Classifications

• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
  • A61B10/00—Instruments for taking body samples for diagnostic purposes; Other methods or instruments for diagnosis, e.g. for vaccination diagnosis, sex determination or ovulation-period determination; Throat striking implements
  • A61B10/02—Instruments for taking cell samples or for biopsy
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
  • G01N15/10—Investigating individual particles
  • G01N15/14—Optical investigation techniques, e.g. flow cytometry
  • G01N15/1429—Signal processing
  • G01N15/1433—Signal processing using image recognition
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T7/00—Image analysis
  • G06T7/0002—Inspection of images, e.g. flaw detection
  • G06T7/0012—Biomedical image inspection
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
  • A61B10/00—Instruments for taking body samples for diagnostic purposes; Other methods or instruments for diagnosis, e.g. for vaccination diagnosis, sex determination or ovulation-period determination; Throat striking implements
  • A61B10/02—Instruments for taking cell samples or for biopsy
  • A61B2010/0216—Sampling brushes
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
  • G01N15/01—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials specially adapted for biological cells, e.g. blood cells
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
  • G01N15/10—Investigating individual particles
  • G01N2015/1006—Investigating individual particles for cytology
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
  • G01N15/10—Investigating individual particles
  • G01N15/14—Optical investigation techniques, e.g. flow cytometry
  • G01N15/1468—Optical investigation techniques, e.g. flow cytometry with spatial resolution of the texture or inner structure of the particle
  • G01N2015/1472—Optical investigation techniques, e.g. flow cytometry with spatial resolution of the texture or inner structure of the particle with colour
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N15/00—Investigating characteristics of particles; Investigating permeability, pore-volume or surface-area of porous materials
  • G01N15/10—Investigating individual particles
  • G01N15/14—Optical investigation techniques, e.g. flow cytometry
  • G01N2015/1488—Methods for deciding
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/20—Special algorithmic details
  • G06T2207/20084—Artificial neural networks [ANN]
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T2207/00—Indexing scheme for image analysis or image enhancement
  • G06T2207/30—Subject of image; Context of image processing
  • G06T2207/30004—Biomedical image processing
  • G06T2207/30024—Cell structures in vitro; Tissue sections in vitro

