CT image processing method, CT image classification method, device, medium and program product

Abstract

The present invention relates to the field of medical image processing, and in particular to a pancreas CT image processing method. The method comprises: obtaining a first non-contrast phase image and a portal venous phase image of a pancreas CT image of a patient with chronic pancreatitis; segmenting the pancreas, the stone and the pancreatic duct in the portal venous phase image to obtain a mask; using the mask and the first non-contrast phase image to train an image segmentation model to obtain a trained image segmentation model; obtaining a second non-contrast phase image of the patient under test, and inputting the second non-contrast phase image into the trained image segmentation model; and processing the second non-contrast phase image on the basis of the trained segmentation model to obtain parameters of a plurality of predetermined features. The present invention also provides a pancreas CT image classification method, a device, a medium and a program product.

Description

Method, device, medium and program product for processing CT images, and method for classifying CT images

[0001] The present application claims priority to the Chinese patent application No. 202410424618.4, filed on April 9, 2024, and entitled "Method for processing CT images, method for classifying CT images, device, medium and program product", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the field of medical image processing, in particular to a method for processing CT images of pancreas, a method for classifying CT images of pancreas, a device, a medium and a program product. BACKGROUND

[0003] The clinical and pathological features of chronic pancreatitis are progressive atrophy, fibrosis, pain response, pancreatic duct distortion, stenosis or dilation, stones, and endocrine and exocrine dysfunction of the pancreas, which gradually worsens as the disease progresses. Accurate classification of chronic pancreatitis is crucial for understanding its severity, monitoring the disease, and guiding treatment.

[0004] The most widely used Cambridge classification standard and M-ANNHEIM classification standard have the following problems: not based on pathological gold standard, only using imaging morphology as the basis for classification, and cannot reflect the function of the pancreas. Therefore, the Cambridge classification standard is not effective for long-term clinical practice and comparison of data between institutions. For patients with chronic pancreatitis, an economic, effective and easy-to-use scoring system that accurately reflects the morphological characteristics of the disease and predicts short-term and medium-term prognosis is essential. It not only helps clinicians accurately understand the disease and provide timely treatment guidance, but also facilitates disease follow-up and comparison between patients.

[0005] In the prior art, enhanced CT of the pancreas is usually used to diagnose chronic pancreatitis. However, enhanced CT has a large radiation dose, and patients may have the risk of iodine allergy. Patients with poor liver and kidney function are not suitable for enhanced CT examination. Compared with enhanced CT, non-enhanced CT has a low radiation dose, is economical, fast, and does not have side effects due to contrast agents. There is currently no scoring system for chronic pancreatitis prognosis based on non-enhanced CT and artificial intelligence. SUMMARY

[0006] The present application provides a method for processing CT images of pancreas, a method for classifying CT images of pancreas, a device, a medium and a program product, which can be applied to the classification and diagnosis of chronic pancreatitis prognosis.

[0007] An embodiment of the present application discloses a method for processing CT images of pancreas, the method comprising:

[0008] obtain a first plain scan image and a portal phase image of a pancreas CT image of a chronic pancreatitis patient;

[0009] segment the pancreas, stone and pancreatic duct in the portal phase image to obtain a mask;

[0010] train an image segmentation model using the mask and the first plain scan image to obtain a trained image segmentation model;

[0011] obtain a second plain scan image of a patient to be detected, and input the second plain scan image into the trained image segmentation model;

[0012] process the second plain scan image based on the trained image segmentation model to obtain parameters of a plurality of predetermined features.

[0013] Optionally, training the image segmentation model using the mask and the first plain scan image further comprises registering the first plain scan image and the portal phase image, and generating a mask corresponding to the first plain scan image based on the mask.

[0014] Optionally, the image segmentation model is trained using a 3D-UNET algorithm based on the mask corresponding to the first plain scan image.

[0015] Embodiments of the present application disclose a classification method of a pancreas CT image, the method comprising:

[0016] obtain a pancreas plain scan image of a chronic pancreatitis patient and an endocrine index and an exocrine index of the chronic pancreatitis patient;

[0017] classify based on an image classification model;

[0018] The image classification model is obtained by the following training method:

[0019] use the pancreas CT image processing method to obtain parameters of a plurality of predetermined features,

[0020] use the parameters of the plurality of predetermined features, the endocrine index and the exocrine index as samples to train the image classification model.

[0021] Embodiments of the present application disclose an electronic device, characterized in that the device comprises a processor and a memory storing computer executable instructions, the processor is configured to execute the instructions to implement the above-mentioned pancreas CT image processing method and the above-mentioned pancreas CT image classification method.

[0022] The embodiment of the present application discloses a computer readable storage medium, characterized in that at least one computer instruction is stored in the computer readable storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-mentioned pancreatic CT image processing method and the above-mentioned pancreatic CT image classification method.

