CN113436150A - Construction method of ultrasound imaging omics model for lymph node metastasis risk prediction - Google Patents
Construction method of ultrasound imaging omics model for lymph node metastasis risk prediction Download PDFInfo
- Publication number
- CN113436150A CN113436150A CN202110632990.0A CN202110632990A CN113436150A CN 113436150 A CN113436150 A CN 113436150A CN 202110632990 A CN202110632990 A CN 202110632990A CN 113436150 A CN113436150 A CN 113436150A
- Authority
- CN
- China
- Prior art keywords
- image
- lymph node
- omics
- swe
- node metastasis
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 208000007433 Lymphatic Metastasis Diseases 0.000 title claims abstract description 47
- 238000012285 ultrasound imaging Methods 0.000 title claims abstract description 24
- 238000010276 construction Methods 0.000 title claims abstract description 10
- 238000000034 method Methods 0.000 claims abstract description 33
- 238000003384 imaging method Methods 0.000 claims abstract description 27
- 238000004458 analytical method Methods 0.000 claims abstract description 25
- 238000003745 diagnosis Methods 0.000 claims abstract description 21
- 238000002604 ultrasonography Methods 0.000 claims abstract description 20
- 238000013058 risk prediction model Methods 0.000 claims abstract description 13
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000012795 verification Methods 0.000 claims abstract description 10
- 238000011282 treatment Methods 0.000 claims abstract description 9
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 238000000586 desensitisation Methods 0.000 claims abstract description 4
- 206010028980 Neoplasm Diseases 0.000 claims description 29
- 210000001165 lymph node Anatomy 0.000 claims description 19
- 206010027476 Metastases Diseases 0.000 claims description 9
- 230000009401 metastasis Effects 0.000 claims description 9
- 238000000611 regression analysis Methods 0.000 claims description 9
- 208000001204 Hashimoto Disease Diseases 0.000 claims description 6
- 208000030836 Hashimoto thyroiditis Diseases 0.000 claims description 6
- 238000000585 Mann–Whitney U test Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000001575 pathological effect Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 238000012546 transfer Methods 0.000 claims description 6
- 238000010219 correlation analysis Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 238000012352 Spearman correlation analysis Methods 0.000 claims description 3
- 208000009453 Thyroid Nodule Diseases 0.000 claims description 3
- 230000003902 lesion Effects 0.000 claims description 3
- 238000007477 logistic regression Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 206010033701 Papillary thyroid cancer Diseases 0.000 description 11
- 208000030045 thyroid gland papillary carcinoma Diseases 0.000 description 11
- 208000024770 Thyroid neoplasm Diseases 0.000 description 3
- 239000004568 cement Substances 0.000 description 3
- 238000003759 clinical diagnosis Methods 0.000 description 3
- 230000007170 pathology Effects 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 210000001685 thyroid gland Anatomy 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000036210 malignancy Effects 0.000 description 2
- 210000000056 organ Anatomy 0.000 description 2
- 238000002271 resection Methods 0.000 description 2
- 238000001356 surgical procedure Methods 0.000 description 2
- 201000002510 thyroid cancer Diseases 0.000 description 2
- 201000011510 cancer Diseases 0.000 description 1
- 230000000747 cardiac effect Effects 0.000 description 1
- 238000002512 chemotherapy Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000007717 exclusion Effects 0.000 description 1
- 230000009545 invasion Effects 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007674 radiofrequency ablation Methods 0.000 description 1
- 238000001959 radiotherapy Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 230000000472 traumatic effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10132—Ultrasound image
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Medical Informatics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Pathology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Epidemiology (AREA)
- Evolutionary Biology (AREA)
- Primary Health Care (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
Abstract
The invention discloses a method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction, which comprises the following steps: 1) and (3) data set construction: acquiring original ultrasonic medical images and clinical information of ultrasonic examination of a PTC patient, carrying out standardization and desensitization treatment on the original ultrasonic medical images and the clinical information, and randomly dividing the treated data into a training set and a verification set; 2) delineating the region of interest: delineating a region of interest in each ultrasound medical image; 3) and (3) image omics feature extraction: performing high-dimensional iconomics feature extraction on each region of interest; 4) and (3) image omics analysis: respectively constructing two imaging omics labels for the training set, and then combining the imaging omics labels with clinical information to construct a lymph node metastasis risk prediction model; 5) and (5) verifying a prediction model. The construction method of the ultrasound imaging omics model for lymph node metastasis risk prediction can accurately quantify, is noninvasive and noninvasive, and does not increase the diagnosis and treatment procedures of patients.
