Method and device for determining historical cases, electronic device and storage medium

Abstract

The present invention discloses a method for determining historical cases and a device, electronic device and storage medium thereof, and relates to the field of image processing technology, wherein the determination method comprises: obtaining N-dimensional medical images of a target object, and performing feature extraction on the medical images in each dimension to obtain a first feature of the medical image, and determining a reference case set of the target object from a feature plan library based on the first feature of the medical image in each dimension. The present invention solves the technical problem in the related art that the most similar historical cases cannot be accurately retrieved, resulting in a low accuracy of the generated radiotherapy plan.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method for determining historical cases, a device thereof, an electronic device and a storage medium. Background Art

[0002] At present, efficient and accurate generation of radiotherapy plans is a key step to ensure the effectiveness of radiotherapy. The generation of radiotherapy plans is affected by many factors, such as case imaging information, field information, and optimization conditions. Taking the IMRT (Intensity-modulated radiation therapy) process as an example, the design process of radiotherapy plans can be simplified as follows: according to the case imaging information, the field is made to cover the PTV (Planning Target Volume) as much as possible, and the irradiation of OAR (Organs at Risk) is reduced as much as possible, and then specific optimization conditions are designed according to the optimization goal. Then, IMRT iterates the complete radiotherapy plan through calculation, and then adjusts the radiotherapy plan and repeats the above steps until the radiotherapy plan meets the requirements. However, the entire radiotherapy plan generation process is affected by many factors, and the use of inappropriate parameters for generation will lead to problems such as long plan optimization time or inaccurate plan generation.

[0003] Currently, deep learning models can be used to predict the dose distribution files used in radiotherapy plans. Specifically, historical dose data can be used as training data to train a deep learning model, and then the patient's medical images can be input into the trained deep learning model to predict the corresponding dose distribution file. However, the prediction accuracy of the deep learning model depends on the quality of the training data, and it needs to rely on a large amount of historical case planning data that meets the quality standards for training. However, historical case planning data is limited by personal privacy and the number of historical cases. Therefore, it is difficult to obtain a high-precision model through training of historical case planning data, and the predicted dose distribution output by the deep learning model is only based on the learned image features. It is difficult to meet the principle of beam dose generation in actual radiotherapy, resulting in differences between the output results and the actual results.

[0004] In the related art, in order to solve the problem of inaccurate predicted dose distribution files, the OVH (Overlap Volume Histogram) method can be used to extract strongly correlated historical plans from the plan library, so as to use the strongly correlated historical plans as a reference for the current plan. Specifically, the OVH plan library can be generated first, and then OVH data can be generated based on the patient's medical images. After that, the machine learning nearest neighbor method is used to calculate the degree of approximation between the OVH data and the data in the plan library to return the plan of the most similar historical case, so that the current plan can be optimized or generated based on the plan of the most similar historical case.

[0005] However, the related art only searches for the most similar cases based on the imaging features of the cases and the changes in the distance relationship between the organs and the target areas, which leads to the problem of low retrieval accuracy.

[0006] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0007] The embodiments of the present invention provide a method for determining historical cases and its device, electronic device and storage medium, so as to at least solve the technical problem in the related art that the most similar historical cases cannot be accurately retrieved, resulting in low accuracy of the generated radiotherapy plan.

[0008] According to one aspect of an embodiment of the present invention, a method for determining historical cases is provided, comprising: acquiring an N-dimensional medical image of a target object, and performing feature extraction on the medical image in each dimension to obtain a first feature of the medical image, wherein the first feature of the medical image comprises a relationship between an overlapping area and a distance range; the overlapping area is an overlapping area between an organ at risk on the medical image and a target area on the medical image under each radiation source; the distance range is a distance range composed of the distances of the organ at risk from the radiation source; N is a positive integer; based on the first feature of the medical image in each dimension, determining a reference case set of the target object from a feature plan library, wherein the feature plan library comprises: a plurality of historical cases; the reference case set comprises: at least one reference case; the reference case is used to provide a reference when determining a radiotherapy plan for the target object.

[0009] Furthermore, the step of performing feature extraction on the medical image in each dimension to obtain the first feature of the medical image includes: determining the irradiation range of each of the field sources based on the target field range of the medical image and the delineation information of the target area; determining the overlapping area of ​​each of the organs at risk and the target area under each of the field sources based on the irradiation range and the delineation information of each of the organs at risk; determining the sub-feature of each of the organs at risk under each of the field sources based on the relationship between the overlapping area and the distance range of the organs at risk from each of the field sources; and constructing the first feature of the medical image under the target field range based on the sub-feature of each of the organs at risk under each of the field sources.