[0023] The embodiment of the present application discloses a computer program product, characterized in that the computer program product comprises computer instructions, and the computer instructions are executed to implement the above-mentioned pancreatic CT image processing method and the above-mentioned pancreatic CT image classification method.

[0024] The present application provides a pancreatic CT image processing method, classification method, device, medium and program product, by annotating the pancreas, pancreatic duct and stone on the portal phase image of the pancreatic CT image and performing segmentation, registering the plain scan phase image and the portal phase image, migrating the mask to the plain scan phase image through transfer learning, training the image segmentation model based on the plain scan phase image using the 3D-UNET algorithm, using the trained image segmentation model to obtain the mask of the plain scan phase image of the patient to be detected, and using the classification model based on the mask to classify the plain scan phase image of the patient to be detected, and finally realizing the prognosis classification diagnosis and judging the severity of chronic pancreatitis.

[0025] The embodiment of the present application compared with the prior art, the main difference and its effect are that: based on the pancreatic CT image processing method of the present application, when making chronic pancreatitis prognosis diagnosis, the patient's pancreatic enhanced CT image does not need to be obtained, only the patient's plain scan CT image is needed, the existing enhanced CT image of the diagnosed patient is used, combined with deep learning to realize the prognosis diagnosis of chronic pancreatitis based on non-enhanced CT image, thereby replacing the method of using enhanced CT, which has the advantages of non-invasive, no risk of iodine allergy, low cost and friendly to people with poor liver and kidney function, and further, the CT features obtained by using the image segmentation model and the chronic pancreatitis clinical image prognosis scoring (CP-CRPS) system generated based on the image classification model, endocrine index and exocrine index generate chronic pancreatitis clinical image prognosis scoring (CP-CRPS) system, which is highly related to chronic pancreatitis prognosis and can predict prognosis. BRIEF DESCRIPTION OF DRAWINGS

[0026] Fig. 1 is a flowchart of a pancreatic CT image processing method according to an embodiment of the present application;

[0027] Fig. 2 is a schematic diagram of a chronic pancreatitis clinical image prognosis scoring (CP-CRPS) system according to an embodiment of the present application;

[0028] Fig. 3 is a hardware structure block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] The application will be further described below in connection with specific embodiments and drawings. It can be understood that the specific embodiments described herein are merely intended to explain the application, but not to limit the application. In addition, only the parts related to the application are shown in the drawings for the convenience of description, rather than all the structures or processes. It should be noted that in the specification, similar reference numbers and letters represent similar items in the following drawings.

[0030] It should be understood that although the terms "first", "second" and the like can be used herein to describe various features, these features should not be limited by these terms. These terms are only used to distinguish one feature from another. For example, a first feature could be termed a second feature, and, similarly, a second feature could be termed a first feature without departing from the scope of the example embodiments.

[0031] To make the objectives, technical solutions and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0032] The clinical pathological process of chronic pancreatitis gradually worsens with the progression of the disease, and may eventually worsen into pancreatic cancer. Pancreatic cancer is a highly lethal malignant tumor with a low survival rate. Therefore, early diagnosis and prognosis are of great relevance to the control of chronic pancreatitis lesions. The classification diagnosis of chronic pancreatitis is the key to mastering the severity of chronic pancreatitis, monitoring the disease, and guiding treatment.

[0033] CT plays a crucial role in the diagnosis, assessment of disease condition, and evaluation of therapeutic effect of chronic pancreatitis, and is the most commonly used examination method. However, previous diagnosis of chronic pancreatitis has been focused on morphological semi-quantitative diagnosis, which is difficult to accurately assess. In recent years, with the widespread application of artificial intelligence in the medical field, full-automatic segmentation and accurate quantification based on CT have become possible. However, so far, there has been no study on the scoring of the severity of chronic pancreatitis based on full-automatic segmentation and quantification of CT, and there is no precise quantification of chronic pancreatitis classification system based on CT. The existing technology mostly uses pancreatic enhanced CT. However, the defects of enhanced CT are large X-ray dose, risk of iodine allergy, and factors such as unsuitable for people with poor liver and kidney function to receive enhanced CT examination, high cost, and the like, which limit the application of enhanced CT in the diagnosis of chronic pancreatitis.

[0034] To solve the above problems, one embodiment of the present application provides a method for processing pancreatic CT images. The method uses CT plain scanning technology, obtains 32 indexes of precise segmentation of pancreas, stones and pancreatic duct based on the 3D-UNET of the plain scan period image, selects 7 most relevant indexes, and combines with the endocrine and exocrine indexes of the pancreas to train the image classification and generate a full-automatic scoring system. The image classification model of the present application is based on plain CT, does not need to inject contrast medium, and has the characteristics of non-invasive, convenient, fast and economical. The full-automatic quantitative scoring of the image classification model of the present application is a precise quantitative chronic pancreatitis classification system based on CT, which is simple and convenient to operate and accurate in evaluation.

