE染色画像における腫瘍領域の情報を取得することが好ましい。
診断対象となるスライドのスライド識別子と検査種別の指定情報との入力を受け付ける入力受付手段と、
前記スライドが格納されたスライドデータベースと、
前記スライド識別子を持つ前記スライドを前記スライドデータベースから取得するスライド取得手段とを備え、
前記画像取得手段は、前記スライド取得手段により取得された前記スライドを撮影することにより、前記HE染色画像および前記癌マーカー染色画像を取得することが好ましい。
端末とサーバとを有し、
前記端末およびサーバは、システム外の通信網を介して接続可能であり、
前記端末は、
前記端末内の情報を前記通信網を介して前記サーバに送信する端末側送信手段と、
前記サーバから送信された情報を前記通信網を介して受信する端末側受信手段とを有し、
前記サーバは、
前記サーバ内の情報を前記通信網を介して前記端末に送信するサーバ側送信手段と、
前記端末から送信された情報を前記通信網を介して受信するサーバ側受信手段と、
診断対象となる前記病理画像として、HE染色画像および前記癌マーカー染色画像を取得する画像取得手段と、
前記HE染色画像における腫瘍領域の情報を取得する情報取得手段と、
前記画像取得手段により取得された、前記HE染色画像と前記癌マーカー染色画像とのマッチング位置を算出するマッチング手段と、
前記情報取得手段により取得された前記HE染色画像における腫瘍領域の情報および前記マッチング手段により算出された前記マッチング位置の情報に基づいて、前記癌マーカー染色画像における腫瘍領域を特定する特定手段と、
前記特定手段により特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞を検出する染色陽性細胞検出手段とを有し、
前記病理画像の情報が、前記端末側送信手段から前記サーバ側受信手段に送信され、かつ、前記サーバの前記染色陽性細胞検出手段により検出された染色陽性細胞の情報が、前記サーバ側送信手段から前記端末側受信手段に送信されることを特徴とする。
前記サーバは、
前記サーバ内の情報を前記通信網を介して端末に送信するサーバ側送信手段と、
前記端末から送信された情報を前記通信網を介して受信するサーバ側受信手段と、
診断対象となる前記病理画像として、HE染色画像および癌マーカー染色画像を取得する画像取得手段と、
前記HE染色画像における腫瘍領域の情報を取得する情報取得手段と、
前記画像取得手段により取得された、前記HE染色画像と前記癌マーカー染色画像とのマッチング位置を算出するマッチング手段と、
前記情報取得手段により取得された前記HE染色画像における腫瘍領域の情報および前記マッチング手段により算出された前記マッチング位置の情報に基づいて、前記癌マーカー染色画像における腫瘍領域を特定する特定手段と、
前記特定手段により特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞を検出する染色陽性細胞検出手段とを有することを特徴とする。
前記端末内の情報を前記通信網を介して前記サーバに送信する端末側送信手段と、
前記サーバから送信された情報を前記通信網を介して受信する端末側受信手段とを有し、
前記病理画像の情報が、前記端末側送信手段から前記サーバ側受信手段に送信され、かつ、前記サーバの前記染色陽性細胞検出手段により検出された染色陽性細胞の情報が、前記サーバ側送信手段から前記端末側受信手段に送信されることを特徴とする。
診断対象となる前記病理画像として、HE染色画像および癌マーカー染色画像を取得する画像取得工程と、
前記HE染色画像における腫瘍領域の情報を取得する情報取得工程と、
前記画像取得工程において取得された、前記HE染色画像と前記癌マーカー染色画像とのマッチング位置を算出するマッチング工程と、
前記情報取得工程において取得された前記HE染色画像における腫瘍領域の情報および前記マッチング工程において算出された前記マッチング位置の情報に基づいて、前記癌マーカー染色画像における腫瘍領域を特定する特定工程と、
前記特定工程において特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞を検出する染色陽性細胞検出工程とを含むことを特徴とする。
前記HE染色画像における腫瘍領域の情報を取得する情報取得工程と、
前記画像取得工程において取得された、前記HE染色画像と前記癌マーカー染色画像とのマッチング位置を算出するマッチング工程と、
前記情報取得工程において取得された前記HE染色画像における腫瘍領域の情報および前記マッチング工程において算出された前記マッチング位置の情報に基づいて、前記癌マーカー染色画像における腫瘍領域を特定する特定工程と、
前記特定工程において特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞を検出する染色陽性細胞検出工程とを、コンピュータに実行させることを特徴とする。
図19は、本発明の画像診断支援システムを備える画像診断支援装置の構成の一例を示すブロック図である。同図に示すように、画像診断支援装置190は、処理部191および記憶部192を備える。画像診断支援装置190は、処理部191において、顕微鏡193に接続されたCCD194、スキャナ195およびディスプレイ196に、接続されている。画像診断支援装置190は、例えば、CPU(Central Processing Unit)、RAM(Random Access Memory)、入力部、ドライブ、入出力インターフェース(I/F)、通信バス等を備える。CPUは、画像診断支援装置の全体の制御を担う。CPUに、例えば、前記各手段の各機能を提供するコンピュータプログラムを組み込むことで、画像診断支援装置190における前記各手段を構築でき、画像診断支援装置190を実現できる。また、画像診断支援装置190は、その動作を、前記各手段の機能を実現するコンピュータプログラムを組み込んだ、LSI(Large Scale Integration)等のハードウェア部品からなる回路部品を実装して、実現することもできる。このようなコンピュータプログラムは、これを格納した記録媒体等の形態であってもよい。前記記録媒体は、例えば、HDD、FD、CD-ROM(CD-R、CD-RW)、MO、DVD、メモリーカード等があげられる。また、記憶部192は、例えば、ROM、HDD、HD等である。HDDは、例えば、CPUの制御下、HDに対するデータの読み込みと書き込みを制御し、HDは、例えば、HDDの制御下、書き込まれたデータを記憶する。
図1は、本発明の画像診断支援システムの一例を示すブロック図である。このシステムは、癌の病理画像に基づく診断を支援するシステムである。前記システムは、診断対象となる病理画像として、HE染色画像およびmiRNA染色画像を取得する画像取得手段と、HE染色画像における腫瘍領域の情報を取得する情報取得手段と、画像取得手段により取得されたHE染色画像とmiRNA染色画像とのマッチング位置を算出するマッチング手段と、情報取得手段により取得されたHE染色画像における腫瘍領域の情報およびマッチング手段により算出されたマッチング位置の情報に基づいて、miRNA染色画像における腫瘍領域を特定する特定手段と、特定手段により特定されたmiRNA染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞含有率を算出する算出手段とを備える。
(a)診断対象となる病理画像として、HE染色画像およびmiRNA染色画像を取得する画像取得工程
(b)HE染色画像における腫瘍領域の情報を取得する情報取得工程
(c)画像取得工程において取得されたHE染色画像とmiRNA染色画像とのマッチング位置を算出するマッチング工程
(d)情報取得工程において取得されたHE染色画像における腫瘍領域の情報およびマッチング工程において算出されたマッチング位置の情報に基づいて、miRNA染色画像における腫瘍領域を特定する特定工程