[0010] Furthermore, based on the first feature of the medical image in each dimension, the step of determining a reference case set of the target object from a feature plan library includes: screening an initial historical case set belonging to the target cancer from the feature plan library, and determining a sub-historical case set corresponding to each target field range from the initial historical case set, wherein the target cancer is the cancer determined for the target object; the target field range is the field range corresponding to the target cancer; calculating the feature similarity between the first feature and the second feature corresponding to each sub-historical case under the same target field range; and determining the reference case set based on all the feature similarities.

[0011] Furthermore, the step of calculating the feature similarity between the first feature under the same target field range and the second feature corresponding to each sub-historical case includes: determining the weight of each organ at risk under each target field range based on the overlapping area and distance between the organ at risk and the target area on the medical image; calculating the first similarity of each organ at risk under the target field range based on the first feature corresponding to the same target field range and the second feature of each sub-historical case; performing weighted processing on the first similarities of all organs at risk according to the weight of each organ at risk to obtain the second similarity under the target field range; accumulating the second similarity of each target field range on the medical image in each dimension to obtain the feature similarity under each target field range.

[0012] Furthermore, based on the first feature corresponding to the same target field range and the second feature of each sub-historical case, the step of calculating the first similarity of each endangered organ under the target field range includes: converting the sub-feature of each endangered organ in the first feature into a first type of curve under the target field range; converting the sub-feature of each endangered organ in the second feature of each sub-historical case into a second type of curve under the target field range; for the same endangered organ, calculating the difference value between the first type of curve and the second type of curve located on the same coordinate axis at each same coordinate value; and accumulating the difference values ​​at all the coordinate values ​​to obtain the first similarity of each endangered organ under the target field range.

[0013] Furthermore, based on all the feature similarities, the step of determining the reference case set includes: sorting the feature similarities, and based on the sorting results, determining a feature similarity set that is within a preset feature similarity threshold range; characterizing the second feature corresponding to each feature similarity in the feature similarity set as a reference feature; and adding the historical case corresponding to each reference feature to the reference case set.

[0014] Furthermore, the determination method also includes: acquiring multiple historical cases and determining the field range corresponding to each of the historical cases; based on the field range, target area delineation information and endangered organ delineation information of the historical cases, performing feature extraction on the historical cases to obtain a second feature of the historical cases; and building the feature plan library based on all the historical cases and the second feature of each of the historical cases.

[0015] According to another aspect of an embodiment of the present invention, a device for determining historical cases is also provided, including: an extraction unit, used to obtain an N-dimensional medical image of a target object, and perform feature extraction on the medical image in each dimension to obtain a first feature of the medical image, wherein the first feature of the medical image includes a relationship between an overlapping area and a distance range; the overlapping area is an overlapping area between an organ at risk on the medical image and a target area on the medical image under each radiation source; the distance range is a distance range composed of the distances of the organ at risk from the radiation source; N is a positive integer; a determination unit, used to determine a reference case set of the target object from a feature plan library based on the first feature of the medical image in each dimension, wherein the feature plan library includes: a plurality of historical cases; the reference case set includes: at least one reference case; the reference case is used to provide a reference when determining a radiotherapy plan for the target object.

[0016] Further, the extraction unit includes: a first determination module, used to determine the irradiation range of each of the field sources based on the target field range of the medical image and the delineation information of the target area; a second determination module, used to determine the overlapping area of ​​each of the organs at risk and the target area under each of the field sources based on the irradiation range and the delineation information of each of the organs at risk; a third determination module, used to determine the sub-features of each of the organs at risk under each of the field sources based on the relationship between the overlapping area and the distance range of the organs at risk from each of the field sources; and a first construction module, used to construct the first feature of the medical image under the target field range based on the sub-features of each of the organs at risk under each of the field sources.

[0017] Furthermore, the determination unit includes: a fourth determination module, used to screen an initial historical case set belonging to the target cancer type from the feature plan library, and determine a sub-historical case set corresponding to each target field range from the initial historical case set, wherein the target cancer type is the cancer type determined for the target object; the target field range is the field range corresponding to the target cancer type; a first calculation module, used to calculate the feature similarity between the first feature and the second feature corresponding to each sub-historical case under the same target field range; and a fifth determination module, used to determine the reference case set based on all the feature similarities.

[0018] Furthermore, the first calculation module includes: a first determination submodule, used to determine the weight of each of the organs at risk under each of the target field ranges based on the overlapping area and distance between the organs at risk on the medical image and the target area; a first calculation submodule, used to calculate the first similarity of each of the organs at risk under the target field range based on the first feature corresponding to the same target field range and the second feature of each of the sub-historical cases; a first processing submodule, used to perform weighted processing on the first similarities of all the organs at risk according to the weight of each organ at risk to obtain the second similarity under the target field range; a first accumulation submodule, used to accumulate the second similarity of each of the target field ranges on the medical image in each dimension to obtain the feature similarity under each of the target field ranges.

