Prostate cancer and hyperplasia prediction system based on multiple modes and construction method

Abstract

The invention provides a multi-mode-based prostate cancer and hyperplasia prediction system and a construction method, relates to the technical field of medical image processing, effectively reduces errors possibly introduced by automatic division through precise combination of preliminary division and secondary identification, improves the accuracy and reliability of prostate region identification, and improves the prostate cancer and hyperplasia prediction efficiency. The time of the whole prediction process can be effectively shortened by adopting a parallel computing processing mode, the processing efficiency is improved, the accuracy of feature recognition can be further improved by matching with fusion and screening of multi-modal features, and a good guarantee is provided for high accuracy and high efficiency of prediction results of prostate cancer conditions and bone metastasis conditions of the prostate cancer conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a multi-modality-based prostate cancer and hyperplasia prediction system and a construction method. Background Art

[0002] In the field of medical imaging, early and accurate diagnosis of prostate cancer and prostate hyperplasia is crucial for the patient's treatment options and prognosis. In addition to occurring within the prostate capsule and causing local symptoms, prostate cancer can also metastasize to other parts of the body, with bones being a common site of metastasis. Bone metastasis is a relatively common and serious condition in prostate cancer, so the prediction of prostate cancer and its bone metastasis is more important. However, in current clinical practice, the prediction of prostate cancer and hyperplasia faces many challenges, such as:

[0003] Traditional automatic segmentation technologies, such as image-based segmentation algorithms, can assist doctors in segmenting prostate areas to a certain extent, but they often have a high error rate. These errors may be caused by unstable image quality, the diversity of prostate morphology, and the complexity of the lesion area. Therefore, the results of automatic segmentation are not accurate enough, which leads to poor accuracy of extracted features, which easily affects the accuracy of actual prediction results. While pursuing high processing efficiency, how to ensure the accuracy of prediction results is a major problem facing current technologies. Traditional processing methods often need to compromise between processing efficiency and accuracy, and it is difficult to meet the needs of clinical applications at the same time.

[0004] Therefore, it is necessary to provide a multimodal prostate cancer and hyperplasia prediction system and construction method to solve the above technical problems. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a multimodal prostate cancer and hyperplasia prediction system and construction method for solving the problem that the existing automatic segmentation results are not accurate enough, resulting in poor accuracy of extracted features, which easily affects the accuracy of actual prediction results and poor processing efficiency.

[0006] The present invention provides a multimodal prostate cancer and hyperplasia prediction system, comprising:

[0007] A region division module is used to preliminarily divide the prostate region according to the position distribution of each organ in each collected inspection image and the shape information of the prostate;

[0008] A model building module is used to extract the correlation between the patient's age, body shape and prostate condition based on historical patient prostate condition data to build a prostate recognition model;

[0009] The identification module is used to identify each of the collected inspection images respectively through the prostate identification model, and to preliminarily identify the preliminarily divided prostate area to obtain the preliminarily identified prostate area;

[0010] An interactive module, for using an interactive operation mode to allow a doctor or an expert to perform a secondary marking on the initially marked prostate region to obtain a secondary marked prostate region;

[0011] A parallel processing module, used to use a plurality of preset processing nodes to perform parallel processing on the prostate region marked twice in each inspection image, and extract features of the prostate region marked twice;

[0012] An evaluation module, used to evaluate the secondary marked prostate region features respectively through processing nodes to obtain evaluation values ​​of each feature;

[0013] A fusion processing module is used to fuse each feature based on the evaluation value of each feature through a preset central processing node and an evaluation value threshold to form a multi-modal feature vector;

[0014] The prediction module is used to input the multimodal feature vector into the trained deep learning model and output the prediction result.