(e)特定工程において特定されたmiRNA染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞含有率を算出する算出工程
図7は、本発明の画像診断支援システムの一例を示すブロック図である。本実施形態のシステムは、図1に示す実施形態1Bのシステムと比べて、染色陽性細胞含有率算出部125が、染色陽性細胞含有率に加えて染色強度をも算出する点で相違する。特に示さない限り、本実施形態において、その他の構成および動作は、実施形態1Bと同様である。
図11は、本発明による画像診断支援システムの一例を示すブロック図である。本実施形態のシステムは、図1に示す前記実施形態1Bに係るシステムと比べて、腫瘍判定および腫瘍領域算出部(腫瘍領域算出手段)1126を設けた点で相違する。特に示さない限り、本実施形態において、その他の構成および動作は、前記実施形態1Bと同様である。染色画像データベース113には、例えば、1つ以上のHE染色画像と、前記HE染色画像の標本に隣接した連続切片の標本であるmiRNA染色画像と、前記HE染色画像と前記miRNA染色画像との標本の隣接情報とが蓄積されている。本実施形態において、前記miRNA染色画像から算出された腫瘍領域情報、あるいは医師等によって判定された腫瘍領域情報は、あってもなくてもよい。
図12は、本発明による画像診断支援システムの一例を示すブロック図である。本実施形態のシステムは、図1等に示す染色画像および腫瘍領域情報取得部122に代えて、スライド撮影部(スライド取得手段)1222を備え、染色画像データベース113に代えて、スライドデータベース1213を備え、診断画像情報および検査対象記憶部131に代えて、診断スライド情報および検査対象記憶部1231を備える。さらに、このシステムには、スライド撮影装置1214および腫瘍判定および腫瘍領域算出部1126を備える。特に示さない限り、本実施形態において、その他の構成および動作は、前記実施形態1Bと同様である。
本発明の癌病理画像診断支援装置(以下、画像診断支援装置という)は、
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から、診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定される特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有し、
前記特徴決定手段は、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、決定した特徴が既知であるとの条件の下、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次決定し、
前記分類テーブル作成手段は、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴とクラス情報とを配置してなる前記分類テーブルを作成し、前記分類テーブルにより前記学習パターンを分類し、
前記特徴抽出手段は、前記特徴のセットを用いて、前記入力パターンの各特徴を算出し、
前記診断手段は、前記算出結果および前記分類テーブルに基づいて、前記入力パターンを診断することを特徴とする。
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有し、
前記特徴決定手段は、特徴の値に応じて遷移させるべき、予め定めた個数の前記学習パターンの集合を用意し、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、前記各学習パターンを、決定した特徴に応じた重みを付けて、分配するとともに、決定した特徴に対応する集合に、順次、遷移させ、前記各学習パターンが含まれる集合についての情報と、決定されている特徴とが既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定し、
前記分類テーブル作成手段は、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴とクラス情報とを配置してなる前記分類テーブルを作成し、前記分類テーブルにより前記学習パターンを分類し、
前記特徴抽出手段は、前記特徴のセットを用いて、前記入力パターンの各特徴を算出し、
前記診断手段は、前記入力パターンの各特徴と、前記特徴のセットの各特徴の決定の際に前記学習パターンが属する集合を逐次記録してなる遷移テーブルとに基づいて、前記入力パターンの遷移を行って、前記遷移の結果、前記入力パターンが属する集合に基づいて前記入力パターンを診断することを特徴とする。
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有し、
前記特徴決定手段は、特徴の値に応じて遷移させるべき、予め定めた個数の前記学習パターンの集合を用意し、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、前記各学習パターンを、決定した特徴に応じた重みを付けて、分配するとともに、決定した特徴に対応する集合に、順次、遷移させ、前記各学習パターンが含まれる集合の情報と、決定されている特徴とが既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定し、
前記分類テーブル作成手段は、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴と、クラス情報とを配置してなる前記分類テーブルにより前記学習パターンを分類し、
前記特徴抽出手段は、前記特徴のセットを用いて、前記順位の特徴が予め定めた値となる確率を示している、前記入力パターンの各特徴を算出し、
前記診断手段は、前記入力パターンの各特徴と、前記特徴のセットの各特徴の決定の際に前記学習パターンが属する集合を逐次記録してなる遷移テーブルとに基づいて前記入力パターンが予め定められたクラス情報を有する確率を算出し、診断することを特徴とする。
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有する画像診断支援装置の画像診断支援プログラムであって、
前記特徴決定手段に、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターン集合のクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、決定した特徴が既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定する処理を実行させ、
前記分類テーブル作成手段に、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴と、クラス情報とを配置してなる前記分類テーブルにより前記パターンを分類する処理を実行させ、
前記特徴抽出手段に、前記特徴のセットを用いて、前記入力パターンの各特徴を算出する処理を実行させ、
前記診断手段に、前記算出結果および前記分類テーブルに基づいて、前記入力パターンを診断する処理を実行させることを特徴とする。具体的には、本発明のプログラムは、前記画像診断支援装置の各種手段に、前述のような処理工程を実行させるためのプログラムである。
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有する画像診断支援装置の画像診断支援プログラムであって、
前記特徴決定手段に、特徴の値に応じて遷移させるべき、予め定めた個数の前記学習パターンの集合を用意し、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、前記各学習パターンを、決定した特徴に応じた重みを付けて分配するとともに、決定した特徴に対応する集合に、順次、遷移させ、前記各学習パターンが含まれる集合についての情報と、決定されている特徴とが既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定する処理を実行させ、
