A thyroid cancer screening method, device, equipment and storage medium

Abstract

The present invention relates to a thyroid cancer screening method, device, equipment and storage medium, which include the following steps: based on the thermal tomography data of thyroid patients and other clinical examination data, and using the Penns biological heat conduction model to construct a TI-RADS grading model; using the ultrasound image and thermal tomography image of the thyroid patient, automatically locate and segment the region of interest in the thermal tomography image to obtain the segmentation result; based on the TI-RADS grading model and the segmentation result, construct a thyroid cancer recognition model; and use the thyroid cancer recognition model to screen for thyroid cancer. Since the thyroid cancer recognition model is constructed based on the TI-RADS grading model and the segmentation result, a non-invasive, contactless, green and environmentally friendly thyroid cancer diagnosis method based on thermal tomography technology can be realized through the thyroid cancer recognition model, which can make up for the defect of insufficient judgment of the nature of the lesion by ultrasonic examination, minimize unnecessary puncture surgery, reduce the patient's pain and medical burden, and has a low operating threshold.

Description

Technical Field

[0001] The present invention relates to the technical field of image data analysis, and in particular to a thyroid cancer screening method, device, equipment and storage medium. Background Art

[0002] At present, thyroid nodules are a common and frequently occurring disease of the endocrine system. With the promotion and application of ultrasound technology and the enhancement of people's health awareness, more and more asymptomatic thyroid nodules are found in health examinations. The detection rate of thyroid nodules in the population is 20%-76%, of which malignant tumors account for 7%-15%. In recent years, thyroid cancer has become one of the malignant tumors with the fastest growing incidence, and the surgical rate of unnecessary thyroid nodules has also increased significantly. Due to the different clinical treatments of benign and malignant thyroid nodules, the impact on the patient's quality of life and the medical expenses involved are also significantly different. Therefore, accurate assessment of the benign and malignant nature of thyroid nodules before surgery can avoid unnecessary surgical resection of benign nodules. At the same time, accurate identification of the type of thyroid nodules before surgery has great reference value for formulating appropriate surgical plans, stratifying prognostic risk factors, and formulating personalized and precise treatment plans.

[0003] In related technologies, since ultrasound imaging has a low diagnostic rate for malignant lesions, fine needle aspiration biopsy (FNAB) is generally used to identify the nature of thyroid nodules in clinical practice. However, FNAB is an invasive surgery with certain surgical risks, which can also cause a great psychological burden on patients. At the same time, the large number of benign lesions in the FNAB examination results also indicates that there are quite a few unnecessary FNAB surgeries.

[0004] Therefore, it is necessary to design a new thyroid cancer screening method to overcome the above problems. Summary of the invention

[0005] The embodiments of the present invention provide a thyroid cancer screening method, device, equipment and storage medium to solve the problem in the related art that FNAB is an invasive surgery and a large number of benign lesions in the FNAB examination results also indicate the existence of quite a lot of unnecessary FNAB surgeries.

[0006] In a first aspect, a thyroid cancer screening method is provided, which includes the following steps: based on the thermal tomography data and other clinical examination data of thyroid patients, a TI-RADS grading model is constructed using the Penns biological heat conduction model; using the ultrasound images and thermal tomography images of thyroid patients, the region of interest in the thermal tomography images is automatically located and segmented to obtain a segmentation result; based on the TI-RADS grading model and the segmentation result, a thyroid cancer recognition model is constructed; and the thyroid cancer recognition model is used to screen for thyroid cancer.

[0007] In some embodiments, the method of using the ultrasound image and thermal tomography image of the thyroid patient to automatically locate and segment the region of interest in the thermal tomography image to obtain the segmentation result includes: automatically aligning the ultrasound image and the thermal tomography image, and automatically obtaining the corresponding annotated region of the thermal tomography image based on the annotation of the lesion and abnormal blood flow signals in the ultrasound image; and automatically locating and segmenting the region of interest in the thermal tomography image based on the corresponding annotated region of the thermal tomography image to obtain the segmentation result.

[0008] In some embodiments, the automatic alignment of the ultrasound image and the thermal tomography image, and the automatic acquisition of the corresponding annotated area of ​​the thermal tomography image based on the annotation of the lesions and abnormal blood flow signals in the ultrasound image, include: clustering each pixel point in the thermal tomography image of each patient according to the Qr curve parsed according to the corresponding temperature; determining the clustering point group of interest with the corresponding annotated ultrasound image as a guide, and aligning the thermal tomography image with the corresponding ultrasound annotated area, and then marking the thermal tomography image accordingly to obtain the corresponding annotated area of ​​the thermal tomography image.