[0015] A method for constructing a prediction of prostate cancer and hyperplasia based on multimodality, the method comprising the following steps:

[0016] The prostate region is preliminarily divided according to the position distribution of each organ in each inspection image collected and the shape information of the prostate;

[0017] Based on the historical patient prostate disease data, the correlation between the patient's age, body shape and prostate disease is extracted to build a prostate recognition model;

[0018] Using the prostate recognition model, each of the collected inspection images is recognized, and the initially divided prostate area is preliminarily marked to obtain a preliminarily marked prostate area;

[0019] Using the interactive operation mode, doctors or experts conduct secondary marking of the initially marked prostate area to obtain the secondary marked prostate area;

[0020] Using a plurality of preset processing nodes to process the secondary marked prostate region in each inspection image in parallel, and extracting the secondary marked prostate region features;

[0021] The features of the prostate region marked by the secondary identification are evaluated respectively through the processing nodes to obtain the evaluation value of each feature;

[0022] Based on the evaluation value of each feature, each feature is fused through a preset central processing node and evaluation value threshold to form a multimodal feature vector;

[0023] Input the multimodal feature vector into the trained deep learning model and output the prediction result.

[0024] Preferably, the preliminary division of the prostate region according to the position distribution of each organ in each collected inspection image and the shape information of the prostate comprises the following steps:

[0025] Collecting various examination images of the patient, including but not limited to MRI images and ultrasound images;

[0026] Use image segmentation technology to identify the location of each organ in each examination image and determine the location of organs adjacent to the prostate;

[0027] Based on the standard morphological features of the prostate, the prostate contour is outlined in each examination image and the prostate area is preliminarily divided.

[0028] Preferably, the extracting the correlation between the patient's age, body shape and prostate condition based on historical patient prostate condition data to construct a prostate recognition model specifically includes the following steps:

[0029] Obtain medical records of historical patients, including data sets of patient age, body shape, and prostate condition information;

[0030] Analyze the correlation between age, body shape and prostate disease through statistical methods, identify the key factors that affect prostate disease, and obtain the analysis results of the correlation;

[0031] The analysis results of the association relationship were used as the training set, and a machine learning algorithm was used to build and train a prostate recognition model.

[0032] Preferably, the method of identifying each collected inspection image by using the prostate recognition model, and preliminarily marking the preliminarily divided prostate area, and obtaining the preliminarily marked prostate area specifically comprises the following steps:

[0033] Inputting each collected inspection image into the prostate recognition model, the prostate recognition model automatically recognizes and marks the prostate area according to each input inspection image;

[0034] The prostate region is automatically identified and marked and output in a visual form to obtain a preliminary identification of the prostate region.

[0035] Preferably, the interactive operation mode is used to perform secondary marking of the initially marked prostate region by a doctor or an expert, and obtaining the secondary marked prostate region specifically comprises the following steps:

[0036] The initially marked prostate area is shown to a doctor or expert, and the doctor or expert performs secondary marking on the initially marked prostate area based on professional knowledge;

[0037] After confirming that the secondary identification result is correct, save and output it as the secondary identified prostate area.

[0038] Preferably, the method of using a plurality of preset processing nodes to perform parallel processing on the prostate regions marked twice in each inspection image and extracting features of the prostate regions marked twice specifically comprises the following steps:

[0039] The prostate region marked by the secondary mark of each inspection image is processed in parallel by using a plurality of preset processing nodes, specifically including denoising, contrast enhancement, normalization and smoothing filtering, to obtain processed inspection image data;

[0040] Based on the processed inspection image data, the prostate region features are extracted, including shape features, texture features and statistical features based on the inspection image data.

[0041] Preferably, the process of evaluating the secondary identified prostate region features respectively through the processing nodes to obtain the evaluation values ​​of each feature specifically comprises the following steps:

[0042] The prostate region features extracted from each processing node are standardized;

[0043] The quality of each extracted prostate region feature was evaluated by calculating feature statistics and analyzing the correlation between features.

[0044] According to the preset evaluation index and the results of quality assessment, the contribution of each extracted prostate region feature is calculated and converted into an evaluation value.

[0045] Preferably, the step of fusing each feature based on the evaluation value of each feature through a preset central processing node and an evaluation value threshold to form a multimodal feature vector specifically includes the following steps:

[0046] Based on the evaluation value of each prostate region feature, the preset threshold is used to screen out features that contribute most to the prediction task;

[0047] The filtered features are fused to form a multimodal feature vector, and the fused multimodal feature vector is overall standardized to obtain a standardized multimodal feature vector.