前記分類テーブル作成手段に、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴と、クラス情報とを配置してなる前記分類テーブルを作成して、前記分類テーブルにより前記学習パターンを分類する処理を実行させ、
前記特徴抽出手段に、前記特徴のセットを用いて、前記入力パターンの各特徴を算出する処理を実行させ、
前記診断手段に、前記入力パターンの各特徴と、前記特徴のセットの各特徴の決定の際に前記学習パターンが属する集合を逐次記録してなる遷移テーブルと、に基づいて前記入力パターンの遷移を行って、前記遷移の結果、前記入力パターンが属する集合に基づいて前記入力パターンを診断する処理を実行させることを特徴とする。具体的には、本発明のプログラムは、前記画像診断支援装置の各種手段に、前述のような処理工程を実行させるためのプログラムである。
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有する画像診断支援装置の画像診断支援プログラムであって、
前記特徴決定手段に、前記学習パターンを特徴の値に応じて遷移させるべき、予め定めた個数の集合を用意し、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、前記各学習パターンを、決定した特徴に応じた重みを付けて分配するとともに、決定した特徴に対応する集合に、順次、遷移させ、前記各学習パターンが含まれる集合の情報と、決定されている特徴とが既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定する処理を実行させ、
前記分類テーブル作成手段に、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴と、クラス情報とを配置してなる前記分類テーブルにより前記学習パターンを分類する処理を実行させ、
前記特徴抽出手段に、前記特徴のセットを用いて、前記順位の特徴が予め定めた値となる確率を示している、前記入力パターンの各特徴を算出する処理を実行させ、
前記診断手段に、前記入力パターンの各特徴と、前記特徴のセットの各特徴の決定の際に前記学習パターンが属する集合を逐次記録してなる遷移テーブルとに基づいて前記入力パターンが予め定められたクラス情報を有する確率を算出し、診断する処理を実行させることを特徴とする。具体的には、本発明のプログラムは、前記画像診断支援装置の各種手段に、前述のような処理工程を実行させるためのプログラムである(以下、同様)。
前記学習パターン入力手段および前記パターン入力手段に、
染色された前記病理画像中の各画素のRGB値から、予め設定された腫瘍細胞核が属する色領域に属する画素を選択する処理と、
前記色領域の分布の中心と前記色領域に属する各画素との距離を算出する処理と、
前記距離に応じて前記各画素に信号を付与する処理と、
前記病理画像中における前記信号の分布からピークを検出する処理と、
前記ピークを中心としたイメージを前記学習パターンとして入力する処理とを実行させることが好ましい。
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有する画像診断支援装置を用いた画像診断支援方法であって、
前記特徴決定手段が、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、決定した特徴が既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定するステップを実施し、
前記分類テーブル作成手段が、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴と、クラス情報と、を配置してなる前記分類テーブルにより前記パターンを分類するステップを実施し、
前記特徴抽出手段が、前記特徴のセットを用いて、前記入力パターンの各特徴を算出するステップを実施し、
前記診断手段が、前記算出結果および前記分類テーブルに基づいて、前記入力パターンを診断するステップを実施することを特徴とする。なお、本発明の方法は、前記画像診断支援装置の使用は必須ではなく、前述のような各種ステップを実施する方法であってもよい。
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有する画像診断支援装置を用いた画像診断支援方法であって、
前記特徴決定手段が、特徴の値に応じて遷移させるべき、予め定めた個数の前記学習パターンの集合を用意し、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、前記各学習パターンを、決定した特徴に応じた重みを付けて、分配するとともに、決定した特徴に対応する集合に、順次、遷移させ、前記各学習パターンが含まれる集合の情報と、決定されている特徴とが既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定するステップを実施し、
前記分類テーブル作成手段が、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴と、クラス情報とを配置してなる前記分類テーブルを作成し、前記分類テーブルにより前記学習パターンを分類するステップを実施し、
前記特徴抽出手段が、前記特徴のセットを用いて、前記入力パターンの各特徴を算出するステップを実施し、
前記診断手段が、前記入力パターンの各特徴と、前記特徴のセットの各特徴の決定の際に前記学習パターンが属する集合を逐次記録してなる遷移テーブルと、に基づいて前記入力パターンの遷移を行って、前記遷移の結果、前記入力パターンが属する集合に基づいて前記入力パターンを診断するステップを実施することを特徴とする。本発明の方法は、前記画像診断支援装置の使用は必須ではなく、前述のような各種ステップを実施する方法であってもよい。
学習に用いる病理画像から腫瘍を中心としたイメージを切り出し、学習パターンとして入力する学習パターン入力手段と、
クラス情報が付された前記学習パターンを記憶保持する学習パターン記憶手段と、
複数の特徴候補を生成する特徴候補生成手段と、
前記特徴候補生成手段によって生成される特徴候補の中から診断に適した特徴のセットを決定する特徴決定手段と、
前記特徴決定手段により決定されている特徴のセットを記憶保持する特徴記憶手段と、
分類テーブルを作成する分類テーブル作成手段と、
診断する病理画像から腫瘍候補を中心としたイメージを切り出し、入力パターンとして入力するパターン入力手段と、
前記入力パターンから特徴を抽出する特徴抽出手段と、
前記特徴に基づいて診断する診断手段とを有する画像診断支援装置を用いた画像診断支援方法であって、
前記特徴決定手段は、特徴の値に応じて遷移させるべき、予め定めた個数の前記学習パターンの集合を用意し、前記各特徴候補に対応する各学習パターンの特徴を計算して、前記学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの最初の特徴として決定し、前記各学習パターンを、決定した特徴に応じた重みを付けて、分配するとともに、決定した特徴に対応する集合に、順次、遷移させ、前記各学習パターンが含まれる集合についての情報と、決定されている特徴とが既知であるとの条件の下で、各特徴候補に対応する各学習パターンの特徴と、前記各学習パターンのクラス情報との相互情報量が最大となる特徴候補を、特徴のセットの次なる特徴として、順次、決定するステップを実施し、
前記分類テーブル作成手段が、前記特徴のセットを用いて、前記各学習パターンの各特徴を計算し、前記各学習パターンの各特徴と、クラス情報とを配置してなる前記分類テーブルを作成し、前記分類テーブルにより前記学習パターンを分類するステップを実施し、
前記特徴抽出手段は、前記特徴のセットを用いて、前記順位の特徴が予め定めた値となる確率を示している、前記入力パターンの各特徴を算出するステップを実施し、