Description
Technical Field
The invention relates to the technical field of medical image processing, in particular to a method for constructing an ultrasonic imaging omics model for lymph node metastasis risk prediction.
Background
Papillary Thyroid Carcinoma (PTC) is the most common primary thyroid malignancy and has rapidly increased incidence worldwide in recent years. Despite the slow clinical course of PTC, cervical Lymph Node (LN) metastasis due to its aggressive lymphoid nature remains a significant problem, and cervical lymph node metastasis is reported in 20-90% of patients under diagnosis. The recent trend toward surgical treatment of papillary thyroid carcinoma has shifted to more individualized, less traumatic treatments, particularly for small unifocal (<1cm) tumors without peri-thyroid invasion and lymph node metastasis. The latest revised american thyroid association guidelines recommend that thyroid single lobe resection is safe and effective for patients without cervical lymph node metastasis. However, in clinical practice, it is difficult for a clinician to ascertain whether a patient has lymph node metastasis. Because cervical lymph node metastasis is an important risk factor for recurrence and death of PTC patients, the cervical lymph node metastasis of papillary thyroid cancer patients can be accurately predicted before an operation, and a basis can be provided for selection of an operation mode and a tumor resection range.
Type B ultrasound is a pre-PTC assessment of cervical LN status with high specificity (85.0% -97.4%) but relatively low sensitivity (36.7% -61.0%). With the development of Ultrasound Shear Wave elasticity imaging (SWE), studies have shown that the quantitative elasticity index of preoperative SWE can help to predict Lymph Node metastasis (Jung, WS., et al, Shear Wave elasticity in evaluation of cardiac simple Node measurements of Clinical pathology, 2015.22(1):111-6, Park, AY., et al, Shear-Wave elasticity for Clinical diagnosis of Clinical pathology, environmental impact diagnosis, UMNALS of Clinical diagnosis, 722, 2016.23(5): chemical elasticity, cement 729, cement viscosity, diagnosis, And diagnosis of Clinical diagnosis, cement viscosity, And diagnosis, 2. environmental viscosity, 2. diagnosis, biological diagnosis, And Use of diagnosis.
The image omics is a process of extracting quantitative image information from medical images in a large scale, can be intuitively understood as converting visual image information into deep features for quantitative research, and assists doctors to make the most accurate diagnosis by performing deeper mining, prediction and analysis on massive image data information through tumor segmentation, feature extraction and model establishment. Therefore, it is necessary to provide a method for constructing an ultrasound imaging omics model for predicting the risk of metastasis of cervical lymph nodes of papillary thyroid cancer patients to solve the above problems.
Disclosure of Invention
The invention aims to overcome the defects of the background technology and provide a method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction, which can accurately quantify, is noninvasive and does not increase more diagnosis and treatment procedures of patients.
In order to achieve the above object, the present invention provides a method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction, comprising the following steps:
1) and (3) data set construction: acquiring original ultrasonic medical images and clinical information of ultrasonic examination of a PTC patient, carrying out standardization and desensitization treatment on the original ultrasonic medical images and the clinical information, and randomly dividing the treated data into a training set and a verification set;
2) delineating the region of interest: delineating a region of interest in each ultrasound medical image;
3) and (3) image omics feature extraction: performing high-dimensional image omics feature extraction on each region of interest in the ultrasonic medical image;
4) and (3) image omics analysis: respectively constructing two imaging omics labels for the training set, and then combining the imaging omics labels with clinical information to construct a lymph node metastasis risk prediction model;
5) and (3) verification of a prediction model: and applying the verification set to a lymph node metastasis risk prediction model to verify the diagnosis efficiency of the lymph node metastasis risk prediction model.
Further, in the step 1), the original ultrasound medical image comprises a B-mode ultrasound image and a SWE image, and is stored in an original DICOM medical image format; the clinical information comprises the age, sex, pathological result, primary tumor position, tumor multifocal, lymph node metastasis conclusion in an ultrasonic diagnosis report, tumor standard grading, tumor image report composition and hashimoto thyroiditis conclusion of the patient.