[0048] Preferably, the step of inputting the multimodal feature vector into a trained deep learning model and outputting a prediction result specifically comprises the following steps:

[0049] Load the deep learning model that has been pre-trained based on historical prostate disease imaging data and prostate bone metastasis data;

[0050] The standardized multimodal feature vector is input into the trained deep learning model to calculate and output the prediction results, which specifically include disease classification, disease severity assessment, and bone metastasis.

[0051] Compared with related technologies, the multimodal prostate cancer and hyperplasia prediction system and construction method provided by the present invention have the following beneficial effects:

[0052] The present invention effectively reduces the errors that may be introduced by automatic division through the precise combination of preliminary division and secondary identification, improves the accuracy and reliability of prostate region identification, and adopts parallel computing processing to effectively shorten the time of the entire prediction process and improve processing efficiency. In combination with the fusion and screening of multimodal features, the accuracy of recognition features can be further improved, providing a good guarantee for the high accuracy and efficiency of the prediction results of prostate cancer and its bone metastasis. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 A system block diagram of a multi-modality-based prostate cancer and hyperplasia prediction system provided by the present invention;

[0054] Figure 2 A schematic diagram of a process for constructing a multimodal prostate cancer and hyperplasia prediction method provided by the present invention;

[0055] Figure 3 A block diagram of a computing device capable of implementing various embodiments of the present disclosure is shown. DETAILED DESCRIPTION

[0056] The present invention will be further described below in conjunction with the accompanying drawings and implementation modes.

[0057] Embodiment 1

[0058] like Figure 1 As shown, a multi-modality-based prostate cancer and hyperplasia prediction system 100 includes:

[0059] The region division module 101 is used to preliminarily divide the prostate region according to the position distribution of each organ in each collected inspection image and the shape information of the prostate;

[0060] Specifically, the region division module is used to collect prostate examination images from different medical imaging devices, such as MRI, CT, ultrasound, etc., and identify the boundaries of organs in the examination images, such as bladder, rectum, etc., to determine the relative position of the prostate, and use edge detection in image processing technology to extract the shape information of the prostate and preliminarily divide it into prostate regions.

[0061] The model building module 102 is used to extract the correlation between the patient's age, body shape and prostate condition based on historical patient prostate condition data to build a prostate recognition model;

[0062] Specifically, the model building module is used to use data mining or statistical analysis methods to explore the correlation between age, body shape and prostate disease, and based on the correlation, train or build a prostate recognition model.

[0063] The identification module 103 is used to identify each of the collected inspection images respectively through the prostate identification model, and to preliminarily identify the preliminarily divided prostate area to obtain a preliminarily identified prostate area;

[0064] Interaction module 104, used to use an interactive operation mode to allow doctors or experts to perform secondary marking on the initially marked prostate region to obtain a secondary marked prostate region;

[0065] Specifically, the interactive module is used to provide an interactive interface, allowing doctors or experts to view the results of the preliminary identification. The doctors or experts use interface tools to reconfirm or adjust the preliminary identification to obtain a more accurate prostate area identification.

[0066] A parallel processing module 105 is used to perform parallel processing on the prostate region marked twice in each inspection image using a plurality of preset processing nodes, and extract features of the prostate region marked twice;

[0067] Specifically, the parallel processing module is used to distribute the image to a plurality of preset processing nodes, and each node is responsible for a portion in each inspection image, and parallel processing is performed on the secondary marked prostate area in each inspection image to extract the secondary marked prostate area features.

[0068] An evaluation module 106, for evaluating the secondary marked prostate region features respectively through processing nodes to obtain evaluation values ​​of each feature;

[0069] A fusion processing module 107 is used to fuse each feature based on the evaluation value of each feature through a preset central processing node and an evaluation value threshold to form a multi-modal feature vector;

[0070] Specifically, the fusion processing module is used to receive the evaluation values ​​of all features through the central processing node, and fuse the features according to a preset fusion algorithm, specifically a feature selection algorithm, combined with the evaluation value threshold, to form a multimodal feature vector.

[0071] The prediction module 108 is used to input the multimodal feature vector into the trained deep learning model and output the prediction result.