Definitions

• the present invention is directed to a minimally invasive apparatus and method for testing lesions of the oral cavity and similar epithelium.
• the present invention is directed to determining whether or not there is a likelihood of detection of precancerous or cancerous conditions in epithelial tissue.
• Detecting precancerous or cancerous conditions in certain body tissue is important. There are various methodologies of obtaining tissue samples for examination and methodologies for analyzing cellular specimens of the tissue samples.
• the dentist or physician who visually detects an oral lesion which is not clearly suggestive of precancer or cancer is faced with a quandary when restricted to the methods and apparatus of the prior art.
• the only accurate tool currently believed to be available in the prior art to distinguish benign from precancerous and cancerous oral lesions is a lacerational or scalpel biopsy of the lesion followed by histological examination of the excised tissue.
• a scalpel biopsy a variety of surgical cutting instruments are used to obtain a tissue sample. I f such a scalpel biopsy removes a part of the lesion it is referred to as an "incisional " biopsy, while if it removes the entire lesion it is referred to as an "excisional " biopsy.
• a scalpel biopsy is a painful, lacerational, highly invasive procedure.
• Typical instruments for this purpose include, but are not limited to a flat scalpel blade, a round scalpel blade (punch biopsy) and scissors.
• Local anesthesia is always required .
• Considerable bleeding from the wound is common and suturing is often necessary.
• primary care dentists and physicians those clinicians who most often encounter benign appearing oral lesions, are reluctant to perform a scalpel biopsy. When necessary, these clinicians will therefore generally refer the patient to an oral surgeon or oral pathologist for the procedure.
• pre-cancerous and cancerous oral lesions often mimic benign lesions .
• these pre-cancerous or cancerous, but benign appearing, oral lesions typically do not receive any immediate diagnostic evaluation and are thus allowed to progress to an advanced stage of oral cancer. Once such progression is underway and continues untreated, the patient' s chances for recovery diminishes .
• cytology A prior art approach which has attempted to address this problem in testing lesions of the oral cavity was the use of cytology.
• cytology is commonl ⁇ used to detect precancer and cancer in other body sites, it has not proven to be useful in the oral cavity because of its low sensitivity, i .e. its high false negative rate. It is believed that this high false negative rate is in part due to the fac; that many oral lesions have an overlying keratin layer which limits availability to the lesion surface of naturally exfoliated abnormal cells . In one large study, ora' cytology was found to have a false negative rate of 30 % .
• a unique processing system to anaivze cellular specimens for abnormal keratinization is described .
• a cytological or cellular sample of an oral lesion is taken from a patient for analysis.
• this sample is obtained by means of a non-scalpel instrument which is sufficiently abrasive to penetrate all three layers (basal, intermediate, and superficial) of the oral epithelium
• this transepithelial sample is obtained by means of pressing and rotating a circular stiff nylon brush several times over the entire lesion surtace
• the sample can be obtained using cytology or histology, and can be any cellular specimen, including cells sloughed off naturally, or cells removed by a health care professional, including a tissue specimen or oral biopsy.
• a cellular sample (preferably trans-epithelial) is examined with the aid of an image recognition system designed to identify minimal evidence of pre-cancerous and cancerous change.
• the system can detect small numbers of abnormal cells distributed among the large number of normal cells obtained during the sampling procedure.
• the subject invention preferably overcomes the limitations and difficulties associated with analysis of cellular specimens for abnormal characteristics by providing an image recognition system which detects characteristics relating to abnormal keratinization of the cells.
• these characteristics include color saturation associated with such abnormal keratinization.
• the system described overcomes the sensitivity limitations of prior art oral cytologic technique by combining innovations in oral pathology, namely 1 ) a non-scalpel cellular sample of all three layers of the oral epithelium, and 2) subjecting this novel sample to inspection by the novel image recognition system specifically designed to detect minimal evidence of early precancerous change in a trans- epithelial sample from an oral lesion, or other lesion with similar epithelia.
• This novel image recognition system which is the suoject of this patent application preferably analyzes for the presence of abnormal keratinization, by detecting predetermined characteristics of color saturation.
• the prior art scalpel procedure is defined as lacerational, whereas the sample collection device herein is non- lacerational and therefore minimally invasive
• the sample collection device herein is non- lacerational and therefore minimally invasive
• the image processing system combines: 1 ) sensitivity to the presence of abnormal cellular morphology obtained from any or all of the three layers of the novel trans-epithelial cellular sample with 2) sensitivity to the presence of abnormally keratinized cells as are commonly found in any or all of the three layers of the epithelia and also obtained by means of the novel trans-epithelial cellular sample of the subject invention
• the keratin component which presented an obstacle to prior art oral cytology, is both penetrated , to ensure that any underlying abnormal basal cell morphology is available for analysis, and productively utilized, as a means of increasing the method's overall sensitivity to evidence of precancerous and cancerous change.
• the image recognition system selects the most suspect abnormal cells and cell clusters among the sample, and displays these cells and ceil clusters on a video monitor for expert review
• the image recognition system also provides a color printout of those suspect cells and cell clusters selected by the expert reviewer as representative of the case.
• the image recognition system selects abnormal cells based on morphological characteristics and on overall resemblance to abnormal cells on which it has been trained.