[0009] In some embodiments, the corresponding annotated areas of the thermal tomography images are used to automatically locate and segment the regions of interest in the thermal tomography images to obtain segmentation results, including: automatically locating and segmenting the regions of interest based on a deep neural network model of multi-scale feature semantic segmentation based on the corresponding annotated areas of the thermal tomography images obtained by alignment.

[0010] In some embodiments, the constructing of a thyroid cancer recognition model based on the TI-RADS grading model and the segmentation results includes: constructing a thyroid cancer recognition model based on thermal tomography images, a thyroid cancer recognition model based on ultrasound images, or a thyroid cancer recognition model based on multimodal data fusion using a deep learning algorithm based on the TI-RADS grading model and the segmentation results.

[0011] In some embodiments, the TI-RADS grading model is constructed based on the thermal tomography data and other clinical examination data of thyroid patients and using the Penns biological heat conduction model, including: obtaining thermal tomography data and other clinical examination data of multiple thyroid patients with quality that meets the requirements; based on the Penns biological heat conduction model, analyzing and obtaining the thermal distribution model of different depths in the thyroid gland; and constructing the TI-RADS grading model based on the thermal tomography data and other clinical examination data and the thermal distribution model of different depths.

[0012] In some embodiments, the method of analyzing and obtaining the thermal distribution model at different depths in the thyroid gland based on the Penns biological heat conduction model includes: analyzing the thermal distribution model at different depths in the thyroid gland layer by layer based on the Penns biological heat conduction model, and performing numerical simulation, and fitting the actual data measured by the thermal tomography equipment under the guidance of expert knowledge to determine the optimal analysis depth; based on the body surface temperature distribution collected by the thermal tomography equipment, using the Petmes biological heat transfer equation, and obtaining the size and distribution information of the internal heat source from the body surface temperature distribution through an inversion algorithm.

[0013] In a second aspect, a thyroid cancer screening device is provided, which includes: a first construction module, which is used to construct a TI-RADS grading model based on the thermal tomography data and other clinical examination data of thyroid patients and using the Penns biological heat conduction model; a segmentation module, which is used to use the ultrasound image and thermal tomography image of the thyroid patient to automatically locate and segment the area of ​​interest in the thermal tomography image to obtain a segmentation result; a second construction module, which is used to construct a thyroid cancer recognition model based on the TI-RADS grading model and the segmentation result; a screening module, which is used to screen for thyroid cancer using the thyroid cancer recognition model.

[0014] According to a third aspect, a computing device is provided, comprising a memory and a processor, wherein the memory is used to store computer instructions; and the processor executes the computer instructions stored in the memory to perform the above method.

[0015] According to a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program code. When the computer program code is executed by a computing device, the computing device executes the above method.

[0016] The beneficial effects brought about by the technical solution provided by the present invention include:

[0017] The embodiments of the present invention provide a thyroid cancer screening method, device, equipment and storage medium. Since a thyroid cancer recognition model is constructed based on the TI-RADS grading model and segmentation results, a non-invasive, non-contact, green and environmentally friendly thyroid cancer diagnosis method based on thermal chromatography technology can be realized through the thyroid cancer recognition model, which can make up for the defect of insufficient judgment of the nature of lesions by ultrasonic examination, minimize unnecessary puncture surgery, and reduce patients' pain and medical burden. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 A flow chart of a thyroid cancer screening method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] The embodiments of the present invention provide a thyroid cancer screening method, device, equipment and storage medium, which can solve the problem in the related art that FNAB is an invasive surgery and a large number of benign lesions in the FNAB examination results also indicate the existence of quite a lot of unnecessary FNAB surgeries.

[0022] See also Figure 1 As shown, a thyroid cancer screening method provided by an embodiment of the present invention may include the following steps:

[0023] S1: Based on the thermal tomography data and other clinical examination data of thyroid patients, and using the Penns biological heat conduction model, a TI-RADS grading model is constructed. In this embodiment, in order to explore the thermal metabolic changes of thyroid lesions through thermal tomography data, it is planned to collect thermal tomography data and other clinical examination data of no less than 10,000 thyroid patients that meet the quality requirements, and the more patients, the better.