前記診断手段は、前記入力パターンの各特徴と、前記特徴のセットの各特徴の決定の際に前記学習パターンが属する集合を逐次記録してなる遷移テーブルと、に基づいて前記入力パターンが予め定められたクラス情報を有する確率を算出し、診断するステップを実施することを特徴とする。本発明の方法は、前記画像診断支援装置の使用は必須ではなく、前述のような各種ステップを実施する方法であってもよい。
染色された前記病理画像中の各画素のRGB値から、予め設定された腫瘍細胞核が属する色領域に属する画素を選択するステップを実施し、
前記色領域の分布の中心と前記色領域に属する各画素との距離を算出するステップと、
前記距離に応じて前記各画素に信号を付与するステップを実施し、
前記病理画像中における前記信号の分布からピークを検出するステップを実施し、
前記ピークを中心としたイメージを前記学習パターンとして入力するステップを実施することが好ましい。
病理画像に患者固有の情報が付加された病理画像データを保持する情報処理端末と、前記病理画像データを診断する画像診断サーバとを有し、
前記画像診断サーバは、
前記病理画像データが有する前記病理画像を診断する前記本発明の画像診断支援装置と、
前記画像診断支援装置による診断結果を前記患者固有の情報とともに記憶する診断結果記憶手段とを有し、
前記情報処理端末は、前記患者固有の情報を伴って前記診断結果の送信を要求し、
前記画像診断サーバは、前記情報処理端末から受信した前記患者固有の情報と前記診断結果とともに記憶された前記患者固有の情報とを比較し、前記情報処理端末から受信した前記患者固有の情報と前記診断結果とともに記憶された前記患者固有の情報とが一致したとき、前記診断結果を前記情報処理端末に送信することを特徴とする。
前記サーバは、
前記サーバ内の情報を通信網を介して端末に送信するサーバ側送信手段と、
前記端末から送信された情報を前記通信網を介して受信するサーバ側受信手段と、
前記病理画像データを用いて被検者を診断する前記本発明の画像診断支援装置と、
前記画像診断支援装置による診断結果を前記患者固有の情報とともに記憶する診断結果記憶手段とを有し、
前記情報処理端末から受信した前記患者固有の情報と前記診断結果とともに記憶された前記患者固有の情報とを比較し、前記情報処理端末から受信した前記患者固有の情報と前記診断結果とともに記憶された前記患者固有の情報とが一致したとき、前記診断結果を前記情報処理端末に送信することを特徴とする。
病理画像に患者固有の情報が付加された病理画像データを保持する情報処理端末であり、前記端末内の情報を前記通信網を介して前記サーバに送信する端末側送信手段と、
前記サーバから送信された情報を前記通信網を介して受信する端末側受信手段とを有し、
前記患者固有の情報を伴った前記診断結果の送信を要求し、前記サーバから送信された前記診断結果を受信することを特徴とする。
図13は、本実施形態に係る画像診断支援装置の構成を示すブロック図である。図13に示すように、本実施形態に係る画像診断支援装置は、学習パターン入力手段1300、学習パターン記憶手段1301、特徴候補生成手段1302、特徴決定手段1303、特徴記憶手段1304、分類テーブル作成手段1305、および分類テーブル1306を有している。
特徴候補生成手段1302は、まず、s番目の特徴パラメータセットを読み出す。s=1~Nであり、s=1からスタートする。s≦N_1の場合は、s番目の特徴パラメータセット(k_s、r0_s、σ_s、th_s)を(k、r0、σ、th)に代入し、パラメータk、r0、σで規定される下記式(1)に例示する複素ガボール関数Gab、およびガウス関数Gを生成する。なお、s≦N_1の場合、学習パターンとして、学習パターン記憶手段1301に記憶されたカラーのサブイメージをグレースケール変換したグレースケール画像を用いて特徴cを算出する。
特徴候補生成手段1302は、まず、s番目の特徴パラメータセットを読み出す。s=1~Nであり、s=1からスタートする。N_1+1≦s≦N_1+N_2の場合は、s番目の特徴パラメータセット(x_s、y_s、color_index)および特徴候補の識別番号sを、特徴決定手段1303に送る(ステップS1402)。
特徴候補生成手段1302は、まず、s番目の特徴パラメータセットを読み出す。s=1~Nであり、s=1からスタートする。N_1+N_2+1≦s≦Nの場合は、s番目の特徴パラメータセット(x_s、y_s、color_index、th_s)および特徴候補の識別番号sを、特徴決定手段1303に送る(ステップS1402)。
本実施形態3は、前記第1の形態または第2の形態の画像診断支援システムにおいて、さらに、miRNA染色画像の染色状態の補正を行う補正手段、補正したmiRNA染色画像における非腫瘍細胞を検出する非腫瘍細胞検出手段、および、検出した非腫瘍細胞のmiRNA染色の有無を判定する判定手段を有する。
プローブを用いてin situ ハイブリダイゼーションを行い、急性骨髄性白血病(AML)患者(n=4)および急性リンパ性白血病(ALL)患者(n=2)の白血球におけるhsa-miR-92aの発現レベルを調べた。
hsa-miR-92a検出用プローブ(配列番号5)
5’-acaggccgggacaagtgcaata-3’
陰性コントロール用プローブ(配列番号6)
5’-gtgtaacacgtctatacgccca-3’
乳房から採取した組織を用いた以外は、前記実施例1と同様にして、切片を作成し、hsa-miR-92aをin situハイブリダイゼーションにより検出した。その結果を、図23に示す。図23の(A)~(D)は、異なる部分から採取した乳房組織についてのmiRNA染色の結果を示す写真である。
肝細胞から採取した組織を用いた以外は、前記実施例1と同様にして、切片を作成し、hsa-miR-92aをin situハイブリダイゼーションにより検出した。
112 出力装置
113 染色画像データベース
120 処理装置
121 入力受付処理部
122 情報取得部
123 画像マッチング処理部
124 腫瘍領域抽出部
125 染色陽性細胞含有率算出部
130 記憶装置
131、132、133、134、135、136、1231 記憶部
725 染色陽性細胞含有率および染色度算出部
1126 腫瘍判定および腫瘍領域算出部
1213 スライドデータベース
1214 スライド装置
1222 スライド撮影部
1300 学習パターン入力手段
1301 学習パターン記憶手段
1302 特徴候補生成手段
1303 特徴決定手段
1304 特徴記憶手段
1305 分類テーブル作成手段
1306 分類テーブル
1701 パターン入力手段
1702 特徴抽出手段
1703 診断手段
190 画像診断支援装置
191 処理部
192 記憶部
193 顕微鏡
194 CCD
195 スキャナ
196 ディスプレイ
2001 画像取得手段
2002 情報取得手段
2003 マッチング手段
2004 腫瘍領域の特定手段
2005 算出手段
Claims (13)
- 試料における癌マーカーを検出する癌マーカー検出工程と、
前記癌マーカー検出工程において検出した前記癌マーカーの発現レベルに基づいて、前記試料の癌の可能性を評価する工程とを含み、
前記試料が、細胞または組織であり、
前記癌マーカーが、hsa-miR-92およびhsa-miR-494の少なくとも一方のmiRNAを含むことを特徴とする、癌の可能性を評価する評価方法。 - 前記hsa-miR-92が、hsa-miR-92a、hsa-miR-92a*、hsa-miR-92bおよびhsa-miR-92b*からなる群から選択された少なくとも一つである、請求の範囲1記載の評価方法。
- 前記hsa-miR-92aが、hsa-miR-92a-1およびhsa-miR-92a-2の少なくとも一方である、請求の範囲2記載の評価方法。
- 前記hsa-miR-92a*が、hsa-miR-92a-1*およびhsa-miR-92a-2*の少なくとも一方である、請求の範囲2記載の評価方法。