• the image recognition system is directed to detection of abnormality of the oral cavity by including a function to detect abnormal keratin as is often found in dysplastic and cancerous oral tissue
• the image recognition system is directed to detection of abnormal keratin by being programmed to detect a threshold level of color saturation associated with "hyalinization” or the stained appearance of such abnormal keratin.
• the image recognition is directed to the combination of morphological cellular change associated with pre-cancer and cancer and the appearance of abnormal keratin as produced by pre-cancerous and cancerous cells of the oral cavity and similar epithelia.
• the image recognition system may be constructed through modification of image recognition systems currently manufactured and sold to detect abnormal cells spontaneously exfoliated from non-keratinized lesions such as cervical lesions.
• the invention consists of a method and system which for the first time provides accurate evaluation of oral lesions without requiring the performance of a scalpel biopsy
• a primary objective of the subject invention is to accurately distinguish pre-cancerous and cancerous oral lesions from benign oral lesions without the pain, bleeding, and tissue wound that typically result from the prior art technique of scalpel biopsy and histology.
• the sensitivity and specificity of the subject invention was compared to the prior art technique of scalpel biopsy and histology.
• the subject invention was found to detect 1 00% of the precancerous and cancerous lesions detected by the prior art technique.
• the subject invention thus had a 0% false negative rate in this study.
• the subject invention had less than a 1 % false positive rate in this study.
• the subject invention also detected pre-cancer or cancer in approximately 1 5 patients whose lesions did not visually appear suspicious enough to expert examiners to warrant a scalpel biopsy. These outcomes represent lives that were potentially saved by the subject invention.
• An additional advantage of the subject invention over the prior art diagnostic technique is greater sensitivity to the detection of pre-cancer and cancer in large multi-focal oral lesions This is because of the larger sampling area obtained by the brush biopsy technique of the subject invention when compared to the smaller area sampled by a traditional incisional scalpel biopsy.
• An additional advantage of the subject invention is that it allows accurate, minimally invasive testing of lesions from areas of the body outside of the oral cavity where a keratin layer that limits the accuracy of prior art cytological WO 00/51066 PCTTJSOO/00743
• An additional advantage of the subject invention is that it allows diagnosis of non-cancerous conditions, such as candidiasis, herpes, geographic tongue, lichenplanus, human papilloma virus, and others.
• Figure 1 is a flowchart illustrating the method of one embodiment of the present invention.
• FIG. 2 is a flowchart illustrating the method of an alternate embodiment, in accordance with the present invention.
• Figure 3 is a flowchart illustrating the method of a third embodiment, in accordance with the present invention.
• Figure 4 is a cross sectional view of an section of epithelium including precancerous or cancerous cells, showing the extent of sampling obtained using exfoliative cytology.
• Figure 5 is a cross sectional view of an section of epithelium including precancerous or cancerous cells, showing the extent of sampling obtained using the brush biopsy technique of the present invention.
• such cells display a central nucleus, a thread-like chromatin pattern, karyopyknosis with maturation, intercellular bridges, stratification, keratinization, thinning, orientation parallel to the basement membrane and lumen, and exfoliation as single cells.
• an automated, computer implemented, system for detection of such differentiation, particularly for cells of the oral cavity and similar epithelium.
• the invention provides a means for alerting physicians to the presence of cancerous or precancerous cells at an early stage by overcoming the disadvantages of the prior art which has tended to avoid early detection.
• a sample of cells from the oral cavity or similar epithelia is processed by an image analysis system for detection of characteristics associated with dysplasia or cancer.
• the sample is a standard sample of cells, taken as known in the prior art.
• the sample is a sample taken using a nonlacerational sampling device.
• the sample can be sent to an automated image processing system for detection of abnormal keratinization.
• a transepithelial sample be taken from the oral cavity or similar epithelia, the sample being obtained using a nonlacerational sampling device.
• This cytological or cellular sample of the entire epithelial thickness of an oral lesion is preferably obtained using a non-scalpel instrument which is sufficiently abrasive to penetrate all three layers (basal, intermediate, and superficial) of the oral epithelium.
• the device is the sampling instrument disclose ⁇ in U .S. Provisional Patent Application Serial No.
• this trans-epithelial sample is obtained by means of pressing and rotating a circular stiff nylon brush several times over the entire lesion surface. This sample is then analyzed by an automated image processing system for detection of morphology or characteristics typical of dysplasia or cancer.
• the transepithelial sample is analyzed by an image processing system for detection of the presence of abnormal keratinization.
• the sample is also analyzed for the presence of other characteristics or morphology associated with dysplasia or cancer. If desired, this analysis can be done with any cellular sample, whether a sample obtained by histology or cytology, and can include cells naturally sloughed off, cells intentionally rubbed off, or a tissue sample from an oral biopsy. Although such techniques can be used in conjunction with the invention, however, they ⁇ o not constitute the preferred embodiment, as it is preferred that a brush biopsy be performed.
• a non-abrasive sweep is conducted of the epithelial surface in a region of interest which typically only captures surface and exfoliated cells from the epithelial area.
• Abnormal cells 37 located below the surface for example, cells located at the basal layer, will theretore often be missed using this superficial sampling.
• the specimen sent for analysis will lack these deeper cells, and may therefore result in a false negative diagnosis.
• a brush biopsy is taken, using the brush biopsy instrument disclosed, the brush being used to penetrate below the surface of the epithelium and obtain cells from all three epithelial layers.