[0072] Embodiment 2

[0073] like Figure 2 As shown, a multi-modality-based prostate cancer and hyperplasia prediction construction method 200 applied to a multi-modality-based prostate cancer and hyperplasia prediction system includes the following steps:

[0074] In block 201, the prostate region is preliminarily divided according to the position distribution of each organ in each collected inspection image and the shape information of the prostate;

[0075] In the specific implementation process, the specific steps of step 201 are:

[0076] Collecting various examination images of the patient, including but not limited to MRI images and ultrasound images;

[0077] Use image segmentation technology to identify the location of each organ in each examination image and determine the location of organs adjacent to the prostate;

[0078] Specifically, a Canny edge detector is used to segment the position area of ​​each organ in each inspection image, and contour smoothing processing is performed on the segmented position area.

[0079] Based on the standard morphological features of the prostate, the prostate contour is outlined in each examination image, and the prostate area is preliminarily divided;

[0080] Specifically, based on the anatomical characteristics of the prostate and clinical knowledge, the standard morphological characteristics of the prostate are determined, including the size range, shape (usually chestnut-shaped), and location (located below the bladder and in front of the rectum). Existing image processing software is used to automatically outline the contours of the prostate in each examination image and preliminarily divide the prostate area.

[0081] In block 202, based on historical patient prostate condition data, the correlation between the patient's age, body shape and prostate condition is extracted to construct a prostate recognition model;

[0082] In the specific implementation process, the specific steps of block 202 are:

[0083] Obtain medical records of historical patients, including data sets of patient age, body shape, and prostate condition information;

[0084] Specifically, the patient's medical record data is extracted from the historical medical record system. The medical record data covers multiple dimensions, including but not limited to the patient's age, body shape (specifically height, weight, BMI, etc.) and prostate condition information. The collected medical record data is cleaned to remove duplicate items, records with too many missing values, and obviously erroneous data, and the cleaned data is organized into a unified format for subsequent analysis.

[0085] Analyze the correlation between age, body shape and prostate disease through statistical methods, identify the key factors that affect prostate disease, and obtain the analysis results of the correlation;

[0086] Specifically, the Pearson correlation coefficient or Spearman rank correlation coefficient in statistical methods is used to analyze the correlation between age, body shape and prostate condition.

[0087] The analysis results of the association relationship were used as the training set, and a machine learning algorithm was used to build and train a prostate recognition model;

[0088] Specifically, the convolutional neural network (CNN) in the machine learning algorithm is used as the architecture, and the analysis results of the association relationship are used as the training set to train and obtain the final prostate recognition model.

[0089] In block 203, the prostate recognition model is used to recognize each of the collected inspection images, and the initially divided prostate regions are initially marked to obtain the initially marked prostate regions;

[0090] In the specific implementation process, the specific steps of block 203 are:

[0091] Inputting each collected inspection image into the prostate recognition model, the prostate recognition model automatically recognizes and marks the prostate area according to each input inspection image;

[0092] Specifically, the collected inspection images are input into the prostate recognition model, and the prostate recognition model automatically extracts feature information in the inspection images through structures such as convolution layers and pooling layers. The classifier or regressor in the prostate recognition model is used to identify the prostate area based on the extracted feature information and give the corresponding coordinate information.

[0093] The prostate region is automatically identified and marked and output in a visual form to obtain a preliminary marked prostate region;

[0094] Specifically, different colors or transparencies are used to mark the prostate region so as to distinguish it from surrounding tissues, and the marked image is displayed in a visual form so as to directly observe the position and morphology of the prostate region.

[0095] In block 204, the doctor or expert performs secondary marking on the initially marked prostate region by using an interactive operation mode to obtain a secondary marked prostate region;

[0096] In the specific implementation process, the specific steps of block 204 are:

[0097] The initially marked prostate area is shown to a doctor or expert, and the doctor or expert performs secondary marking on the initially marked prostate area based on professional knowledge;

[0098] Specifically, through the existing medical image processing interactive software, the image of the prostate area preliminarily marked is displayed to the doctor or expert in a visual form. The doctor or expert uses his or her professional knowledge and experience to review the preliminarily marked prostate area. If the review finds that the preliminary marking is inaccurate or missing, the doctor or expert will use the tools provided by the software, such as adjusting the boundary line, to perform a second marking of the prostate area for correction.