[0024] S2: Using the ultrasound image and thermal tomography image of the thyroid patient, the region of interest in the thermal tomography image is automatically located and segmented to obtain the segmentation result.

[0025] S3: Based on the TI-RADS grading model and segmentation results, a thyroid cancer recognition model was constructed.

[0026] S4: Screening for thyroid cancer using the thyroid cancer identification model.

[0027] Among the related technologies, thyroid ultrasound examination currently has a low diagnostic rate for malignant lesions (thyroid cancer). In general, cytopathology, namely fine needle aspiration biopsy (FNAB) of thyroid nodules, is currently the most sensitive and specific method for preoperative evaluation of benign and malignant thyroid nodules, and has rich evidence-based medicine. However, FNAB is an invasive examination with certain risks, and requires sampling by experienced doctors under ultrasound guidance. Due to the lack of professional medical equipment and experienced doctors, FNAB is unlikely to be used for large-scale physical examination screening. On the other hand, in cases where FNAB surgery is indicated based on ultrasound diagnostic results, a considerable portion of the cytological examination results are benign lesions.

[0028] Since the embodiment of the present invention constructs a thyroid cancer recognition model based on the TI-RADS grading model and segmentation results, the thyroid cancer recognition model can realize a non-invasive, non-contact, green and environmentally friendly thyroid cancer diagnosis method based on thermal tomography technology, which can make up for the defect of insufficient judgment of the nature of lesions by ultrasonic examination, minimize unnecessary puncture operations, and reduce patients' pain and medical burden.

[0029] Thyroid nodules are a relatively common disease of the thyroid gland, mainly due to the appearance of lumps after abnormal proliferation of cells inside the thyroid gland. Thyroid nodules may become malignant, and malignant nodules lead to thyroid cancer. Ultrasound examination obtains information such as the size, morphology, and contour boundaries of the lesion by analyzing the echo signals generated after emitting ultrasound to different tissues. Thermal tomography does not emit any signals, but performs functional imaging based on the captured cell metabolic radiation heat data, that is, it detects the functional signals of human organs or tissues, and therefore cannot intuitively reflect the morphological information of the lesion area. Therefore, it is necessary to determine whether there is a lump in the thyroid gland based on thermal tomography data, and further determine the nature of the lump, how the metabolic heat distribution of the thyroid lump differs from that of normal tissue, and determine the metabolic heat differences between benign and malignant masses.

[0030] The method for obtaining the TI-RADS grading model in step S1 may include the following steps: obtaining thermal tomography data and other clinical examination data of multiple thyroid patients that meet the quality requirements; analyzing and obtaining the thermal distribution model of different depths in the thyroid gland based on the Penns biological heat conduction model; and constructing the TI-RADS grading model based on the thermal tomography data and other clinical examination data and the thermal distribution model of different depths. After the TI-RADS grading model is constructed, any thermal tomography data can be automatically graded.

[0031] In this embodiment, since the thermal distribution model of different depths within the thyroid gland is obtained by analysis, based on the thermal distribution model and through thermal tomography data and other clinical examination data, the pattern of changes in cell metabolic thermal distribution with thyroid disease can be explored, and objective TI-RADS grading recommendations that do not rely on the doctor's subjective experience and operating techniques can be obtained.

[0032] Furthermore, the other clinical examination data may include: ultrasonic medical imaging diagnosis data, thyroid cytology diagnosis data, five thyroid function test data, etc. Of course, other clinical examination data may also be included. More data is conducive to the accurate assessment of thyroid nodules. In this embodiment, by fusing multiple types of clinical data, the thyroid thermal tomography data can be fully interpreted.

[0033] Preferably, in order to establish an artificial intelligence model, the collected data needs to be preprocessed, that is, after obtaining thermal tomography data and other clinical examination data of multiple thyroid patients that meet the quality requirements, it can also include: preprocessing the acquired thermal tomography data and other clinical examination data, wherein the preprocessing includes the annotation of abnormal blood flow signals in the ultrasound image and the manual annotation of the thyroid lesion area. In order to facilitate manual annotation and reduce the workload of annotation, it can be performed based on existing annotation tools.