[0035] Fig. 1 is a flowchart of a method for processing pancreatic CT images according to an embodiment of the present application.

[0036] S110: Obtain a first plain scan period image and a portal period image of the pancreatic CT image of a chronic pancreatitis patient.

[0037] In enhanced CT scanning, the contrast medium is injected into the blood vessels, and then flows to various organs and lesions with the blood flow. The plain scan period image refers to the CT scan image obtained before the contrast medium is developed. The contrast medium development consists of three periods, namely, the arterial phase, the portal phase, and the delay phase. Among them, the arterial phase refers to the arterial blood vessel filling development period, which can show the characteristics of the contrast medium flowing into the artery; the portal phase refers to the portal blood vessel filling development period, which can show the characteristics of the blood flowing through the portal vein after the arterial phase; the delay phase shows the characteristics after the arterial phase and the portal phase. The arterial phase time range after injecting the contrast medium is 20-25 seconds, the portal phase time range is 60-70 seconds, and the delay phase time range is 110-130 seconds.

[0038] S120: Segment the pancreas, stones and pancreatic duct in the portal period image to obtain a mask.

[0039] The image segmentation model 210 can be used to segment the portal period image outlined in S120 to obtain a segmentation mask. The mask is the segmented pancreas, stone and pancreatic duct mask.

[0040] In some embodiments, the image segmentation model 210 can be trained based on any one of 3D-Unet, Segnet, MaskRCNN and DeepLabv3+.

[0041] S130: Train the image segmentation model 210 using the mask and the first plain scan period image to obtain the trained image segmentation model 210.

[0042] Because some structures in the CT plain scan phase image are not clear, it is difficult to distinguish, therefore, the pancreas, stone and pancreatic duct are first outlined on the portal phase image, the pancreas is divided into head, body and tail when outlining, a mask is generated based on the annotation of the outline, the plain scan phase image and the portal phase image are registered, the mask outlined on the portal phase image is migrated to the plain scan phase image, and the image segmentation model 210 is trained based on the mask corresponding to the plain scan phase image.

[0043] As an implementation, the first plain scan phase image and the portal phase image are registered, the portal phase image and the first plain scan phase image of the CT image of the pancreas of the chronic pancreatitis patient are image-registered, the registration principle is to take the portal phase image as the reference, the pancreas, stone and pancreatic duct in the portal phase image are segmented to obtain a mask, and the segmentation mask of the pancreas, stone and pancreatic duct outlined on the portal phase image is migrated to the first plain scan phase image. The mask corresponding to the first plain scan phase image is generated based on the mask.

[0044] The chronic pancreatitis patient in S130 refers to a chronic pancreatitis patient diagnosed by clinic. In addition, in the above implementation, the first plain scan phase image and the portal phase image are registered, the registration principle is to take the portal phase image as the reference, and in another implementation, the plain scan phase image, the arterial phase image, the portal phase image and the delay phase image can be calibrated based on the portal phase image.

[0045] Because the first plain scan phase image has been registered with the portal phase image, the mask corresponding to the first plain scan phase image can be generated according to the annotation of the segmented mask. The image segmentation model is trained according to the first plain scan phase image and the pancreas, stone and pancreatic duct mask therein, to obtain the trained image segmentation model. In this way, the trained image segmentation model is a pancreas, stone and pancreatic duct segmentation model based on the plain scan phase image.

[0046] In some implementations, the image segmentation model 210 is obtained based on the 3D-Unet algorithm. The Unet network mainly includes two parts, the first half is feature extraction, and the second half is used for up-sampling. This structure can also be called an encoder-decoder structure. Since the overall structure of the Unet network is a capital English letter U, it is called Unet. Since medical images are often block-shaped, that is, composed of many slices. In this case, if a 2D image processing model 2D-Unet is used for processing, the medical image pictures must be processed into individual slices, and then sent into the model for training, so the process is relatively cumbersome and low in efficiency.

[0047] 3DUNet is created based on 2D-UNet, and also contains an encoder part and a decoder part. The encoder part is used to analyze the whole picture and perform feature extraction and analysis, and the corresponding decoder part is to generate a segmented block diagram. Compared with 2D-UNet, the difference is that all 2D operations are replaced by 3D operations. In this way, instead of inputting a single slice, the whole picture can be taken as input to the model.

[0048] In some embodiments of the present application, the chronic pancreatitis patients include chronic pancreatitis patients and recurrent acute pancreatitis patients, wherein the acute pancreatitis patients are considered as mild suspected chronic pancreatitis patients, and here, no distinction is made, and they are collectively referred to as chronic pancreatitis patients. Therefore, the image segmentation model 210 is trained using the mask and the plain scan images of the pancreas CT images of the chronic pancreatitis patients, and the trained image segmentation model 210 further includes obtaining the plain scan images of the pancreas CT images of the chronic pancreatitis patients, training the image segmentation model 210 according to the mask and the plain scan images of the pancreas CT images of the chronic pancreatitis patients, that is, using the mask and the plain scan images of the pancreas CT images of the chronic pancreatitis patients as a training set, training the image segmentation model 210 through the training set, and additionally finding a batch of the plain scan images of the pancreas CT images of the chronic pancreatitis patients as a validation set, and verifying the image segmentation model 210 by the validation set.