[0099] After confirming that the secondary identification result is correct, save and output it as the secondary identified prostate area.

[0100] In block 205, a plurality of preset processing nodes are used to perform parallel processing on the prostate region marked twice in each inspection image, and features of the prostate region marked twice are extracted;

[0101] In the specific implementation process, the specific steps of block 205 are:

[0102] The prostate region marked by the secondary mark of each inspection image is processed in parallel by using a plurality of preset processing nodes, specifically including denoising, contrast enhancement, normalization and smoothing filtering, to obtain processed inspection image data;

[0103] Specifically, a parallel computing framework containing multiple processing nodes is constructed to process the prostate areas marked twice in each inspection image in parallel. Specifically, filters are used to remove noise in the image, reduce random interference in the image, enhance the contrast of the image through histogram equalization, Laplace sharpening and other methods to make the boundary between the prostate area and the surrounding tissue clearer, and normalize the pixel value of the image to a specific range (specifically 0-1 or -1 to 1) to facilitate subsequent processing and analysis. The Gaussian filter in the smoothing filter is used to smooth the image to reduce the detail noise in the image while retaining important edge information.

[0104] Extracting prostate region features based on the processed inspection image data, including shape features, texture features, and statistical features based on the inspection image data;

[0105] Specifically, the shape features of the prostate region are extracted using edge detection and contour extraction in image processing technology, where the shape features specifically include area, perimeter, circularity, and aspect ratio; and the gray level co-occurrence matrix and local binary pattern in the texture analysis method are used to extract the texture features of the prostate region; and the statistical features of the prostate region are calculated, specifically the mean, variance, and histogram distribution.

[0106] In block 206, the secondarily identified prostate region features are evaluated respectively by the processing nodes to obtain evaluation values ​​of each feature;

[0107] In the specific implementation process, the specific steps of block 206 are:

[0108] The prostate region features extracted from each processing node are standardized;

[0109] Specifically, all prostate region features extracted from each processing node are standardized through Z-score standardization (zero mean standardization) and Min-Max standardization (maximum minimum standardization).

[0110] For example, if three features of the prostate area (Area), perimeter (Perimeter) and mean intensity (Mean Intensity) are extracted from a certain processing node, the Z-score standardization method can be used to convert the original value of each feature into a standardized value relative to its mean and standard deviation.

[0111] The quality of each extracted prostate region feature was evaluated by calculating feature statistics and analyzing the correlation between features.

[0112] Specifically, the statistics of each feature, such as the mean, standard deviation, median, interquartile range, etc., are calculated, and the linear or monotonic relationship between the features is evaluated by calculating the Pearson correlation coefficient or Spearman rank correlation coefficient in the correlation coefficient between the features, and the quality of each feature is evaluated.

[0113] According to the preset evaluation index and the result of quality evaluation, the contribution of each extracted prostate region feature is calculated and converted into an evaluation value;

[0114] Specifically, appropriate evaluation indicators are selected according to the analysis objectives and data characteristics, specifically information gain and Gini index, to quantify the contribution of features to the predictive ability of the target variable. Then, based on the preset evaluation indicators and the results of quality assessment, the contribution of each prostate region feature is calculated, and finally the calculated contribution is converted into a unified evaluation value.

[0115] In block 207, based on the evaluation values ​​of each feature, each feature is fused through a preset central processing node and an evaluation value threshold to form a multimodal feature vector;

[0116] In the specific implementation process, the specific steps of block 207 are:

[0117] Based on the evaluation value of each prostate region feature, the preset threshold is used to screen out features that contribute most to the prediction task;

[0118] Specifically, based on historical data or expert experience, one or more thresholds of evaluation values ​​are set, and the evaluation values ​​of each prostate region feature obtained are compared with the set thresholds. Features that are higher than the set thresholds are screened as features with high contribution to the prediction task.

[0119] The filtered features are fused to form a multimodal feature vector, and the fused multimodal feature vector is overall standardized to obtain a standardized multimodal feature vector;

[0120] Specifically, according to the correlation and complementarity between the features, the screened features are fused into a multimodal feature vector by using a series or parallel method, and the fused multimodal feature vector is standardized as a whole.