- 前記miRNAが、hsa-miR-92aである、請求の範囲1記載の評価方法。
- 前記癌が、大腸癌、直腸癌、胆嚢癌、胃癌、乳癌、白血病、膵癌、肝臓癌、脳腫瘍および骨肉腫からなる群から選択された少なくとも一つの癌である、請求の範囲1から5のいずれか一項に記載の評価方法。
- 前記評価方法が、癌の発症の有無、癌の進行度または予後の状態を判断する評価方法である、請求の範囲1から6のいずれか一項に記載の評価方法。
- 前記癌マーカー検出工程において、前記癌マーカーを、発色、蛍光およびオートラジオグラフィーからなる群から選択された少なくとも一つにより可視化する、請求の範囲1から7のいずれか一項に記載の評価方法。
- 前記試料を固定化し、in situハイブリダイゼーション法により前記癌マーカーを検出する、請求の範囲1から8のいずれか一項に記載の評価方法。
- 前記癌マーカーの発現レベルが、前記試料における前記癌マーカーの発現量で表わされる、請求の範囲1から9のいずれか一項に記載の評価方法。
- 前記評価工程において、検出した前記癌マーカーの発現レベルに基づいて、下記(1)、(2)および(3)からなる群から選択された少なくとも一つの方法により、癌の可能性を評価する、請求の範囲1から10のいずれか一項に記載の評価方法。
(1)被検者の試料における前記癌マーカーの発現レベルを、正常者の試料における前記癌マーカーの発現レベルと比較し、前記正常者の発現レベルよりも高い場合に、前記被検者は、前記癌の可能性が高いと決定する。
(2)被検者の試料における前記癌マーカーの発現レベルを、正常者の試料における前記癌マーカーの発現レベルと比較し、前記正常者の発現レベルよりも相対的に高い程、前記被検者は、前記癌が相対的に進行していると決定する。
(3)被検者の試料における前記癌マーカーの発現レベルを、進行期別の各癌患者の試料における前記癌マーカーの発現レベルと比較し、前記被検者は、同程度の発現レベルを示す患者と同じ進行期であると決定する。 - 前記癌マーカー検出工程において、固定化した前記試料について、前記癌マーカーを染色した癌マーカー染色画像を取得し、
さらに、
前記固定化した試料について、HE染色画像を取得するHE染色画像取得工程、
前記HE染色画像における腫瘍領域の情報を取得する情報取得工程、
前記HE画像取得工程において取得された前記HE染色画像と、前記癌マーカー検出工程において取得された前記癌マーカー染色画像とのマッチング位置を算出するマッチング工程、
前記情報取得工程において取得された前記HE染色画像における腫瘍領域の情報および前記マッチング工程において算出された前記マッチング位置の情報に基づいて、前記癌マーカー染色画像における腫瘍領域を特定する特定工程、および、
前記特定工程において特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞を検出する染色陽性細胞検出工程を含む、
請求の範囲9から11のいずれか一項に記載の評価方法。 - 前記染色陽性細胞検出工程が、前記特定工程において特定された前記癌マーカー染色画像における腫瘍領域の情報に基づいて、前記腫瘍領域内の染色陽性細胞含有率を算出する算出工程である、請求の範囲12記載の評価方法。
Priority Applications (5)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN2010800179242A CN102421916A (zh) | 2009-04-21 | 2010-04-21 | 用于评估癌症的方法 |
EP10767100A EP2423323A4 (en) | 2009-04-21 | 2010-04-21 | METHOD OF TUMOR ASSESSMENT |
JP2011510351A JPWO2010123043A1 (ja) | 2009-04-21 | 2010-04-21 | 癌の評価方法 |
US13/265,680 US20120115139A1 (en) | 2009-04-21 | 2010-04-21 | Method for evaluating cancer |
US14/032,651 US20140178874A1 (en) | 2009-04-21 | 2013-09-20 | Method for evaluating cancer |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2009-103332 | 2009-04-21 | ||
JP2009103332 | 2009-04-21 |
Related Child Applications (2)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US13/265,680 A-371-Of-International US20120115139A1 (en) | 2009-04-21 | 2010-04-21 | Method for evaluating cancer |
US14/032,651 Continuation US20140178874A1 (en) | 2009-04-21 | 2013-09-20 | Method for evaluating cancer |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2010123043A1 true WO2010123043A1 (ja) | 2010-10-28 |
Family
ID=43011159
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2010/057102 WO2010123043A1 (ja) | 2009-04-21 | 2010-04-21 | 癌の評価方法 |
Country Status (5)
Country | Link |
---|---|
US (2) | US20120115139A1 (ja) |
EP (1) | EP2423323A4 (ja) |
JP (1) | JPWO2010123043A1 (ja) |
CN (1) | CN102421916A (ja) |
WO (1) | WO2010123043A1 (ja) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2013188208A (ja) * | 2012-02-14 | 2013-09-26 | Okayama Univ | Atllの診断のためのデータ取得方法、atll診断用キットおよびatll診断システム |
JP2016518813A (ja) * | 2013-03-15 | 2016-06-30 | ベンタナ メディカル システムズ, インコーポレイテッド | デジタル・ホール・スライドの自動採点のための組織物体に基づく機械学習システム |
CN105734155A (zh) * | 2016-04-27 | 2016-07-06 | 固安博健生物技术有限公司 | 软骨母细胞型骨肉瘤致病基因及其应用 |
KR20170018412A (ko) | 2014-06-13 | 2017-02-17 | 도레이 카부시키가이샤 | 유방암 검출 키트 또는 디바이스 및 검출 방법 |
KR20170018411A (ko) | 2014-06-18 | 2017-02-17 | 도레이 카부시키가이샤 | 간암의 검출 키트 또는 디바이스 및 검출 방법 |
JP2022512981A (ja) * | 2018-11-09 | 2022-02-07 | エム・エム・18 メディカル アー・ベー | 生検サンプル画像の異なるカテゴリを識別するための方法 |
Families Citing this family (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103003840A (zh) * | 2010-09-30 | 2013-03-27 | 富士通先端科技株式会社 | 登记程序、登记装置以及登记方法 |