Further, in the step 2), the ITK-snap software is adopted to delineate thyroid nodules in the B-mode ultrasound image, and then the region of interest is mapped into the corresponding SWE image in a registration manner, wherein the mapping formula is as follows:
ROISWE(x,y)＝ROIBMUS(x+w,y+h)
in the formula, ROIBMUS(x, y) denotes the region of interest, ROI, in B-mode ultrasonographySWE(x, y) denotes a region of interest in the SWE, w and h denote offsets between the SWE image and the B-mode image, respectively, w is half the width of the image, and h is 0.
Further, in the step 3), a PyRadiomics package of python software is adopted to extract a first-order histogram feature, a shape feature, a texture feature and a wavelet feature of each region of interest.
Further, in the step 4), the imaging omics analysis specifically includes the following steps:
4.1) feature screening: respectively analyzing B ultrasonic image features and SWE image features of the training set by using a significance analysis method, a correlation analysis method and a Lasso regression analysis method, and removing features irrelevant to an LN state;
4.2) constructing a B-ultrasonic imaging omics label: performing Lasso analysis result coefficients according to B-ultrasonic image characteristics, and forming a B-ultrasonic image omics label in a linear weighting mode;
4.3) constructing a SWE imaging omics tag: carrying out Lasso analysis result coefficients according to the SWE image characteristics, and forming a SWE image omics tag in a linear weighting mode;
4.4) constructing a risk prediction model: and (3) constructing a prediction model of LN transfer risk by performing binary multi-factor Logistic regression analysis on the clinical information, the B-ultrasonic imaging group tag and the SWE imaging group tag.
Further, in the step 4.1), the significance analysis is to perform T test or Mann-Whitney U test on the image features, perform T test on the features conforming to normal distribution, perform Mann-Whitney U test on the features not conforming to normal distribution, and reject the features with a P value greater than 0.05; performing Spearman correlation analysis on the features which are not removed in the previous step of analysis, and removing the features with the correlation coefficient smaller than 0.8; the Lasso regression analysis is to perform Lasso regression analysis on the features which are not eliminated in the previous step of analysis, and eliminate the features with coefficient of 0.
Further, in the step 4.2), the formula for constructing the B-mode ultrasound imaging omics label in a linear weighting manner is as follows:
RadScore_BMUS＝0.058
+0.106×original_shape2D_Perimeter
+0.039×wavelet.LH_glszm_ZoneEntropy
wherein RadsCore _ BMUS represents a B-mode ultrasonography tag, origi _ shape2D _ Perimeter represents lesion side length, and wavelet.
Further, in the step 4.3), the formula for constructing the SWE imaging omics tag in a linear weighting manner is as follows:
RadScore_SWE＝0.041
+0.133×logarithm_glszm_ZoneEntropy_R
-0.045×gradient_glcm_Imc1_B
-0.060×wavelet.LH_glszm_SmallAreaLowGrayLevelEmphasis_GREY
-0.032×wavelet.LH_ngtdm_Contrast_G
wherein, RadsCore _ SWE represents a SWE image omics label, logarihm _ glszm _ ZoneEntrol _ R represents the regional entropy of a logarithmic filtering gray scale region size matrix under a red channel, gradient _ glcm _ Imc1_ B represents the related information measurement of a gradient filtering gray scale symbiotic matrix under a blue channel, wavelet.
Still further, in the step 4.4), constructing a prediction model of LN transfer risk is:
wherein NomoScore is a prediction model of LN metastasis Risk, Multifocality is tumor multifocal, US-ported LN status is lymph node metastasis conclusion in an ultrasonic diagnosis report, Radsore _ SWE is a SWE imaging omics label, e represents a natural constant, and Risk represents LN metastasis Risk probability.
Further, if Risk <0.574, it indicates no metastasis has occurred in the lymph node; if Risk ≧ 0.574, it indicates no metastasis has occurred in the lymph node.
Compared with the prior art, the invention has the following advantages:
firstly, the ultrasound imaging omics model for predicting the lymph node metastasis risk constructed by the method can accurately predict the lymph node metastasis risk of thyroid cancer patients, and can provide more comprehensive information of focuses by combining multi-modal information fusion of clinic and images.
Secondly, the construction method of the ultrasound imaging omics model for lymph node metastasis risk prediction can be accurately quantified, noninvasive and noninvasive, and does not increase more diagnosis and treatment procedures for patients, and the use method is simple and easy to popularize and apply in actual clinic.
Drawings
FIG. 1 is a process flow of a method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to the present invention;
FIG. 2 is a diagram of a model for predicting lymph node metastasis risk established by the present invention;
FIG. 3 is a graph of a fit of the present invention prediction to the patient's actual lymph node metastasis status;
FIG. 4 is a decision curve of the clinical utility value of the predictive model of the invention.