[0034] In some embodiments, the thermal tomography device is to capture the far-infrared light sensing data of the inspected part, convert it into temperature data through digital analog, and then describe the thermal field distribution at different depths through the biological heat transfer model. Therefore, it is necessary to determine the biological heat transfer model and the thyroid analysis depth (number of layers). Generally, the normal thyroid ultrasound measurement value is 4.0 to 5.0 cm in long diameter; 2.0 to 2.5 cm in wide diameter; 1.0 to 1.5 cm in thick diameter, and the isthmus thickness is ≤ 0.4 cm. The method of analyzing and obtaining the thermal distribution model at different depths in the thyroid gland based on the Penns biological heat conduction model may include: the method intends to analyze the thermal distribution model at different depths in the thyroid gland layer by layer based on the Penns biological heat conduction model, and perform numerical simulation, and fit it with the actual data measured by the thermal tomography device under the guidance of expert knowledge to determine the optimal analysis depth; based on the body surface temperature distribution collected by the thermal tomography device, the Petmes biological heat transfer equation is used to obtain the size and distribution information of the internal heat source from the body surface temperature distribution through the inversion algorithm.

[0035] In the field of bioheat transfer research and application, researchers have always believed that the Petmes bioheat transfer equation better reflects the thermal mechanism and heat transfer law inside the human body and is a more ideal bioheat transfer model:

[0036]

[0037] Where T(x, y, z) is the temperature field distribution function in the body; ρ and c are the density and heat capacity of the body; k is the heat transfer coefficient; w b , b 、c b is the blood perfusion rate, blood density, and blood heat capacity; T a is the arterial blood temperature; Q m is the metabolic heat in the body; b ρ b c b (T a -T) is the blood perfusion term, and the last two terms are the internal heat source terms.

[0038] Based on the body surface temperature distribution collected by thermal tomography equipment, the above model can be used to obtain the size and distribution information of the internal heat source value from the body surface temperature distribution through the inversion algorithm. The size of the heat value directly reflects the state of metabolism in the body and can help determine the location, nature and degree of the lesion.

[0039] The present invention mainly adopts two methods to obtain the TI-RADS grading model, and which one to adopt can be determined based on experimental results.

[0040] In some embodiments, one method is to obtain a TI-RADS grading model based on a statistical law method, which is similar to the breast Qr curve grading diagnosis method currently summarized based on thermal tomography. That is, the TI-RADS grading model is constructed based on thermal tomography data and other clinical examination data and thermal distribution models at different depths, including: inputting the thermal tomography data and other clinical examination data collected from all patients into thermal distribution models at different depths for analysis to determine the thermal intensity distribution at different depths; and drawing Qr curves based on the thermal intensity distribution at different depths, and classifying and statistically analyzing them according to different diagnostic types, summarizing the general laws of thermal intensity and depth distribution, and deriving the TI-RADS grading model accordingly. If, after experiments, different grades are statistically clearly distinguishable, this relatively simple and intuitive method is adopted.

[0041] In some optional embodiments, another method is to use machine learning methods to build a grading model, that is, the TI-RADS grading model is built based on thermal tomography data and other clinical examination data and thermal distribution models at different depths, including: labeling thermal tomography data according to ultrasound grading results, collecting patients' thermal tomography analysis data as input to the machine learning algorithm, building a TI-RADS grading model, and finally realizing automatic grading of any thermal tomography data. Among them, the ultrasound grading result can be the ultrasound TI-RADS grading obtained by clinical physicians based on the experience of ultrasound medical imaging features of thyroid patients in the early stage.

[0042] Furthermore, thermal tomography equipment is a new type of medical imaging equipment. Although it has many clinical applications in breast cancer screening, its clinical application for thyroid cancer screening has not yet been carried out. This method requires the use of mature thyroid examination methods such as ultrasound to build models and discover patterns. Especially in the study of automatic positioning of thyroid lesion areas, ultrasound medical images are needed to guide data labeling in the early stage. Therefore, it is necessary to align two different types of images, ultrasound medical images and thermal tomography images.

[0043] In step S2, the method of using the ultrasound image and thermal tomography image of the thyroid patient to automatically locate and segment the region of interest in the thermal tomography image to obtain the segmentation result may include: automatically aligning the ultrasound image and the thermal tomography image, automatically obtaining the corresponding annotated region of the thermal tomography image based on the annotation of the lesion and abnormal blood flow signal in the ultrasound image, so that the automatic positioning and segmentation algorithm of the target of interest can be established based on the annotated data later; based on the corresponding annotated region of the thermal tomography image, automatically locating and segmenting the region of interest in the thermal tomography image to obtain the segmentation result. In this embodiment, by studying the automatic positioning and segmentation method of the lesion site in the thermal tomography image, the lesion region of interest is extracted from the thyroid thermal tomography image for key analysis, thereby improving the accuracy of diagnosis.