[0049] As an example, referring to the data characteristics of the sample patients in Table 1, the sample data of the training set is 841 pancreatitis patients, and the sample data of the validation set is 84 pancreatitis patients. The pancreatitis patients include chronic pancreatitis patients and recurrent acute pancreatitis patients, wherein 841 is used as a training set for establishing a chronic pancreatitis clinical image prognosis scoring system, and 84 is used as a validation set for the model to verify the availability of the chronic pancreatitis clinical image prognosis scoring system.

[0050] Table 1: Reference characteristics of all patients

[0051] S130: Obtain the second plain scan image of the patient to be detected, and input the second plain scan image to the trained image segmentation model 210.

[0052] In some embodiments, the second plain scan image of the patient to be detected is input to the trained image segmentation model 210 to obtain a mask corresponding to the pancreas. For the patient to be detected, only the plain scan image needs to be obtained, and the enhanced CT image does not need to be used. After inputting the plain scan image of the patient to be detected to the trained image segmentation model 210, the parameters of the plurality of predetermined features of the trained image segmentation model 210 are obtained.

[0053] S150: Based on the trained image segmentation model 210, the second plain scan period image is processed to obtain parameters of a plurality of predetermined features. Specifically, based on the trained image segmentation model 210, a calcification segmentation mask of the second plain scan period image is obtained, and the calcification mask is three-dimensionally segmented using multi-dimensional data analysis of a Python software package to obtain calcification points. The number of calcifications and other information can be obtained, such as the location information of the calcifications determined according to the intersection-over-union of the segmentation mask of the calcification part and the head, body, and tail of the pancreas.

[0054] In some embodiments, the second plain scan period image can be classified based on a convolutional neural network. For example, the classification model can be any one of a VGGNet, an Inception, a ResNet, or a Densenet model. The VGGNet and the Inception are conventional convolutional neural network (CNN) models. In theory, the deeper the neural network, the better the effect of the classification model. However, a neural network that is too deep can cause problems such as gradient vanishing, gradient explosion, and model degradation, thereby reducing accuracy. To this end, ResNet can effectively solve these problems by using a residual network module. ResNet can train a deeper CNN model, thereby achieving higher accuracy. The core of the ResNet model is to establish a "short circuit connection" between the front and back layers, which helps the gradient to be back-propagated during the training process, thereby enabling a deeper CNN network to be trained. The Densenet model is similar to the ResNet, but the Densenet model establishes a dense connection between all previous layers and subsequent layers. In addition, the Densenet can achieve feature reuse through the connection of features in the channel. These characteristics allow the Densenet to reduce the parameter and computational cost.

[0055] In some embodiments, any one of DenseNet-121, DenseNet-161, DenseNet-169, and DenseNet-201 can be used as the image classification model 220.

[0056] The CT image processing method obtains a mask corresponding to the pancreas based on the registered plain scan period CT image and portal phase CT image of a patient with chronic pancreatitis, and trains an image segmentation model 210 based on the plain scan period image by using the plain scan period image and the mask through transfer learning. The trained image segmentation model 210 is used to obtain a mask of the plain scan period image of a patient to be detected, and a classification model is used to classify the plain scan period image of the patient to be detected based on the mask. In this way, when making a prognosis diagnosis of chronic pancreatitis, the patient's enhanced CT image of the pancreas does not need to be obtained, only the patient's plain CT image needs to be obtained, which has the advantages of being non-invasive, having no risk of iodine allergy, being inexpensive, and being friendly to people with poor liver and kidney function.

[0057] As an implementation, in combination with the sample data of chronic pancreatitis patients in Table 1 above, the second plain scan period image is processed based on the trained image segmentation model 210 to obtain parameters of a plurality of predetermined features, and referring to Tables 2-3, it can be seen that the parameters of the plurality of predetermined features.