US8977017B2 (en) * | 2011-09-15 | 2015-03-10 | The General Hospital Corporation | System and method for support of medical diagnosis |
US9798918B2 (en) | 2012-10-05 | 2017-10-24 | Cireca Theranostics, Llc | Method and system for analyzing biological specimens by spectral imaging |
US9396532B2 (en) * | 2012-11-15 | 2016-07-19 | Siemens Healthcare Diagnostics, Inc. | Cell feature-based automatic circulating tumor cell detection |
CA2907405A1 (en) * | 2013-03-19 | 2014-09-25 | Cireca Theranostics, Llc | Method and system for analyzing biological specimens by spectral imaging |
CN112391468B (zh) * | 2013-07-09 | 2024-02-06 | 中央兰开夏大学 | 脑癌检测 |
CN104195238B (zh) * | 2014-08-15 | 2017-08-01 | 深圳市晋百慧生物有限公司 | 用于检测肠癌的标记物及其应用 |
CN104498496B (zh) * | 2014-11-27 | 2017-11-14 | 深圳市第二人民医院 | 小分子novel‑miR‑1和骨肉瘤药物 |
CN104372007B (zh) * | 2014-11-27 | 2017-12-22 | 深圳市第二人民医院 | 骨肉瘤的生物标记物及其应用 |
GB2543029A (en) * | 2015-09-23 | 2017-04-12 | Pathxl Ltd | Method and apparatus for tissue recognition |
US11021756B2 (en) | 2016-01-04 | 2021-06-01 | Yang Zuozhang | MiRNA markers for the diagnosis of osteosarcoma |
WO2017221592A1 (ja) * | 2016-06-23 | 2017-12-28 | コニカミノルタ株式会社 | 画像処理装置、画像処理方法及び画像処理プログラム |
WO2018216618A1 (ja) | 2017-05-25 | 2018-11-29 | 日本電気株式会社 | 情報処理装置、制御方法、及びプログラム |
CN107217055B (zh) * | 2017-06-26 | 2019-01-18 | 生工生物工程(上海)股份有限公司 | 一种癌症诊断芯片及其试剂盒 |
WO2019208703A1 (ja) * | 2018-04-26 | 2019-10-31 | 日本電気株式会社 | 情報処理装置、制御方法、及びプログラム |
CN109609630B (zh) * | 2018-12-03 | 2023-01-13 | 哈尔滨医科大学 | 用于早期胃癌诊断的分子标志物及其应用 |
CN112132166B (zh) * | 2019-06-24 | 2024-04-19 | 杭州迪英加科技有限公司 | 一种数字细胞病理图像智能分析方法、系统及装置 |
US20230142909A1 (en) * | 2020-04-10 | 2023-05-11 | Koninklijke Philips N.V. | Clinically meaningful and personalized disease progression monitoring incorporating established disease staging definitions |
CN112798567B (zh) * | 2021-02-01 | 2022-03-18 | 山东大学 | 一种基于吖啶橙和碳点的比率荧光用于miRNA体外检测的方法 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007081680A2 (en) * | 2006-01-05 | 2007-07-19 | The Ohio State University Research Foundation | Microrna expression abnormalities in pancreatic endocrine and acinar tumors |
WO2008108059A1 (ja) * | 2007-03-01 | 2008-09-12 | Nec Corporation | 乳癌病理画像診断支援システム、乳癌病理画像診断支援方法、乳癌病理画像診断支援プログラム、及び、乳癌病理画像診断支援プログラムを記録した記録媒体 |
JP2009103332A (ja) | 2007-10-22 | 2009-05-14 | Panasonic Corp | 加熱調理器 |
WO2009133915A1 (ja) * | 2008-04-30 | 2009-11-05 | 日本電気株式会社 | 癌マーカー、それを用いた癌の評価方法および評価試薬 |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2471923B1 (en) * | 2004-05-28 | 2014-08-20 | Asuragen, Inc. | Methods and compositions involving microRNA |
US20080269072A1 (en) * | 2004-10-21 | 2008-10-30 | Hart Ronald P | Rational Probe Optimization for Detection of MicroRNAs |
ES2420973T3 (es) * | 2007-07-25 | 2013-08-28 | University Of Louisville Research Foundation, Inc. | Micro-ARN asociado a exosoma como marcador de diagnóstico |
-
2010
- 2010-04-21 US US13/265,680 patent/US20120115139A1/en not_active Abandoned
- 2010-04-21 CN CN2010800179242A patent/CN102421916A/zh active Pending
- 2010-04-21 JP JP2011510351A patent/JPWO2010123043A1/ja active Pending
- 2010-04-21 WO PCT/JP2010/057102 patent/WO2010123043A1/ja active Application Filing
- 2010-04-21 EP EP10767100A patent/EP2423323A4/en not_active Withdrawn
-
2013
- 2013-09-20 US US14/032,651 patent/US20140178874A1/en not_active Abandoned
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2007081680A2 (en) * | 2006-01-05 | 2007-07-19 | The Ohio State University Research Foundation | Microrna expression abnormalities in pancreatic endocrine and acinar tumors |