[0044] However, ultrasound images and thermal tomography images belong to two different imaging mechanisms after all. Therefore, the above alignment does not require each pixel to be matched one by one. Preferably, the automatic alignment of ultrasound images and thermal tomography images, based on the annotation of lesions and abnormal blood flow signals in the ultrasound image, automatically obtains the corresponding annotated area of ​​the thermal tomography image, which can include: clustering each pixel in the thermal tomography image of each patient according to the Qr curve parsed from the corresponding temperature; using the corresponding annotated ultrasound image as a guide, determining the cluster of points of interest, and aligning the thermal tomography image with the corresponding ultrasound annotated area, and then marking the thermal tomography image accordingly to obtain the corresponding annotated area of ​​the thermal tomography image. Among them, the clustering method initially considers a density-based clustering algorithm, and adjustments and improvements based on actual results will also be considered during the research process.

[0045] In some embodiments, the automatic positioning and segmentation of the region of interest in the thermal tomography image based on the corresponding annotated region of the thermal tomography image to obtain the segmentation result may include: automatically positioning and segmenting the region of interest based on a deep neural network model of multi-scale feature semantic segmentation based on the corresponding annotated region of the thermal tomography image obtained by alignment. In this embodiment, the deep neural network model is preferably a U-Net network model.

[0046] The original U-Net network was published in 2015 and is used for medical cell segmentation. The U-Net network is a classic fully convolutional network with a U-shaped structure, divided into two parts. The input of the network is a 572*572 image with mirrored edges (input image tile). The left side of the network is called the contracting path, which is a series of downsampling operations consisting of convolution and Max Pooling. The contracting path consists of 4 blocks, each of which uses 3 effective convolutions and 1 Max Pooling downsampling. After each downsampling, the number of feature maps is multiplied by 2. The right part of the network is called the expansive path. It also consists of 4 blocks. Before each block starts, the size of the feature map is multiplied by 2 through deconvolution, and its number is halved (the last layer is slightly different), and then merged with the feature map of the symmetrical contracting path on the left. Since the size of the feature map of the left contracting path and the right expansion path is different, U-Net crops the feature map of the contracting path to the same size as the expansion path and merges them to finally get the segmentation result.

[0047] Furthermore, since the thermal tomography image represents a heat map of temperature information, in order to obtain more semantic information describing the pixels in the image, in the specific implementation, it can be considered to incorporate information such as the thermal intensity distribution model of different depths of the thermal tomography data analysis to characterize each pixel to achieve better segmentation results. For blood vessels, it can be considered to calculate the convex hull of the point set based on the relevant algorithm in computer geometry, and then calculate the longest diameter and shortest diameter of the convex hull, and post-process the segmentation results to identify the blood vessels. On the one hand, the metabolism of blood cells is generally more active, and on the other hand, the vascular morphology of the tumor site is different from that of normal tissue. Therefore, it is necessary to automatically identify blood vessels by analyzing the thermal tomography data.

[0048] In some optional embodiments, in step S3, the construction of a thyroid cancer recognition model based on the TI-RADS grading model and the segmentation results may include: based on the TI-RADS grading model and the segmentation results, using a deep learning algorithm to construct a thyroid cancer recognition model based on thermal tomography images, a thyroid cancer recognition model based on ultrasound images, or a thyroid cancer recognition model based on multimodal data fusion. Specifically, the results of the patient's cytological examination can be used as the category labels of the corresponding thermal tomography images and ultrasound images. It can be considered to adopt a comparative study method, using a deep learning algorithm to respectively construct a thyroid cancer recognition model based on thermal tomography images, a thyroid cancer recognition model based on ultrasound images, and a thyroid cancer recognition model based on multimodal data fusion such as thermal tomography, ultrasound, and thyroid function tests, perform specific analysis and improvement based on the test results, and finally build a thyroid cancer screening system suitable for different data scenarios. For the problem of multimodal data fusion, it can be considered to design a special convolution kernel, or to use a graph embedded method to achieve the organic fusion of different types of data. This embodiment utilizes machine learning algorithms such as deep learning, and integrates the functional information obtained from the aforementioned study on the relationship between thyroid disease and metabolic thermal changes, as well as the morphological information of the segmented lesion area, to construct an intelligent screening method for thyroid cancer that is highly consistent with the cell puncture results, providing a decision-making reference for the selection of further clinical treatment plans.