[0058] Table 2 Quantitative feature table of CT image

[0059] Table 3 Semi-quantitative quantitative feature table of CT image

[0060] In some embodiments, classifying the second plain scan period image based on the output of the trained image segmentation model 210 includes extracting features layer by layer based on the output of the trained image segmentation model, and then classifying the second plain scan period image based on the extracted features using the image classification model 220, and the parameters of the plurality of predetermined features are obtained according to the output of the image segmentation model 210. As an implementation, the average annual hospitalization days of chronic pancreatitis patients are used as the outcome indicator, and linear correlation analysis (Spearman) is used to test the correlation between the 32 indicators and the average annual hospitalization days. It is found that the total pancreatic stone volume, the maximum stone diameter, the average stone diameter, the average stone CT value, the intrapancreatic duct stone volume, the pancreatic involvement and range, and the maximum pancreatic thickness are correlated with the average annual hospitalization days. The above 7 indicators are combined with endocrine function indicators such as diabetes or not, and exocrine indicators such as intermittent diarrhea and fatty diarrhea. After pancreatic enzyme replacement therapy, the symptoms are improved, which indicates that the exocrine dysfunction is improved, and the CP-CRPS system 200 is formed, as shown in FIG. 2.

[0061] FIG. 2 is a schematic diagram of an image classification model according to an embodiment of the present application.

[0062] The chronic pancreatitis clinical image prognosis scoring system 200 includes an image segmentation model 210 and an image classification model 220. The image segmentation model 210 can receive a plain scan period image of a patient to be detected and output a pancreatic mask. The image classification model 220 classifies the plain scan period image according to the pancreatic mask. The image segmentation model 210 is trained by the following method.

[0063] First, the first plain scan period image and the portal phase image of the pancreatic CT image of the chronic pancreatitis patient are obtained.

[0064] Second, the pancreas, stones and pancreatic duct in the portal phase image are segmented to obtain a mask.

[0065] Again, the image segmentation model 210 is trained using the mask and the first plain scan image.

[0066] Based on the parameters of the plurality of predetermined features output by the image segmentation model 210, under the assistance of the image classification model 200, the doctor can combine the segmentation and classification results with the clinical and CT features, and give the final prognosis result.

[0067] The image classification model 220 classifies the plain scan image according to the pancreatic mask, and the image classification model 220 is trained by the following method.

[0068] First, the pancreatic plain scan image of the patient with chronic pancreatitis and the endocrine and exocrine indicators of the patient with chronic pancreatitis are obtained;

[0069] Second, the parameters of a plurality of predetermined features are obtained using the processing method of the pancreatic CT image, and the image classification model is trained using the parameters of the plurality of predetermined features, the endocrine and exocrine indicators as samples, and classified based on the image classification model.

[0070] Based on the CP-CRPS system 200, as an embodiment, referring to Table 4, combined with the sample data of Tables 1-3 described above, the result of the chronic pancreatitis prognosis score can be obtained. In order to generate the chronic pancreatitis prognosis score table 4, Spearman analysis needs to be performed first to determine the parameters significantly correlated with pALOS, and then the CP-CRPS scoring system is established using the related parameters. In order to simplify the scoring system, the CT features significantly correlated with pALOS are divided into four categories: none, low, medium, and high. In addition to the maximum thickness of the pancreas, each category has a score. Specifically, 0 points for none, 1 point for low, 2 points for medium, and 3 points for high. As an example, the maximum thickness of the pancreas is ≥30mm, 0 points, the maximum thickness of the pancreas is in the range of 20-30mm, 1 point, the maximum thickness of the pancreas is ≤20mm, 2 points, in addition, there is no diabetes and exocrine insufficiency, 0 points, and there is 1 point. The involvement of the pancreas is evaluated as 0, 1, 2, and 3 points according to normal, 30%, 30%-70%, and 70%, respectively. Finally, the scoring rules are written into the artificial intelligence system. The critical value of the distribution is determined according to the reporting standard of CP, and the rest of the features are divided into four categories with an average of four, rounded to an integer.

[0071] Table 4 Chronic Pancreatitis Clinical Imaging Prognosis Scoring Table

[0072] According to the verification results of Tables 1-4, there are a total of 925 patients with recurrent acute pancreatitis and chronic pancreatitis, with a median age of 44 years, a total range of 5.7-81.2 years, and 667 males. All patients are divided into a training set and a verification set, with 841 in the training set and 84 in the verification set. In the training set and the verification set, the median of pALOS is 9.0 days, with an actual range of 1.0-36, and the median is 11.0 days, with an actual range of 2.0-34. The chronic pancreatitis clinical imaging prognosis score (CP-CRPS) system is formulated in combination with 7 CT quantitative parameters of endocrine disease and exocrine dysfunction. In the training set and the verification set, the median of CP-CRPS is 15.0, with a range of 6.0-22.0, and the median is 17.0, with a range of 6.0-21.0. In the training set, CP-CRPS is correlated with pALOS, with a Spearman correlation coefficient of 0.33, P<0.00001, and the correlation is externally verified, with a Spearman correlation coefficient of 0.35, P=0.001.

[0073] Figure 2 provides an end-to-end image classification system, which inputs the plain scan image of the patient to be detected into the image segmentation model, and inputs the output of the image segmentation model into the image classification model, without obtaining the enhanced CT image of the patient's pancreas. Only the CT plain scan image of the patient is needed, and the existing enhanced CT image of the confirmed patient is used in combination with deep learning to realize the prognosis diagnosis of pancreatitis based on non-enhanced CT images, instead of using the enhanced CT method, which has the advantages of non-invasive, no risk of iodine allergy, low cost, and friendly to people with poor liver and kidney function.