[0019] Furthermore, the first calculation submodule includes: a first conversion submodule, used to convert the sub-feature of each of the organs at risk in the first feature into a first type of curve within the target field range; a second conversion submodule, used to convert the sub-feature of each of the organs at risk in the second feature of each of the sub-historical cases into a second type of curve within the target field range; a second calculation submodule, used to calculate, for the same organ at risk, the difference value between the first type of curve and the second type of curve located on the same coordinate axis at each same coordinate value; and a second accumulation submodule, used to accumulate the difference values ​​at all the coordinate values ​​to obtain the first similarity of each of the organs at risk within the target field range.

[0020] Furthermore, the fifth determination module includes: a first sorting submodule, used to sort the feature similarities, and based on the sorting results, determine a feature similarity set within a preset feature similarity threshold range; a first characterization submodule, used to characterize the second feature corresponding to each feature similarity in the feature similarity set as a reference feature; and a first adding submodule, used to add the historical cases corresponding to each reference feature to the reference case set.

[0021] Furthermore, the determination device also includes: a sixth determination module, used to obtain multiple historical cases and determine the field range corresponding to each of the historical cases; a first extraction module, used to extract features of the historical cases based on the field range, target area delineation information and endangered organ delineation information of the historical cases to obtain the second feature of the historical cases; a first building module, used to build the feature plan library based on all the historical cases and the second feature of each of the historical cases.

[0022] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned methods for determining historical cases.

[0023] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned methods for determining historical cases.

[0024] In the present invention, an N-dimensional medical image of the target object is obtained, and feature extraction is performed on the medical image in each dimension to obtain the first feature of the medical image. Based on the first feature of the medical image in each dimension, a reference case set of the target object is determined from a feature plan library, thereby solving the technical problem in related technologies that the most similar historical cases cannot be accurately retrieved, resulting in low accuracy of the generated radiotherapy plan.

[0025] In the present invention, by combining the field angle and the number of fields, the characteristics of the multidimensional medical image of the target object are extracted, and then based on the extracted characteristics, the closest reference case set is retrieved from the feature plan library for subsequent radiotherapy plan optimization. In this way, the extracted features integrate the field information and dimensional information, so that the most similar historical cases can be searched efficiently and accurately according to the extracted features, achieving the technical effect of accurately retrieving the most similar historical cases for subsequent optimization of radiotherapy plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0027] Figure 1 is a flow chart of an optional method for determining historical cases according to an embodiment of the present invention;

[0028] Figure 2 is a schematic diagram of an optional overlapping area of ​​an organ at risk and a target area under a radiation field source according to an embodiment of the present invention;

[0029] Figure 3 is a schematic diagram of overlapping areas of organs at risk and target areas under multiple radiation field sources according to an optional embodiment of the present invention;

[0030] Figure 4 is a schematic diagram of an optional bladder_Dice according to an embodiment of the present invention;

[0031] Figure 5 is a schematic diagram of an optional bone marrow_Dice according to an embodiment of the present invention;

[0032] Figure 6 is a schematic diagram of an optional rectum_Dice according to an embodiment of the present invention;

[0033] Figure 7 is a schematic diagram of an optional femoral head l_Dice according to an embodiment of the present invention Figure 1 ;

[0034] Figure 8 is a schematic diagram of an optional femoral head r_Dice according to an embodiment of the present invention Figure 2 ;

[0035] Fig. 9 is a schematic diagram of an optional most similar historical case retrieval process according to an embodiment of the present invention;

[0036] Fig.10 is a schematic diagram of an optional historical case determination device according to an embodiment of the present invention;

[0037] Fig.11 This is a hardware structure block diagram of an electronic device (or mobile device) for a method for determining historical cases according to an embodiment of the present invention. DETAILED DESCRIPTION

[0038] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0039] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0040] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) collected and involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data are in compliance with the relevant laws, regulations and standards of the relevant regions, necessary confidentiality measures are taken, and public order and good customs are not violated, and corresponding operation portals are provided for users to choose to authorize or refuse. For example, an interface is set up between this system and relevant users or organizations. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0041] In order to address the problem that the related art only considers the change in the distance relationship between the organ and the target area, but does not take into account the possible angular deviation of the endangered organ, and also does not consider the field information that has a significant impact on the dose, the present invention proposes a method for fusing the field information and multi-dimensional angle information, which can effectively improve the retrieval accuracy and obtain the most similar historical cases for reference in optimizing the radiotherapy plan.