[0049] This method provides a data set for establishing an artificial intelligence model by collecting, preprocessing and annotating clinical examination data of thyroid patients; determines the thyroid thermal tomography depth through expert knowledge and the Penns model and obtains the thermal distribution model of different layers through inversion analysis to help determine the location of the lesion, the nature and extent of the lesion, etc. The TI-RADS grading model is constructed through the thermal distribution model; the corresponding annotation information of the thermal tomography image is automatically obtained by annotating the lesions and (abnormal) blood flow signals in the ultrasound image, and this is used as the basis of the data set to realize the automatic positioning and segmentation of the region of interest in the thermal tomography image, and to build a thyroid cancer screening model.

[0050] The embodiment of the present invention also provides a thyroid cancer screening device, which may include: a first construction module, which is used to construct a TI-RADS grading model based on the thermal tomography data and other clinical examination data of thyroid patients and using the Penns biological heat conduction model; a segmentation module, which is used to use the ultrasound image and thermal tomography image of the thyroid patient to automatically locate and segment the area of ​​interest in the thermal tomography image to obtain the segmentation result; a second construction module, which is used to construct a thyroid cancer recognition model based on the TI-RADS grading model and the segmentation result; a screening module, which is used to screen for thyroid cancer using the thyroid cancer recognition model. Among them, the first construction module, the segmentation module and the second construction module can correspondingly implement the method involved in any of the above embodiments, which will not be repeated here.

[0051] The present invention integrates relevant data such as ultrasonic medical imaging, thyroid function blood tests, cytological tests, etc., interprets the thyroid biological metabolic thermal data obtained by thermal tomography equipment, reveals the relationship between thyroid disease and metabolic thermal changes, and further intelligently analyzes information such as the heat source depth, distribution, and thermal radiation value of the thyroid lesion site to obtain an objective judgment on the nature of the lesion, provide a basis for further selection of clinical treatment plans, and lay a theoretical and methodological foundation for the future development of intelligent thermal tomography thyroid screening equipment.

[0052] The embodiment of the present invention further provides a computing device, the computing device comprising a memory and a processor, the memory being used to store computer instructions; the processor executing the computer instructions stored in the memory to execute the method described in any of the above embodiments. In addition, the present invention helps to develop harmless, non-invasive, green and environmentally friendly thyroid cancer screening equipment, and has a low operating threshold, which can be widely used in a large number of physical examination institutions and grassroots hospitals for large-scale health censuses, and has significant economic and social benefits.

[0053] An embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer program code. When the computer program code is executed by a computing device, the computing device executes the method described in any one of the above embodiments.

[0054] Medical AI technology has developed rapidly in medical imaging, and more and more landing projects have emerged. However, a thyroid cancer diagnosis system based on thermal tomography technology that integrates multi-source heterogeneous data such as thyroid color ultrasound data, thyroid cytology data, and thyroid function test data has not yet been developed. The thyroid cancer diagnosis system based on thermal tomography technology and artificial intelligence technology can make up for the shortcomings of ultrasound examination due to its non-invasive, contactless, green and environmentally friendly, simple and convenient operation, and can be widely used in various hospitals to achieve early screening of thyroid cancer. The present invention proposes to study artificial intelligence methods such as machine learning, and construct an automatic segmentation model for lesions and an automatic classification model for thyroid diseases based on thyroid thermal tomography data. This method basically has no inspection blind spots and is advanced.

[0055] The descriptions of the processes corresponding to the above-mentioned figures have different emphases. For parts that are not described in detail in a certain process, please refer to the relevant descriptions of other processes.

[0056] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The above computer program product includes one or more computer instructions. When the above computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The above computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The above computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the above computer instructions can be transmitted from a website site, a computer, a server or a data center by wired (such as coaxial cable, optical fiber, digital subscriber line) or wireless (such as infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center. The above computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, a data center, etc. that contains one or more available media integrated. The above available medium can be a magnetic medium, (such as a floppy disk, a storage disk, a tape), an optical medium (such as a DVD), or a semiconductor medium (such as an SSD), etc. In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0057] In the several embodiments provided in the present application, it should be understood that the disclosed device can also be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the indirect coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be electrical or other forms.