Detailed Description
The following describes the embodiments of the present invention in detail with reference to the embodiments, but they are not intended to limit the present invention and are only examples. While the advantages of the invention will be apparent and readily appreciated by the description.
The invention discloses a method for constructing an ultrasound imaging omics model for predicting lymph node metastasis risk, which comprises the following steps of:
step 1), data set construction: acquiring original medical images and clinical information of ultrasonic examination of a PTC patient, wherein the original ultrasonic medical images comprise B-mode ultrasonic images and SWE images and are stored in an original DICOM medical image format, carrying out standardization and desensitization treatment on the original medical DICOM images and the clinical information, randomly dividing the processed data into a training set and a verification set according to the ratio of 8:2, wherein the data in the training set is used for constructing a model, and the data in the verification set is used for verifying the prediction efficiency of the model;
inclusion criteria were: the patient is diagnosed with papillary thyroid carcinoma through the pathology in the operation; b ultrasonic and SWE simultaneous imaging is carried out within two weeks before operation; ultrasonic examination of the longest axis of the target tumor; performing cervical lymphadenectomy and pathological examination;
exclusion criteria: the pathological result of the operation specimen is uncertain; the patient has undergone radio frequency ablation surgery, radiation therapy or chemotherapy; the target tumor has artifacts on the image; with other malignancies.
Clinical information: the clinical information of the patients comprises Age (Age), Gender (Gender), pathological results, Primary position of Tumor (Primary site), Tumor position (Tumor location), Tumor multifocal (Multifocality), lymph node metastasis conclusion in ultrasonic diagnosis report (US-ported LN status), Tumor standard grade (TI-RADS level), Tumor image report Composition (Composition) and Hashimoto thyroiditis conclusion (Hashimoto thyroiditis);
step 2), delineating an interested area: delineating a thyroid nodule region of interest (ROI) in the B-mode ultrasound image and the SWE image, because the organ in the SWE image has the same structure as the organ in the B-mode ultrasound image, the ROI delineation only needs to be carried out on the B-mode ultrasound image, and the ROI is mapped into the SWE image in a registration mapping mode:
ROISWE(x,y)＝ROIBMUS(x+w,y+h)
wherein, ROIBMUS(x, y) denotes the region of interest, ROI, in B-mode ultrasonographySWE(x, y) denotes the region of interest in the SWE, w and h denote the offset between the SWE image and the B-mode image, respectively, with w typically being half the width of the image and h being 0.
Step 3), performing image omics feature extraction: extracting 479 features from each ROI of the B-mode and SWE images respectively through a PyRadiomics package of python software, wherein the 479 features comprise 18 first-order histogram features, 14 shape (shape) features, 24 gray level co-occurrence matrix (GLCM) features, 16 gray level run step matrix (GLRLM) features, 16 gray level region size matrix (GLSZM) features, 14 Gray Level Difference Matrix (GLDM) features, 5 field gray level difference matrix (NGTDM) features and 372 first-order histogram, GLCM, RLM, GLTDM features of wavelet filtered image (wavelet);
step 4), performing image omics analysis: two image labels RadScore _ BMUS and RadScore _ SWE are constructed respectively based on B-ultrasound and SWE images, the image omics labels are combined with clinical characteristics to construct a multi-modal lymph node metastasis risk prediction model, and the specific analysis process comprises the following steps:
step 4.1), feature screening: respectively and sequentially carrying out significance analysis, correlation analysis and Lasso regression analysis on B-ultrasound and SWE image omics characteristics in a training set, wherein the significance analysis represents that T test or Mann-Whitney U test is carried out on image characteristics, T test is carried out on characteristics conforming to normal distribution, Mann-Whitney U test is carried out on characteristics not conforming to normal distribution, and characteristics with a P value larger than 0.05 are removed; the correlation analysis means that the features which are not removed in the previous step are analyzed and subjected to Spearman correlation analysis, and the features with the correlation coefficient smaller than 0.8 are removed; the Lasso regression analysis shows that Lasso regression analysis is carried out on the features which are not eliminated in the previous step of analysis, and the elimination coefficient is 0;
step 4.2), constructing a B-ultrasonic imaging omics label: performing Lasso analysis result coefficients according to B-ultrasonic image characteristics, and forming a B-ultrasonic image omics label in a linear weighting mode; the imaging group signature of B-ultrasound (Radscore _ BMUS) is:
wherein, original _ shape2D _ Perimeter represents the side length of the lesion, and wavelet.