[0074] As an implementation, the correlation between the indicators is analyzed by using statistical methods, the 3D-U-Net is trained by using non-enhanced CT for segmentation model, the quantitative features of CT images and functional indicators are developed, for specific multiple characteristic parameters and indicators, the correlation between variables and outcomes is analyzed by using Spearman. The baseline data of continuous variables is expressed by mean ± standard deviation (SD) or median (range), and the baseline data of classification variables is expressed by proportion. Cases with missing clinical and imaging data are excluded. In order to determine the single quantitative features of CT images, linear correlation analysis is carried out, and parameters with significant correlation with pALOS are selected for further analysis. In order to establish a convenient scoring system, the CT features with significant correlation are divided into low, medium and high three segments, and linear correlation is used again between all chronic pancreatitis classification systems and outcomes, and variance analysis is used for comparison between all chronic pancreatitis classification systems and outcomes. Finally, 7 parameters are found, including performance distribution, total stone volume, total stone maximum diameter, total stone average diameter, average CT value of all stones, main pancreatic duct stone volume and pancreatic maximum diameter, which have significant correlation with pALOS. The 7 parameters combine with endocrine index and exocrine index to establish a classification system.

[0075] It is worth noting that the above statistical methods are aimed at analyzing the correlation between indicators, and should not be regarded as a limitation of the present application. Other analysis methods available to those skilled in the art can be used for analysis.

[0076] In clinical practice, accurate evaluation of the severity of pancreatitis helps to accurately predict the short-term and medium-term prognosis. Imaging is a necessary condition for the diagnosis of pancreatitis, but its application has been hindered due to the lack of accurate quantification. The scheme of the present application accurately quantifies the non-enhanced CT features, especially the 32 CT features of pancreatic stones, parenchyma and ducts. In addition, 5 of the 7 CT features are related to stones, which are related to the course and complexity of stone fragmentation, and the final score also selects the distribution of pancreatic parenchyma, which can indicate the degree of pancreatic fibrosis and reflect the pathological features of pancreatitis, including interlobular and intralobular fibrosis, acinar cell loss, structural distortion and duct dilation. Considering the disturbance of endocrine and exocrine function, the endocrine index and the exocrine index are included. Finally, the CP-CRPS system combines 7 CT features with 2 functional indicators, which shows significant correlation with pALOS. It is worth noting that the system of the present application is output by artificial intelligence, which is easier to use in clinical practice than other classification systems of pancreatitis.

[0077] Fig. 3 is a hardware structure block diagram of an electronic device according to an embodiment of the present application.

[0078] As shown in FIG. 3, the electronic device 300 can include one or more processors 302, system control logic 308 connected to at least one of the processors 302, system memory 304 connected to the system control logic 308, non-volatile memory (NVM) 306 connected to the system control logic 308, and a network interface 310 connected to the system control logic 308.

[0079] The processor 302 can include one or more single-core or multi-core processors. The processor 302 can include any combination of general-purpose processors and dedicated processors (e.g., graphics processors, application processors, baseband processors, etc.). In embodiments of the present application, the processor 302 can be configured to perform a processing method according to a pancreatic CT image as shown in FIG. 1.

[0080] In some embodiments, the system control logic 308 can include any suitable interface controllers to provide for any suitable interface to at least one of the processors 302 and / or any suitable device or component in communication with the system control logic 308.

[0081] In some embodiments, the system control logic 308 can include one or more memory controllers to provide an interface to connect to the system memory 304. The system memory 304 can be used to load and store data and / or instructions. In some embodiments, the system memory 304 of the electronic device 300 can include any suitable volatile memory, such as suitable dynamic random access memory (DRAM).

[0082] The non-volatile memory 306 can include one or more tangible, non-transitory computer-readable media for storage of data and / or instructions. In some embodiments, the non-volatile memory 306 can include any suitable non-volatile memory, such as flash memory, and / or any suitable non-volatile storage device, such as at least one of a HDD (Hard Disk Drive), a CD (Compact Disc) drive, a DVD (Digital Versatile Disc) drive.

[0083] The non-volatile memory 306 can include a portion of the storage resources installed on the device of the electronic device 300, or it can be accessed by the device but not necessarily part of the device. For example, the non-volatile memory 306 can be accessed over a network via the network interface 310.

[0084] In particular, system memory 304 and non-volatile memory 306 can include, respectively, a temporary copy and a permanent copy of instructions 320. Instructions 320 can include instructions that, when executed by at least one of processors 302, cause electronic device 300 to implement a method of processing a pancreatic CT image as shown in FIG. 1. In some embodiments, instructions 320, hardware, firmware, and / or software components thereof can additionally / alternatively be located in system control logic 308, network interface 310, and / or processors 302.