[0058] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present application.

[0059] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0060] If the above-mentioned integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium may include, for example: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory, a random access memory, a magnetic disk or an optical disk.

[0061] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for screening thyroid cancer, characterized in that: It includes the following steps: Obtain thermal tomography data and other clinical examination data of multiple thyroid patients that meet the quality requirements; Based on the Penns biological heat conduction model, the heat distribution model at different depths in the thyroid gland is analyzed layer by layer, and numerical simulation is performed. Under the guidance of expert knowledge, the actual data measured by the thermal tomography equipment is fitted to determine the optimal analysis depth; Based on the body surface temperature distribution collected by thermal tomography equipment, the Petmes bioheat transfer equation is used to obtain the size and distribution information of the internal heat source from the body surface temperature distribution through the inversion algorithm; Construct a TI-RADS grading model based on thermal tomography data and other clinical examination data as well as thermal distribution models at different depths; Cluster the Qr curves obtained by analyzing the corresponding temperature for each pixel in the thermal tomography image of each patient; Using the corresponding annotated ultrasound image as a guide, determine the cluster of points of interest, align the thermal tomography image with the corresponding ultrasound annotated area, and then mark the thermal tomography image accordingly to obtain the corresponding annotated area of ​​the thermal tomography image; Based on the corresponding annotated area of ​​the thermal tomography image, the region of interest in the thermal tomography image is automatically located and segmented to obtain a segmentation result; Based on the TI-RADS grading model and segmentation results, a thyroid cancer recognition model was constructed; Screening for thyroid cancer using the thyroid cancer identification model.

2. The thyroid cancer screening method according to claim 1, characterized in that: The method of automatically locating and segmenting the region of interest in the thermal tomography image based on the corresponding annotated region of the thermal tomography image to obtain a segmentation result includes: By aligning the corresponding annotated areas of the thermal tomography images, the region of interest is automatically located and segmented based on a deep neural network model with multi-scale feature semantic segmentation.

3. The method for screening thyroid cancer according to claim 1, characterized in that: The thyroid cancer recognition model is constructed based on the TI-RADS grading model and the segmentation results, including: According to the TI-RADS grading model and segmentation results, a deep learning algorithm is used to construct a thyroid cancer recognition model based on thermal tomography images, a thyroid cancer recognition model based on ultrasound images, or a thyroid cancer recognition model based on multimodal data fusion.

4. A thyroid cancer screening device, characterized in that: It includes: A first building module is used to obtain thermal chromatography data and other clinical examination data of multiple thyroid patients that meet quality requirements; Based on the Penns biological heat conduction model, the heat distribution model at different depths in the thyroid gland is analyzed layer by layer, and numerical simulation is performed. Under the guidance of expert knowledge, the model is fitted with the actual data measured by the thermal tomography equipment to determine the optimal analysis depth. Based on the body surface temperature distribution collected by the thermal tomography equipment, the Petmes biological heat transfer equation is used to obtain the size and distribution information of the internal heat source from the body surface temperature distribution through the inversion algorithm. Based on the thermal tomography data and other clinical examination data as well as the heat distribution models at different depths, the TI-RADS grading model is constructed. The segmentation module is used to cluster the Qr curves parsed from the corresponding temperature for each pixel in the thermal tomography image of each patient; determine the clusters of interest based on the corresponding annotated ultrasound image, align the thermal tomography image with the corresponding ultrasound annotated area, and then mark the thermal tomography image accordingly to obtain the corresponding annotated area of ​​the thermal tomography image; Based on the corresponding annotated area of ​​the thermal tomography image, the region of interest in the thermal tomography image is automatically located and segmented to obtain a segmentation result; The second building module is used to build a thyroid cancer recognition model based on the TI-RADS grading model and segmentation results; A screening module is used to screen thyroid cancer using a thyroid cancer recognition model.

5. A computing device, characterized in that: The computing device comprises a memory and a processor, wherein the memory is used to store computer instructions; the processor executes the computer instructions stored in the memory to perform the method according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program code. When the computer program code is executed by a computing device, the computing device executes the method according to any one of claims 1 to 3.

Images