• the specimen sent for testing includes abnormal cells regardless of whether they are cells from the superficial layer or the deeper intermediate and basal layers, providing a more complete and accurate cellular sample for analysis.
• the subject invention thus combines two innovations in oral pathology, namely, a cellular sample of all three layers of the oral epithelium obtained without the use of a scalpel or similar lacerational instrument, with an analysis of that sample by a novel image recognition system specifically designed to detect minimal evidence of early precancerous change in that trans-epithelial sample from an oral lesion, or other lesion with similar epithelia.
• the image processing system combines sensitivity to the presence of abnormal cellular morphology obtained from any of the three layers of the epithelia with sensitivity to the presence of abnormally keratinized cells as are found in any of the layers of the epithelia.
• the keratin component which previously presented an obstacle to prior art oral cytology, is penetrated to ensure that any underlying abnormal intermediate and basal cell morphology is available for analysis and productively utilized as a means of increasing the method's overall sensitivity to evidence of precancerous and cancerous change.
• the invention utilizes a modified version of the commercially available PAPNET system, currently sold by Neuromedical Systems, Inc. of Upper Saddle River, New Jersey and Suffern, New York Further details of such systems are described in U.S. Patent No. 4,965,725, entitled “Neural Network Based Automated Cytological Specimen Classification System and Method ", U.S. Patent No. 5,257, 1 82, entitled “Morphological Classification System and Method"; U.S. Patent No 5,287,272, entitled WO 00/51066 PCT/TJSOO/00743
• the invention utilizes a modified computer image recognition system having similar or equivalent capability to detect and present abnormal cells within certain predetermined parameters.
• such testing is conducted to obtain a transepithelial sample of the lesion which can them be sent for staining and subsequent analysis by a computer implemented system. Accordingly, in the first step of the present invention, a transepithelial sample is taken from the patient's oral lesion.
• the cytological or cellular sample is taken of the entire epithelial thickness of an oral lesion by means of a nonlacerational or nonscalpel instrument which is sufficiently abrasive to penetrate all three layers (basal, intermediate, and superficial) of the oral epithelium
• this trans- epithelial sample is obtained by means of pressing and rotating a circular stiff nylon brush several times over the entire lesion surface.
• the sample is taken using the brush disclosed in pending U.S. Provisional Patent Application Serial No.
• the sample Upon sampling of the lesion, the sample is stained preferably using the modified Papanicolaou stain, as is well known in the art. This stained sample is then imaged and analyzed using an image recognition system which selects abnormal cells based on a combination of abnormal morphology and abnormal keratinization particularly the characteristics associated with lesions of the oral cavity and similar epithelia.
• the image recognition system first processes the image through an algorithmic classifier and then sends the processed data to a neural net work
• the algorithmic classifier locates a first group of candidate objects within the image which could be the nuclei of cells.
• the algorithmic classifier has been modified to also locate a second group of candidate objects within the image, these being cells displaying the abnormal keratinization typical of precancerous and cancerous lesions of the oral cavity
• the neural net then scores cells for the presence of other morphological features associated with cancerous cells.
• these two candidate groups are both sent to the neural net for scoring, and the highest ranking objects from the combined two groups are displayed to the expert reviewer.
• those ceils displaying abnormal keratinization do not compete with other cells in the neural net stage of analysis; rather, all abnormally keratinized cells are forwarded for display to the expert reviewer.
• the stained sample is analyzed by a modified PAPNET system, the system being modified as described herein.
• PAPNET systems are commercially available image recognition systems which select the most suspect abnormal cells and cell clusters among those presented in a stained sample, and then display these cells and cell clusters on a video monitor for expert review.
• the programming of the system is set to detect and display the top 64 glandular cells or top 64 clusters of cells, and the top 64 squamous ceils or top 64 single cells, the ratio of glandular to squamous cells or cell clusters to single cells being in a 1 : 1 ratio, these top cells being those cells ranked highest for abnormalities by the system.
• the programming of the system is modified to vary this detection ratio to enhance the system's performance on oral brush biopsies of the oral mucosa.
• the system is reprogrammed to detect and display the top 64 clusters of cells and the top 1 28 single cells, in a 1 :2 ratio.
• a 1 :2 ratio is used in the preferred embodiment, in other embodiments modifications can be made to the controller such that a 1 :3 ratio, or some other ratio higher than 1 . 1 can be employed. Accordingly, by modification of this detection parameter the detection capabilities for abnormalities in oral lesions is improved.
• the system is reprogrammed such that a further set of parameters are introduced to enhance the system's detection capabilities, by isolating a second group of candidate cells.
• the image recognition system is directed to enhance its detection of abnormalities of the oral cavity by including a function to detect abnormalities such as are characteristic of dysplastic and cancerous oral tissue.
• the system detects abnormal keratin. These keratinized cells are then either forwarded to the neural net for detection of abnormalities consistent with cancer or precancer, or are forwarded directly to an expert for analysis.
• Another method to detect abnormal samples is to set threshold levels for morphology and/or color, which threshold levels indicate, if exceeded, that there is a likelihood of abnormal cells in the sample being analyzed.
• the PAPNET system is redesigned or reprogrammed to detect "glowing" or glassy, orange - red cytoplasms, as are associated with abnormal keratinization of oral and similar epithelia.