[0085] Network interface 310 can include a transceiver to provide a radio interface for electronic device 300 to communicate with any other suitable device (e.g., a front-end module, an antenna, etc.) over one or more networks. In some embodiments, network interface 310 can be integrated with other components of electronic device 300. For example, network interface 310 can be integrated with at least one of processors 302, system memory 304, NVM 306, and a firmware device (not shown) having instructions that, when executed by at least one of processors 302, cause electronic device 300 to implement one or more embodiments of various embodiments shown in FIG. 1.

[0086] Network interface 310 can further include any suitable hardware and / or firmware to provide a multiple-input multiple-output radio interface. For example, network interface 310 can be a network adapter, a wireless network adapter, a telephone modem, and / or a wireless modem.

[0087] In one embodiment, at least one of processors 302 can be packaged with one or more controllers for system control logic 308 to form a system-in-a-package (SiP). In one embodiment, at least one of processors 302 can be integrated on the same die with one or more controllers for system control logic 308 to form a system-on-a-chip (SoC).

[0088] Electronic device 300 can further include input / output (I / O) devices 312 coupled with system control logic 308. I / O devices 312 can include a user interface to enable a user to interact with electronic device 300; a peripheral component interface to enable peripheral components to interact with electronic device 300. In some embodiments, electronic device 300 also includes sensors to determine at least one of environmental conditions and location information related to electronic device 300.

[0089] In some embodiments, input / output (I / O) devices 312 can include, but are not limited to, a display (e.g., a liquid crystal display, a touch screen display, etc.), a speaker, a microphone, one or more cameras (e.g., a still image camera and / or a video camera), a flashlight (e.g., a light-emitting diode flash), and a keypad.

[0090] In some embodiments, the peripheral component interface can include, but is not limited to, a non-volatile memory port, an audio jack, and a power supply interface.

[0091] It can be understood that the structural schematic of the embodiments of the present application does not constitute a specific limitation to the electronic device 300. In other embodiments of the present application, the electronic device 300 can include more or fewer components than shown, or a combination of some components, or splitting of some components, or different arrangement of components. The components shown can be implemented in hardware, software, or a combination of software and hardware.

[0092] The program code can be applied to input instructions to perform the functions described in the present application and to generate output information. The output information can be applied to one or more output devices in a known manner. For the purpose of the present application, the system for processing instructions including the processor 302 includes any system having a processor such as a digital signal processor (DSP), a microcontroller, an application specific integrated circuit (ASIC) or a microprocessor.

[0093] The program code can be implemented in a high-level programming language or an object-oriented programming language to communicate with the processing system. When necessary, the program code can also be implemented in assembly language or machine language. In fact, the mechanisms described in the present application are not limited to the scope of any specific programming language. In any case, the language can be a compiled language or an interpreted language.

[0094] According to one embodiment of the present application, a computer readable storage medium is also proposed, wherein at least one computer instruction is stored in the computer readable storage medium, and the at least one instruction is loaded and executed by a processor to implement the aforementioned processing method of the pancreatic CT image.

[0095] According to one embodiment of the present application, a computer program product is also proposed, wherein the computer program product includes computer instructions, and the computer instructions are executed to implement the aforementioned processing method of the pancreatic CT image.

[0096] The illustrative embodiments of the present application include, but are not limited to, a processing method, a classification method, a device, a medium and a program product of the pancreatic CT image.

[0097] Various aspects of the illustrative embodiments will be described using terminology commonly employed by those skilled in the art and having a basic understanding of the technology. It will be apparent, however, to one skilled in the art that the features described in the specification can be implemented in a variety of alternate embodiments. For purposes of explanation and ease of understanding, specific details of the illustrative embodiments are set forth including a particular sequence of steps implemented in one or more processes. However, it will be apparent to those skilled in the art that the illustrative embodiments can be practiced without the specific details. In some instances, detailed descriptions of well-known methods associated with computing, software, programming, and / or engineering are omitted so as not to obscure the illustrative embodiments.

[0098] Furthermore, various operations will be described as multiple discrete operations, in a manner that is most helpful in understanding the illustrative embodiments; however, the order of description should not be construed as to imply that these operations are necessarily order dependent. Many of the operations can be handled in parallel or concurrently. Additionally, the description neither implies that the various blocks are the only ones that can be employed, nor that all of the blocks must be employed. Skilled persons having the benefit of this disclosure will appreciate that other operations can be implicit to the described operations and are omitted for the sake of brevity.

[0099] Reference throughout this specification to "an example", "one example", "exemplary embodiment", "one embodiment", or the like, means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. The appearances of the phrases in various places in the specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics can be combined in any suitable manner on the various embodiments without limitation.

[0100] The terms "comprise", "have" and "include" are synonymous and are used in their open-ended sense. The phrase "A and / or B" means "(A), (B), or (A and B)".