[0121] In box 208, the multimodal feature vector is input into the trained deep learning model and the prediction result is output.

[0122] In the specific implementation process, the specific steps of block 208 are:

[0123] Load the deep learning model that has been pre-trained based on historical prostate disease imaging data and prostate bone metastasis data;

[0124] Specifically, historical prostate disease imaging data and prostate bone metastasis data are obtained, and the correlation analysis method is used to obtain the correlation between individual features of the prostate disease imaging data and the correlation between the prostate bone metastasis data and the prostate disease. Based on the corresponding results obtained as a training set, a deep learning model is used as an architecture training to obtain a deep learning model for prediction.

[0125] The standardized multimodal feature vector is input into the trained deep learning model to calculate and output the prediction results, which specifically include disease classification, disease severity assessment, and bone metastasis.

[0126] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0127] Figure 3 A schematic block diagram of an example electronic device 300 that can be used to implement an embodiment of the present disclosure is shown. The electronic device can be used, for example, for one or more operations in the embodiments of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0128] like Figure 3 As shown, the electronic device 300 includes a computing unit 301, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 302 or a computer program loaded from a storage unit 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the device can also be stored. The computing unit 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0129] Multiple components in the electronic device are connected to the I / O interface 305, including: an input unit 306, such as a keyboard, a mouse, etc.; an output unit 307, such as various types of displays, speakers, etc.; a storage unit 308, such as a disk, an optical disk, etc.; and a communication unit 309, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 309 allows the device to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0130] The computing unit 301 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 301 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 301 performs the various methods and processes described above, such as the methods of various embodiments of the present disclosure. For example, in some embodiments, the methods of various embodiments of the present disclosure may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 308. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via ROM 302 and / or communication unit 309. When the computer program is loaded into RAM 303 and executed by the computing unit 301, one or more steps of the methods of various embodiments of the present disclosure described above may be performed. Alternatively, in other embodiments, the computing unit 301 may be configured to perform the methods of various embodiments of the present disclosure in any other appropriate manner (e.g., by means of firmware).

[0131] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0132] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0133] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0135] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0136] A computer system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The relationship between the client and the server is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server may also be a server of a distributed system, or a server combined with a blockchain.

[0137] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0138] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

[0139] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

Claims

1. A multimodal prostate cancer and hyperplasia prediction system, characterized in that: The prediction system comprises: A region division module is used to preliminarily divide the prostate region according to the position distribution of each organ in each collected inspection image and the shape information of the prostate; A model building module is used to extract the correlation between the patient's age, body shape and prostate condition based on historical patient prostate condition data to build a prostate recognition model; The identification module is used to identify each of the collected inspection images respectively through the prostate identification model, and to preliminarily identify the preliminarily divided prostate area to obtain the preliminarily identified prostate area; An interactive module, for using an interactive operation mode to allow a doctor or an expert to perform a secondary marking on the initially marked prostate region to obtain a secondary marked prostate region; A parallel processing module, used to use a plurality of preset processing nodes to perform parallel processing on the prostate region marked twice in each inspection image, and extract features of the prostate region marked twice; An evaluation module, used to evaluate the secondary marked prostate region features respectively through processing nodes to obtain evaluation values ​​of each feature; A fusion processing module is used to fuse each feature based on the evaluation value of each feature through a preset central processing node and an evaluation value threshold to form a multi-modal feature vector; The prediction module is used to input the multimodal feature vector into the trained deep learning model and output the prediction result.

2. A method for constructing a multi-modal prostate cancer and hyperplasia prediction system, applied to a multi-modal prostate cancer and hyperplasia prediction system as claimed in claim 1, characterized in that: The prediction construction method comprises the following steps: The prostate region is preliminarily divided according to the position distribution of each organ in each inspection image collected and the shape information of the prostate; Based on the historical patient prostate disease data, the correlation between the patient's age, body shape and prostate disease is extracted to build a prostate recognition model; Using the prostate recognition model, each of the collected inspection images is recognized, and the initially divided prostate area is preliminarily marked to obtain a preliminarily marked prostate area; Using the interactive operation mode, doctors or experts conduct secondary marking of the initially marked prostate area to obtain the secondary marked prostate area; Using a plurality of preset processing nodes to process the secondary marked prostate region in each inspection image in parallel, and extracting the secondary marked prostate region features; The features of the prostate region marked by the secondary identification are evaluated respectively through the processing nodes to obtain the evaluation value of each feature; Based on the evaluation value of each feature, each feature is fused through a preset central processing node and evaluation value threshold to form a multimodal feature vector; Input the multimodal feature vector into the trained deep learning model and output the prediction result.