WO2008108059A1 (ja) * | 2007-03-01 | 2008-09-12 | Nec Corporation | 乳癌病理画像診断支援システム、乳癌病理画像診断支援方法、乳癌病理画像診断支援プログラム、及び、乳癌病理画像診断支援プログラムを記録した記録媒体 |
JP2009103332A (ja) | 2007-10-22 | 2009-05-14 | Panasonic Corp | 加熱調理器 |
WO2009133915A1 (ja) * | 2008-04-30 | 2009-11-05 | 日本電気株式会社 | 癌マーカー、それを用いた癌の評価方法および評価試薬 |
Non-Patent Citations (12)
Title |
---|
CALIN GA; DUMITRU CD; SHIMIZU ME: "Frequent deletions and down-regulation of micro-RNA genes miR15 and miR16 at 13ql4 in chronic lymphocytic leukemia", PROC NATL ACAD SCI USA, vol. 99, 2002, pages 15524 - 9 |
CALIN,G.A. ET AL.: "MicroRNA profiling reveals distinct signatures in B cell chronic lymphocytic leukemias", PROC.NATL.ACAD.SCI.USA, vol. 101, no. 32, 2004, pages 11755 - 60 * |
DUAN,H. ET AL.: "MiR-320 and miR-494 affect cell cycles of primary murine bronchial epithelial cells exposed to benzo[a]pyrene", TOXICOL.IN VITRO., vol. 24, no. 3, 2010, pages 928 - 35 * |
LAGOS-QUINTANA ET AL., CURR BIOL, vol. 12, 2002, pages 735 - 739 |
LUI,W.O. ET AL.: "Patterns of known and novel small RNAs in human cervical cancer", CANCER RES., vol. 67, no. 13, 2007, pages 6031 - 43, XP002544106 * |
MICHAEL MZ; SM OC; VAN HOLST PELLEKAAN NG; YOUNG GP; JAMES RJ: "Reduced accumulation of specific microRNAs in colorectal neoplasia", MOL CANCER RES, vol. 1, 2003, pages 882 - 91 |
NEILSON ET AL., GENES DEV, vol. 21, 2007, pages 578 - 589 |
OBEMOSTER ET AL., RNA, vol. 12, 2006, pages 1161 - 1167 |
PABLO LANDGRAF ET AL., CELL, vol. 129, 2007, pages 1401 - 1414 |
RUBY ET AL., CELL, vol. 127, 2006, pages 1193 - 1207 |
See also references of EP2423323A4 |
WU H. ET AL., PLOS ONE, vol. 2, no. 10, 2007, pages EL 020 |
Cited By (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2013188208A (ja) * | 2012-02-14 | 2013-09-26 | Okayama Univ | Atllの診断のためのデータ取得方法、atll診断用キットおよびatll診断システム |
JP2016518813A (ja) * | 2013-03-15 | 2016-06-30 | ベンタナ メディカル システムズ, インコーポレイテッド | デジタル・ホール・スライドの自動採点のための組織物体に基づく機械学習システム |
US10176579B2 (en) | 2013-03-15 | 2019-01-08 | Ventana Medical Systems, Inc. | Tissue object-based machine learning system for automated scoring of digital whole slides |
KR20220070557A (ko) | 2014-06-13 | 2022-05-31 | 도레이 카부시키가이샤 | 유방암 검출 키트 또는 디바이스 및 검출 방법 |
US11859255B2 (en) | 2014-06-13 | 2024-01-02 | Toray Industries, Inc. | Breast cancer detection kit or device, and detection method |
KR20170018412A (ko) | 2014-06-13 | 2017-02-17 | 도레이 카부시키가이샤 | 유방암 검출 키트 또는 디바이스 및 검출 방법 |
KR20230113854A (ko) | 2014-06-13 | 2023-08-01 | 도레이 카부시키가이샤 | 유방암 검출 키트 또는 디바이스 및 검출 방법 |
KR20230042128A (ko) | 2014-06-13 | 2023-03-27 | 도레이 카부시키가이샤 | 유방암 검출 키트 또는 디바이스 및 검출 방법 |
US10597726B2 (en) | 2014-06-13 | 2020-03-24 | Toray Industries, Inc. | Breast cancer detection kit or device, and detection method |
EP3816303A2 (en) | 2014-06-13 | 2021-05-05 | Toray Industries, Inc. | Breast cancer detection kit or device, and method for detecting breast cancer |
US11479822B2 (en) | 2014-06-13 | 2022-10-25 | Toray Industries, Inc. | Breast cancer detection kit or device, and detection method |
KR20220092645A (ko) | 2014-06-18 | 2022-07-01 | 도레이 카부시키가이샤 | 간암의 검출 키트 또는 디바이스 및 검출 방법 |
EP3862439A2 (en) | 2014-06-18 | 2021-08-11 | Toray Industries, Inc. | Liver cancer detection kit or device, and detection method |
US11512355B2 (en) | 2014-06-18 | 2022-11-29 | Toray Industries, Inc. | Liver cancer detection kit or device, and detection method |