4.3) constructing a SWE imaging omics tag: carrying out Lasso analysis result coefficients according to the SWE image characteristics, and forming a SWE image omics tag in a linear weighting mode; the imaging omics signature for SWE (RadScore _ SWE) is:
RadScore_SWE＝0.041
+0.133×logarithm_glszm_ZoneEntropy_R
-0.045×gradient_glcm_Imc1_B
-0.060×wavelet.LH_glszm_SmallAreaLowGrayLevelEmphasis_GREY
-0.032×wavelet.LH_ngtdm_Contrast_G
the log _ glszm _ ZoneEntropy _ R represents the regional entropy of the log-filtered gray scale region size matrix in the red channel, the gradient _ glcm _ Imc1_ B represents the gradient-filtered gray scale co-occurrence matrix correlation information measure in the blue channel, the wavelet.
Step 4.3), constructing a risk prediction model: as shown in fig. 2, the final multi-modal imaging group risk prediction model (nomocore) was constructed by analyzing clinical indexes Age (Age), Gender (Gender), pathological outcome, Primary Tumor position (Primary site), Tumor position (Tumor location), Tumor multifocal (Multifocality), lymph node metastasis conclusion in ultrasonic diagnosis report (US-ported LN status), Tumor standard grade (TI-RADS level), Tumor image report Composition (Composition), Hashimoto thyroiditis (Hashimoto thyroiditis), B-supermicro tag (radius _ BMUS), and SWE tag (radius _ SWE) through Logistic regression method:
wherein e represents a natural constant, Risk represents LN transition Risk probability; the ROC curve of the prediction result of the training set is analyzed, the AUC is 0.851, the maximum johnson index (sensitivity + specificity-1) is 0.574 is used as an optimal threshold, Risk <0.574 indicates that the LN does not transfer, and Risk ≧ 0.574 indicates that the LN does not transfer.
Step 5), verifying a prediction model, applying the image group risk prediction model to data in a verification set, wherein a fitting curve is shown in fig. 3, a decision curve of clinical application value is shown in fig. 4, the area AUC under the ROC curve is 0.832 (95% CI,0.749-0.916), the prediction probability is counted by taking 0.574 as a threshold point, the sensitivity is 0.868 (95% CI,0.727-0.943), the specificity is 0.731 (95% CI,0.598-0.832), and the accuracy is 0.789 (95% CI, 0.694-0.861).
The ultrasound imaging omics model for predicting the lymph node metastasis risk, which is constructed by the method, can accurately predict the lymph node metastasis risk of a thyroid cancer patient, and can provide more comprehensive information of a focus by combining multi-modal information fusion of clinic and image; the method belongs to a detection method which can be accurately quantified, is noninvasive and noninvasive, does not increase more diagnosis and treatment procedures of a patient, is simple to use, and is easy to popularize and apply in actual clinic.
The above description is only an embodiment of the present invention, and it should be noted that any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the protection scope of the present invention, and the rest that is not described in detail is the prior art.
Claims (10)
1. A construction method of an ultrasound imaging omics model for lymph node metastasis risk prediction is characterized by comprising the following steps: the method comprises the following steps:
1) and (3) data set construction: acquiring original ultrasonic medical images and clinical information of ultrasonic examination of a PTC patient, carrying out standardization and desensitization treatment on the original ultrasonic medical images and the clinical information, and randomly dividing the treated data into a training set and a verification set;
2) delineating the region of interest: delineating a region of interest in each ultrasound medical image;
3) and (3) image omics feature extraction: performing high-dimensional image omics feature extraction on each region of interest in the ultrasonic medical image;
4) and (3) image omics analysis: respectively constructing two imaging omics labels for the training set, and then combining the imaging omics labels with clinical information to construct a lymph node metastasis risk prediction model;
5) and (3) verification of a prediction model: and applying the verification set to a lymph node metastasis risk prediction model to verify the diagnosis efficiency of the lymph node metastasis risk prediction model.
2. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 1, characterized in that: in the step 1), the original ultrasound medical image comprises a B-mode ultrasound image and a SWE image and is stored in an original DICOM medical image format; the clinical information comprises the age, sex, pathological result, primary tumor position, tumor multifocal, lymph node metastasis conclusion in an ultrasonic diagnosis report, tumor standard grading, tumor image report composition and hashimoto thyroiditis conclusion of the patient.
3. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 2, characterized in that: in the step 2), the ITK-SNAPE software is adopted to delineate thyroid nodules in the B-mode ultrasound image, and then the region of interest is mapped to the corresponding SWE image in a registration mode, wherein the mapping formula is as follows:
ROISWE(x,y)＝ROIBMUS(x+w,y+h)
in the formula, ROIBMUS(x, y) denotes the region of interest, ROI, in B-mode ultrasonographySWE(x, y) denotes a region of interest in the SWE, w and h denote offsets between the SWE image and the B-mode image, respectively, w is half the width of the image, and h is 0.
4. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 3, wherein: in the step 3), a PyRadiomics package of python software is adopted to extract first-order histogram features, shape features, texture features and wavelet features of each region of interest.
5. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 4, wherein: in the step 4), the image omics analysis specifically comprises the following steps:
4.1) feature screening: respectively analyzing B ultrasonic image features and SWE image features of the training set by using a significance analysis method, a correlation analysis method and a Lasso regression analysis method, and removing features irrelevant to an LN state;
4.2) constructing a B-ultrasonic imaging omics label: performing Lasso analysis result coefficients according to B-ultrasonic image characteristics, and forming a B-ultrasonic image omics label in a linear weighting mode;
4.3) constructing a SWE imaging omics tag: carrying out Lasso analysis result coefficients according to the SWE image characteristics, and forming a SWE image omics tag in a linear weighting mode;
4.4) constructing a risk prediction model: and (3) constructing a prediction model of LN transfer risk by performing binary multi-factor Logistic regression analysis on the clinical information, the B-ultrasonic imaging group tag and the SWE imaging group tag.
6. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 5, wherein: in the step 4.1), the significance analysis is to perform T test or Mann-Whitney U test on the image characteristics, perform T test on the characteristics conforming to normal distribution, perform Mann-Whitney U test on the characteristics not conforming to normal distribution, and reject the characteristics with the P value larger than 0.05; performing Spearman correlation analysis on the features which are not removed in the previous step of analysis, and removing the features with the correlation coefficient smaller than 0.8; the Lasso regression analysis is to perform Lasso regression analysis on the features which are not eliminated in the previous step of analysis, and eliminate the features with coefficient of 0.
7. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 5, wherein: in the step 4.2), the formula for forming the B-mode ultrasound imaging omics label in a linear weighting mode is as follows:
RadScore_BMUS＝0.058+0.106×original_shape2D_Perimeter+0.039×wavelet.LH_glszm_ZoneEntropy
wherein RadsCore _ BMUS represents a B-mode ultrasonography tag, origi _ shape2D _ Perimeter represents lesion side length, and wavelet.
8. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 5, wherein: in the step 4.3), the formula for forming the SWE imaging omics tag in a linear weighting manner is as follows:
RadScore_SWE＝0.041+0.133×logarithm_glszm_ZoneEntropy_R-0.045×gradient_glcm_Imc1_B-0.060×wavelet.LH_glszm_SmallAreaLowGrayLevelEmphasis_GREY-0.032×wavelet.LH_ngtdm_Contrast_G
wherein, RadsCore _ SWE represents a SWE image omics label, logarihm _ glszm _ ZoneEntrol _ R represents the regional entropy of a logarithmic filtering gray scale region size matrix under a red channel, gradient _ glcm _ Imc1_ B represents the related information measurement of a gradient filtering gray scale symbiotic matrix under a blue channel, wavelet.
9. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 5, wherein: in the step 4.4), the construction of the prediction model of LN transfer risk is as follows:
wherein NomoScore is a prediction model of LN metastasis Risk, Multifocality is tumor multifocal, US-ported LN status is lymph node metastasis conclusion in an ultrasonic diagnosis report, Radsore _ SWE is a SWE imaging omics label, e represents a natural constant, and Risk represents LN metastasis Risk probability.