3. A method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 2, characterized in that: The preliminary division of the prostate region according to the position distribution of each organ in each collected inspection image and the shape information of the prostate comprises the following steps: Collecting various examination images of the patient, including but not limited to MRI images and ultrasound images; Use image segmentation technology to identify the location of each organ in each examination image and determine the location of organs adjacent to the prostate; Based on the standard morphological features of the prostate, the prostate contour is outlined in each examination image and the prostate area is preliminarily divided.

4. A method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 2, characterized in that: The method of extracting the correlation between the patient's age, body shape and prostate condition based on the historical patient prostate condition data to construct a prostate recognition model specifically includes the following steps: Obtain medical records of historical patients, including data sets of patient age, body shape, and prostate condition information; Analyze the correlation between age, body shape and prostate disease through statistical methods, identify the key factors that affect prostate disease, and obtain the analysis results of the correlation; The analysis results of the association relationship were used as the training set, and a machine learning algorithm was used to build and train a prostate recognition model.

5. The method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 2, characterized in that: The method of identifying each collected inspection image by using the prostate recognition model, and preliminarily marking the preliminarily divided prostate area, and obtaining the preliminarily marked prostate area specifically includes the following steps: Inputting each collected inspection image into the prostate recognition model, the prostate recognition model automatically recognizes and marks the prostate area according to each input inspection image; The prostate region is automatically identified and marked and output in a visual form to obtain a preliminary identification of the prostate region.

6. A method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 2, characterized in that: The interactive operation mode is used to perform secondary marking of the initially marked prostate area by a doctor or an expert, and obtaining the secondary marked prostate area specifically includes the following steps: The initially marked prostate area is shown to a doctor or expert, and the doctor or expert performs secondary marking on the initially marked prostate area based on professional knowledge; After confirming that the secondary identification result is correct, save and output it as the secondary identified prostate area.

7. A method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 2, characterized in that: The method of using a plurality of preset processing nodes to perform parallel processing on the prostate region marked twice in each inspection image and extracting the features of the prostate region marked twice specifically includes the following steps: The prostate region marked by the secondary mark of each inspection image is processed in parallel by using a plurality of preset processing nodes, specifically including denoising, contrast enhancement, normalization and smoothing filtering, to obtain processed inspection image data; Based on the processed inspection image data, the prostate region features are extracted, including shape features, texture features and statistical features based on the inspection image data.

8. The method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 2, characterized in that: The process of evaluating the secondary marked prostate region features respectively through the processing nodes to obtain the evaluation values ​​of each feature specifically comprises the following steps: The prostate region features extracted from each processing node are standardized; The quality of each extracted prostate region feature was evaluated by calculating feature statistics and analyzing the correlation between features. According to the preset evaluation index and the results of quality assessment, the contribution of each extracted prostate region feature is calculated and converted into an evaluation value.

9. A method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 2, characterized in that: The method of fusing each feature based on the evaluation value of each feature through a preset central processing node and an evaluation value threshold to form a multimodal feature vector specifically includes the following steps: Based on the evaluation value of each prostate region feature, the preset threshold is used to screen out features that contribute most to the prediction task; The filtered features are fused to form a multimodal feature vector, and the fused multimodal feature vector is overall standardized to obtain a standardized multimodal feature vector.

10. A method for constructing a multimodal prostate cancer and hyperplasia prediction system according to claim 9, characterized in that: The step of inputting the multimodal feature vector into the trained deep learning model and outputting the prediction result specifically includes the following steps: Load the deep learning model that has been pre-trained based on historical prostate disease imaging data and prostate bone metastasis data; The standardized multimodal feature vector is input into the trained deep learning model to calculate and output the prediction results, which specifically include disease classification, disease severity assessment, and bone metastasis.

Images