• the algorithmic classifier is programmed to take the cells of the stained sample and to further perform an analysis of their color.
• each pixel of the image is separately screened or tested for distinctive color characteristics.
• the cells are first screened and detected for abnormality by the system using the standard parameters of the PAPNET system (with a modified detection ratio as described above) and are then analyzed for distinctive color.
• every pixel of the image is preferably measured to detect the pixel's Hue (H), Intensity (I) and Saturation (S) .
• the target Hue sought is an orange - red, such as that commonly associated with abnormally keratinized cytoplasm.
• the system searches for a Hue of approximately 46-82 on a scale ranging from 0-255.
• the target pixels which the system searches for are those ceils which further have a high Intensity and a high Saturation.
• the system is set to search for and detect pixels with an Intensity of 1 00-255 on a range of 0-255 Likewise, in the preferred embodiment, the system is set to detect a Saturation of 43-255, also on a range of 0-255.
• the system is programmed to analyze each pixel of the sample image and to narrow down the target pixels of interest to those pixels displaying all of the preferred parameters of desired Hue, Intensity and Saturation. RGB analysis could also be employed .
• a morphological closing i.e. a dilation and an erosion
• a morphological closing is mathematically performed by the system to find round objects.
• Such closings are known in the art, and are used, for example, elsewhere in the prior unmodified PAPNET system for purposes unrelated to the detection of abnormal keratin.
• Pratt describes such closings and the mathematics of the same. See, William K. Pratt, Digital Image Processing, Second Edition, Chapter 1 5, Morphological Image Processing (New York: John Wiley & Sons) .
• the morphological closing operation on these pixels with the desired color characteristics is performed using a 9x9 (octogonal) structuring element.
• the 1 4x 1 4 structuring element used by the PAPNET system in a separate step unrelated to color processing can still be used, however, in that separate step of the image processing
• the 9x9 structuring element used in this stage of the process tends to find round objects, and will find those which are approximately 2 - 1 8 microns in diameter.
• the first group of candidate objects (those objects which have been isolated by the algorithmic classifier without reference to color characteristics) and the second group of candidate objects (those cells isolated based on color characteristics that indicate the presence of abnormal keratinization) are both sent to the neural net for scoring, and the highest ranking objects from the combined two groups are displayed to the expert reviewer.
• those cells displaying abnormal keratinization do not compete with other cells in the neural net stage of analysis. Rather, the first group of candidate cells is scored by the neural net, and the second group of candidate cells bypasses the neural net and is directly forwarded for display to the expert reviewer.
• those cells displaying abnormal keratinization are sent to a separate, second neural net for independent ranking by a neural net, but also without having to compete with other cells in the neural net stage of analysis.
• the first group of candidate cells is scored by a first neural net
• the second group of candidate cells is scored by a second neural net, with each group of cells being forwarded for display to the expert reviewer
• a pathologist, cytologist or other expert reviewer can then visually examine the objects forwarded for the display, the objects being displayed on a video monitor Upon visual inspection, the reviewer can then make a WO 00/51066 PCT/TJSOO/00743
• the image recognition system also provides a color printout of those suspect cells and cell clusters selected by the expert reviewer as representative of the case.
• An alternative embodiment to scoring and displaying the cells to an expert reviewer is to program the system to set certain threshold levels both for morphological abnormalities and/or abnormal keratinization . When those threshold levels are exceeded, the system will automatically produce a visualization of only such specific cells and not those highest scored or highest ranked on the specimen regardless of whether or not the scores have exceeded the threshold level.
• the system can be programmed to detect and display basal cells using a basal cell detector in the algorithmic classifier.
• the pathologist can merely examine the cells in the sample with a standard microscope to detect the presence of basal cells.
• This embodiment provides a "fail-safe feature" within the overall testing process to ensure that a transepithelial sample has been obtained, i.e. that the cellular sample includes cells from all three layers of the epithelia. Should the pathologist determine that basal cells are not, in fact, present, the incomplete sample can be disregarded if negative and sampling can be repeated to attempt to obtain a transepithelial sample for analysis.
• the pathologist will not rely on the incomplete sample for the test results
• the initial sampling fails to obtain a transepithelial sample, that deficiency is detected in advance and the patient will not obtain a false sense of security from a false negative.
• the PAPNET system as is commercially available was designed for a platform with a neural network which has relatively slow processing speeds Because of the large number of cells in the original sample requiring examination, and the speed of the original PAPNET neural network, a multiple stage analysis was performed in which the population of cells to be analyzed by the neural network was necessarily reduced at each stage to allow the overall process to run in a reasonable time For example, if 500,000 cells were in a sample, a first analytical stage applying algorithmic, first stage neural network or other analytical techniques could reduce that population to approximately 50,000 cells, which could then be realistically examined by the processing speed of the then available neural networks.
• a single stage neural network is meant to indicate that there may be different layers within the neural network, but the entirely of the process can occur in a single stage without prior elimination of objects requiring neural network analysis.
• a single neural network stage can be implemented in which the entirety of the population of sample cells can be analyzed in a sufficiently short period of time to realize practical operation.