10. The method for constructing an ultrasound imaging omics model for lymph node metastasis risk prediction according to claim 9, wherein: if Risk is less than 0.574, the lymph node is not metastasized; if Risk ≧ 0.574, it indicates no metastasis has occurred in the lymph node.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110632990.0A CN113436150A (en) | 2021-06-07 | 2021-06-07 | Construction method of ultrasound imaging omics model for lymph node metastasis risk prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110632990.0A CN113436150A (en) | 2021-06-07 | 2021-06-07 | Construction method of ultrasound imaging omics model for lymph node metastasis risk prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113436150A true CN113436150A (en) | 2021-09-24 |
Family
ID=77803833
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110632990.0A Pending CN113436150A (en) | 2021-06-07 | 2021-06-07 | Construction method of ultrasound imaging omics model for lymph node metastasis risk prediction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113436150A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115132327A (en) * | 2022-05-25 | 2022-09-30 | 中国医学科学院肿瘤医院 | Microsatellite instability prediction system, construction method thereof, terminal equipment and medium |
| CN116416235A (en) * | 2023-04-12 | 2023-07-11 | 北京建筑大学 | Feature region prediction method and device based on multi-mode ultrasonic data |
| CN117173167A (en) * | 2023-11-02 | 2023-12-05 | 中日友好医院(中日友好临床医学研究所) | Method and device for predicting tumor prognosis by image histology machine learning survival model |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130023479A1 (en) * | 2011-07-22 | 2013-01-24 | Protgen Ltd. | Use of nucleolin as a biomarker for lymphangiogenesis in cancer prognosis and therapy |
| CN108921821A (en) * | 2018-06-01 | 2018-11-30 | 中国人民解放军战略支援部队信息工程大学 | Method of discrimination based on the LASSO mammary cancer armpit lymph gland transfering state returned |
| CN110472629A (en) * | 2019-08-14 | 2019-11-19 | 青岛大学附属医院 | A kind of pathological image automatic recognition system and its training method based on deep learning |
| CN111862085A (en) * | 2020-08-03 | 2020-10-30 | 徐州市肿瘤医院 | Method and system for predicting latent N2 lymph node metastasis of peripheral NSCLC |
| CN112102343A (en) * | 2020-08-12 | 2020-12-18 | 海南大学 | Ultrasound image-based PTC diagnostic system |
| CN112216395A (en) * | 2020-09-11 | 2021-01-12 | 中山大学孙逸仙纪念医院 | Axillary lymph node metastasis prediction model for breast cancer patient and construction method thereof |
| RU2740850C1 (en) * | 2020-06-16 | 2021-01-21 | Федеральное государственное бюджетное научное учреждение "Томский национальный исследовательский медицинский центр Российской академии наук" (Томский НИМЦ) | Method for the synchronous lymphatic metastasis prediction in the patients with non-small cell lung cancer following the neoadjuvant chemotherapy |
-
2021
- 2021-06-07 CN CN202110632990.0A patent/CN113436150A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130023479A1 (en) * | 2011-07-22 | 2013-01-24 | Protgen Ltd. | Use of nucleolin as a biomarker for lymphangiogenesis in cancer prognosis and therapy |
| CN108921821A (en) * | 2018-06-01 | 2018-11-30 | 中国人民解放军战略支援部队信息工程大学 | Method of discrimination based on the LASSO mammary cancer armpit lymph gland transfering state returned |
| CN110472629A (en) * | 2019-08-14 | 2019-11-19 | 青岛大学附属医院 | A kind of pathological image automatic recognition system and its training method based on deep learning |
| RU2740850C1 (en) * | 2020-06-16 | 2021-01-21 | Федеральное государственное бюджетное научное учреждение "Томский национальный исследовательский медицинский центр Российской академии наук" (Томский НИМЦ) | Method for the synchronous lymphatic metastasis prediction in the patients with non-small cell lung cancer following the neoadjuvant chemotherapy |
| CN111862085A (en) * | 2020-08-03 | 2020-10-30 | 徐州市肿瘤医院 | Method and system for predicting latent N2 lymph node metastasis of peripheral NSCLC |
| CN112102343A (en) * | 2020-08-12 | 2020-12-18 | 海南大学 | Ultrasound image-based PTC diagnostic system |
| CN112216395A (en) * | 2020-09-11 | 2021-01-12 | 中山大学孙逸仙纪念医院 | Axillary lymph node metastasis prediction model for breast cancer patient and construction method thereof |
Non-Patent Citations (4)
| Title |
|---|
| CHRISTOPH F DIETRICH 等: "Nomogram Based on Shear-Wave Elastography Radiomics Can Improve Preoperative Cervical Lymph Node Staging for Papillary Thyroid Carcinoma", 《HTTPS://WWW.RESEARCHGATE.NET/PUBLICATION/339086022》 * |