[0101] As used herein, the term "module" can refer to, be part of, or include: a memory (shared, dedicated, or group) that stores one or more software or firmware programs, an application specific integrated circuit (ASIC), an electronic circuit, and / or a processor (shared, dedicated, or group) that executes the software or firmware programs to perform the functions described herein, a combination of more than one of the above, and / or other suitable components.

[0102] In the drawings, some of the structures or method features can be shown in particular arrangements and / or orders. However, it will be appreciated that such specific arrangements and / or orders are not required. Rather, in some embodiments, the features can be arranged and / or ordered differently than shown in the illustrative drawings. Moreover, the inclusion of a structural or method feature in a particular drawing is not meant to imply that such a feature is required in all embodiments. In some embodiments, the features can be included or excluded from such a feature in combination with other features.

[0103] In the description provided herein, numerous specific details are set forth. However, it is understood that embodiments of the application can be practiced without these specific details. In some instances, well-known methods, structures and techniques have not been described in detail in order to not obscure the understanding of this description.

[0104] Similarly, it is to be understood that the mechanical features of the application sometimes are grouped together in a single embodiment, figure or description of related features, for the purpose of streamlining the disclosure and aiding in the understanding of one or more aspects of the application. In this respect, no implication can be inferred that the described examples of the application require a greater number of features than are explicitly recited in each claim. To the contrary, structural and / or logical features are often listed in single embodiments, figures or descriptions when it is practical to do so and not intended to imply that each available feature must be used in any single embodiment. Rather, logical and / or structural features can be varied freely among embodiments, to form desired structural and logical combinations. Therefore, the following claims are hereby expressly incorporated into this detailed description, with each claim acting as a separate embodiment of the application.

[0105] Those skilled in the art will appreciate that modules in the apparatuses of the embodiments can be adapted and placed in one or more apparatuses other than the embodiments. Modules or units or components in the embodiments can be combined into one module or unit or component and further can be divided into more modules or units or components. Any combination of all or some of the disclosed features and any method or device of the embodiments disclosed in the specification (including accompanying claims, abstract and drawings), or any method or device so disclosed, can be adopted in any combination, except that at least some of such features and / or processes or units are mutually exclusive. Unless explicitly stated otherwise, each feature disclosed in the specification (including accompanying claims, abstract and drawings) can be replaced by alternative features that serve the same, equivalent or similar purpose.

[0106] Furthermore, those skilled in the art will recognize that, while certain embodiments described herein include certain features that are not included in other embodiments, combinations of features of different embodiments are meant to be within the scope of the application and form different embodiments, for example, in the claims. For example, in the claims, any of the claimed embodiments can be used in any combination.

Claims

1. A method for processing pancreatic CT images, used in electronic equipment, characterized in that: The method comprises: Acquiring the first plain scan phase image and portal venous phase image of pancreatic CT images of patients with chronic pancreatitis; segmenting the pancreas, stones, and pancreatic duct in the portal venous phase image to obtain a mask; training an image segmentation model using the mask and the first plain scan image to obtain a trained image segmentation model; Acquire a second plain scan image of the patient to be detected, and input the second plain scan image into the trained image segmentation model; The second plain scan image is processed based on the trained image segmentation model to obtain parameters of multiple predetermined features.

2. The method for processing pancreatic CT images according to claim 1, characterized in that: Training the image segmentation model using the mask and the first plain scan image also includes registering the first plain scan image and the portal venous phase image, and generating a mask corresponding to the first plain scan image based on the mask.

3. The method for processing pancreatic CT images according to claim 2, characterized in that: The image segmentation model is trained using a 3D-UNET algorithm based on the mask corresponding to the first plain scan image.

4. A method for classifying pancreatic CT images, used in electronic equipment, characterized in that: The method comprises: Obtaining a plain scan image of the pancreas of a patient with chronic pancreatitis and endocrine and exocrine indicators of the patient; Classification based on image classification model; The image classification model is obtained by the following training method: Obtaining parameters of a plurality of predetermined features using the method according to claim 1, The image classification model is trained using the parameters of the plurality of predetermined features, the endocrine index, and the exocrine index as samples.

5. An electronic device, characterized in that: The device includes a processor and a memory storing computer-executable instructions, wherein the processor is configured to execute the instructions to implement the pancreatic CT image processing method according to any one of claims 1 to 3 or the pancreatic CT image classification method according to claim 4.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one computer instruction, which is loaded and executed by a processor to implement the pancreatic CT image processing method according to any one of claims 1 to 3 or the pancreatic CT image classification method according to claim 4.

7. A computer program product, characterized in that The computer program product includes computer instructions, and when the computer instructions are executed, the pancreatic CT image processing method according to any one of claims 1 to 3 or the pancreatic CT image classification method according to claim 4 is implemented.

Images