Landscapes

• Engineering & Computer Science (AREA)
• Health & Medical Sciences (AREA)
• General Health & Medical Sciences (AREA)
• Life Sciences & Earth Sciences (AREA)
• General Physics & Mathematics (AREA)
• Medical Informatics (AREA)
• Chemical & Material Sciences (AREA)
• Pathology (AREA)
• Physics & Mathematics (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Signal Processing (AREA)
• Quality & Reliability (AREA)
• Radiology & Medical Imaging (AREA)
• Biomedical Technology (AREA)
• Heart & Thoracic Surgery (AREA)
• Molecular Biology (AREA)
• Surgery (AREA)
• Animal Behavior & Ethology (AREA)
• Public Health (AREA)
• Veterinary Medicine (AREA)
• Theoretical Computer Science (AREA)
• Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
• Dispersion Chemistry (AREA)
• Analytical Chemistry (AREA)
• Biochemistry (AREA)
• Immunology (AREA)
• Investigating Or Analysing Biological Materials (AREA)
• Sampling And Sample Adjustment (AREA)
• Medicines Containing Antibodies Or Antigens For Use As Internal Diagnostic Agents (AREA)
• Testing Or Measuring Of Semiconductors Or The Like (AREA)
• Sewage (AREA)

Priority Applications (10)

Applications Claiming Priority (6)

Publications (1)

Family

ID=27382597

Family Applications (1)

Country Status (12)

Cited By (1)

Families Citing this family (7)

Citations (2)

Family Cites Families (10)

• 2000
  • 2000-01-12 DE DE60044604T patent/DE60044604D1/de not_active Expired - Lifetime
  • 2000-01-12 DK DK00902395.3T patent/DK1155381T3/da active
  • 2000-01-12 IL IL14493300A patent/IL144933A0/xx active IP Right Grant
  • 2000-01-12 WO PCT/US2000/000743 patent/WO2000051066A1/en not_active Ceased
  • 2000-01-12 EP EP00902395A patent/EP1155381B1/en not_active Expired - Lifetime
  • 2000-01-12 PT PT00902395T patent/PT1155381E/pt unknown
  • 2000-01-12 JP JP2000601597A patent/JP4748628B2/ja not_active Expired - Fee Related
  • 2000-01-12 AU AU24120/00A patent/AU769471B2/en not_active Ceased
  • 2000-01-12 CN CNB008040230A patent/CN100339862C/zh not_active Expired - Fee Related
  • 2000-01-12 CA CA002363833A patent/CA2363833C/en not_active Expired - Fee Related
  • 2000-01-12 AT AT00902395T patent/ATE472775T1/de active
• 2001
  • 2001-08-03 US US09/922,013 patent/US20020012938A1/en not_active Abandoned
  • 2001-08-16 IL IL144933A patent/IL144933A/en not_active IP Right Cessation
• 2004
  • 2004-04-27 AU AU2004201764A patent/AU2004201764B2/en not_active Ceased

Patent Citations (3)

Non-Patent Citations (2)

Also Published As

Similar Documents

Legal Events