US10590487B2 (en) | 2014-06-18 | 2020-03-17 | Toray Industries, Inc. | Liver cancer detection kit or device, and detection method |
KR20230053005A (ko) | 2014-06-18 | 2023-04-20 | 도레이 카부시키가이샤 | 간암의 검출 키트 또는 디바이스 및 검출 방법 |
KR20170018411A (ko) | 2014-06-18 | 2017-02-17 | 도레이 카부시키가이샤 | 간암의 검출 키트 또는 디바이스 및 검출 방법 |
KR20230146105A (ko) | 2014-06-18 | 2023-10-18 | 도레이 카부시키가이샤 | 간암의 검출 키트 또는 디바이스 및 검출 방법 |
US11827941B2 (en) | 2014-06-18 | 2023-11-28 | Toray Industries, Inc. | Liver cancer detection kit or device, and detection method |
KR20240036134A (ko) | 2014-06-18 | 2024-03-19 | 도레이 카부시키가이샤 | 간암의 검출 키트 또는 디바이스 및 검출 방법 |
KR20240105494A (ko) | 2014-06-18 | 2024-07-05 | 도레이 카부시키가이샤 | 간암의 검출 키트 또는 디바이스 및 검출 방법 |
CN105734155A (zh) * | 2016-04-27 | 2016-07-06 | 固安博健生物技术有限公司 | 软骨母细胞型骨肉瘤致病基因及其应用 |
JP2022512981A (ja) * | 2018-11-09 | 2022-02-07 | エム・エム・18 メディカル アー・ベー | 生検サンプル画像の異なるカテゴリを識別するための方法 |
JP7569786B2 (ja) | 2018-11-09 | 2024-10-18 | エム・エム・18 メディカル アー・ベー | 生検サンプル画像の異なるカテゴリを識別するための方法 |
Also Published As
Publication number | Publication date |
---|---|
US20120115139A1 (en) | 2012-05-10 |
CN102421916A (zh) | 2012-04-18 |
EP2423323A4 (en) | 2012-11-28 |
JPWO2010123043A1 (ja) | 2012-10-25 |
US20140178874A1 (en) | 2014-06-26 |
EP2423323A1 (en) | 2012-02-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2010123043A1 (ja) | 癌の評価方法 | |
Fischer et al. | Intraductal papillary mucinous neoplasms arise from multiple independent clones, each with distinct mutations | |
Graham et al. | Gene expression profiles of estrogen receptor–positive and estrogen receptor–negative breast cancers are detectable in histologically normal breast epithelium | |
CA2623775A1 (en) | Methods and materials for identifying the origin of a carcinoma of unknown primary origin | |
EP2304630A1 (en) | Molecular markers for cancer prognosis | |
Tsuchiya | Fluorescence in situ hybridization | |
US20200109457A1 (en) | Chromosomal assessment to diagnose urogenital malignancy in dogs | |
US20230257826A1 (en) | Methods for predicting prostate cancer and uses thereof | |
US20220042106A1 (en) | Systems and methods of using cell-free nucleic acids to tailor cancer treatment | |
CN106574307A (zh) | 用于评估前列腺癌的进展的材料和方法 | |
US20150329911A1 (en) | Nucleic acid biomarkers for prostate cancer | |
CN102027128A (zh) | 作为癌症标志物的结肠癌相关的转录物1(ccat-1) | |
CN110607370B (zh) | 一种用于人体肿瘤分子分型的基因组合及其应用 | |
Portier et al. | From morphologic to molecular: established and emerging molecular diagnostics for breast carcinoma | |
CA2722424A1 (en) | Cytological methods for detecting cancer | |
WO2015121663A1 (en) | Biomarkers for prostate cancer | |
US20220042108A1 (en) | Systems and methods of assessing breast cancer | |
Tsuchiya et al. | FISH Testing of Cytology Specimens: Pre-analytic, Analytic, and Post-analytic Considerations | |
CN117144003A (zh) | 用于mrd检测的癌种特异核心探针组的设计方法及其应用 | |
Chand | The prognostic role of extramural venous invasion in post-chemoradiotherapy rectal cancer | |
Albertson et al. | Comparative genomic hybridization | |
US20140172317A1 (en) | Method and apparatus for performing drug screening | |
AU2004219989A1 (en) | Expression profiling of tumours |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 201080017924.2 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 10767100 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2011510351 Country of ref document: JP Kind code of ref document: A |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2010767100 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13265680 Country of ref document: US |