| ZHIGUO ZHOU 等: "Predicting Lymph Node Metastasis in Head and Neck Cancer by Combining Many-objective Radiomics and 3-dimensioal Convolutional Neural Network through Evidential Reasoning", 《2018 40TH ANNUAL INTERNATIONAL CONFERENCE OF THE IEEE ENGINEERING IN MEDICINE AND BIOLOGY SOCIETY (EMBC)》 * |
| 王伟镇 等: "影像组学预测甲状腺乳头状癌中央区淋巴结转移", 《南方医科大学学报》 * |
| 赵泓博 等: "基于CT双期增强影像组学预测甲状腺乳头状癌淋巴结转移", 《放射学实践》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115132327A (en) * | 2022-05-25 | 2022-09-30 | 中国医学科学院肿瘤医院 | Microsatellite instability prediction system, construction method thereof, terminal equipment and medium |
| WO2023226217A1 (en) * | 2022-05-25 | 2023-11-30 | 中国医学科学院肿瘤医院 | Microsatellite instability prediction system and construction method therefor, terminal device, and medium |
| CN116416235A (en) * | 2023-04-12 | 2023-07-11 | 北京建筑大学 | Feature region prediction method and device based on multi-mode ultrasonic data |
| CN116416235B (en) * | 2023-04-12 | 2023-12-05 | 北京建筑大学 | Feature region prediction method and device based on multi-mode ultrasonic data |
| CN117173167A (en) * | 2023-11-02 | 2023-12-05 | 中日友好医院(中日友好临床医学研究所) | Method and device for predicting tumor prognosis by image histology machine learning survival model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chang et al. | Computer‐aided diagnosis for classifying benign versus malignant thyroid nodules based on ultrasound images: a comparison with radiologist‐based assessments | |
| Chou et al. | Stepwise logistic regression analysis of tumor contour features for breast ultrasound diagnosis | |
| CN113436150A (en) | Construction method of ultrasound imaging omics model for lymph node metastasis risk prediction | |
| He et al. | A review on automatic mammographic density and parenchymal segmentation | |
| CN112071418B (en) | Gastric cancer peritoneal metastasis prediction system and method based on enhanced CT image histology | |
| CN115661107B (en) | Image analysis method, system and equipment based on bladder cancer risk stratification | |
| US20180053297A1 (en) | Methods and Apparatuses for Detection of Abnormalities in Low-Contrast Images | |
| US20160106388A1 (en) | Breast cancer detection system, breast cancer detection method, breast cancer detection program, and computer-readable recording medium having breast cancer detection program recorded thereon | |
| Pareek et al. | Predicting the spread of vessels in initial stage cervical cancer through radiomics strategy based on deep learning approach | |
| Zhou et al. | Preoperative prediction of axillary lymph node metastasis in patients with breast cancer based on radiomics of gray-scale ultrasonography | |
| Xing et al. | Automatic detection of A‐line in lung ultrasound images using deep learning and image processing | |
| Gunasekaran | Leveraging object detection for the identification of lung cancer | |
| CN104835155A (en) | Fractal-based early-stage breast cancer calcification point computer auxiliary detection method | |
| CN116630680B (en) | Dual-mode image classification method and system combining X-ray photography and ultrasound | |
| Zhang et al. | Deep learning based on ultrasound to differentiate pathologically proven atypical and typical medullary thyroid carcinoma from follicular thyroid adenoma | |
| Hsieh et al. | Stage classification in chronic kidney disease by ultrasound image | |
| CN103606142B (en) | A kind of elastic strain appraisal procedure based on ultrasonic elastic image and system | |
| Lu et al. | A Review of the Role of Ultrasound Radiomics and Its Application and Limitations in the Investigation of Thyroid Disease | |
| CN114764855A (en) | Intelligent cystoscope tumor segmentation method, device and equipment based on deep learning | |
| WO2022153100A1 (en) | A method for detecting breast cancer using artificial neural network | |
| Fan et al. | The Value of CT Radiomics Nomogram Analysis of Parenchyma Surrounding Pulmonary Nodules in the Differentiation of Inflammatory Nodules from Tuberculous Nodules | |
| Tang et al. | The effect of image resolution on convolutional neural networks in breast ultrasound | |
| KR102620046B1 (en) | Method and system for breast ultrasonic image diagnosis using weakly-supervised deep learning artificial intelligence | |
| CN112884759B (en) | Method and related device for detecting metastasis state of axillary lymph nodes of breast cancer | |
| Alam et al. | Automated Breast Cancer Diagnosis Based on Neural Network Algorithms |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210924 |
|
| RJ01 | Rejection of invention patent application after publication |