[0042] In the present invention, since the effectiveness of the deep learning model is strongly correlated with the consistency of the distribution of training data, in order to maintain the effectiveness of the model in the case of frequent changes in the database, it is necessary to regularly retrain the model to ensure the effectiveness of the newly added data in the database, which consumes a high amount of training time and resources, and regular maintenance is troublesome. However, the image features extracted by fusing the field information and multi-dimensional angle information in the present invention can continuously supplement and improve the database, and when new data is entered into the database, its features can be calculated and saved to realize an effective search function, effectively saving training time and resources.

[0043] In the present invention, the most similar historical cases can be retrieved accurately and effectively, which can provide effective dose references when manually optimizing radiotherapy plans, and can also be used as initial constraints when automatically optimizing radiotherapy plans. Compared with the dose distribution predicted by deep learning in related technologies, the historical distribution can be fully reproduced under the premise of appropriate settings, and has higher reliability.

[0044] The present invention is described in detail below in conjunction with various embodiments.

[0045] Embodiment 1

[0046] According to an embodiment of the present invention, an embodiment of a method for determining historical cases is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0047] Figure 1 is a flow chart of an optional method for determining historical cases according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0048] Step S101, obtaining an N-dimensional medical image of the target object, and performing feature extraction on the medical image in each dimension to obtain a first feature of the medical image, wherein the first feature of the medical image includes a relationship between an overlapping area and a distance range; the overlapping area is an overlapping area between an organ at risk on the medical image and a target area on the medical image under each radiation source; the distance range is a distance range consisting of the distance between the organ at risk and the radiation source; and N is a positive integer.

[0049] Step S102, based on the first feature of the medical image in each dimension, determine a reference case set of the target object from the feature plan library, wherein the feature plan library includes: multiple historical cases; the reference case set includes: at least one reference case; the reference case is used to provide a reference when determining a radiotherapy plan for the target object.

[0050] In an embodiment of the present invention, the features of the multidimensional medical images of the target object are extracted by combining the field angle and the number of fields, and then the closest reference case set is retrieved from the feature plan library based on the extracted features to optimize the subsequent radiotherapy plan. In this way, the extracted features integrate the field information and the dimensional information, so that the most similar historical cases can be searched efficiently and accurately based on the extracted features, thereby solving the technical problem that the most similar historical cases cannot be accurately retrieved in the related technology, resulting in low accuracy of the generated radiotherapy plan, and achieving the technical effect of accurately retrieving the most similar historical cases to facilitate the subsequent optimization of the radiotherapy plan.

[0051] The embodiment of the present invention is described in detail below in combination with the above steps.

[0052] Step S101, obtaining an N-dimensional medical image of the target object, and performing feature extraction on the medical image in each dimension to obtain a first feature of the medical image, wherein the first feature of the medical image includes a relationship between an overlapping area and a distance range; the overlapping area is an overlapping area between an organ at risk on the medical image and a target area on the medical image under each radiation source; the distance range is a distance range consisting of the distance between the organ at risk and the radiation source; and N is a positive integer.

[0053] In step S101, the case system can obtain N-dimensional medical images (N is a positive integer, i.e., multi-dimensional medical images, such as three-dimensional and four-dimensional medical images) of the target object (e.g., the patient for whom the radiotherapy plan is to be generated or optimized) from the case database (the case in which the patient is stored) and then extract the features of the medical images in each dimension (i.e., the first features). It should be noted that the features extracted in this embodiment integrate the field information and the dimension information, i.e., the first features extracted from the medical images include the relationship between the overlapping area (i.e., the overlapping area of ​​the organ at risk on the medical image and the target area on the medical image under each field source) and the distance range (i.e., the distance range composed of the distance between the organ at risk and the field source).

[0054] In some optional embodiments, the AB-OVH (Angle Based Overlap Volume Histogram) algorithm can be used to extract the overlapping volume histogram of OAR (organ at risk) and target area under different beams (used to represent the relationship between the overlapping shadow area and the distance from the radiation source), and the overlapping volume histogram and the beam are used as features of the medical image. In this way, when extracting the medical image features, both the angle relationship information between the organ and the target area and the field information are considered, which can improve the retrieval accuracy of subsequent historical cases.

[0055] In the embodiment of the present invention, the features corresponding to the medical image may include all the fields, and each field may correspond to a list, so as to record the overlapping volume histograms corresponding to all OARs in the list.

[0056] For example, the N-dimensional medical image of the target object may be a three-dimensional CT (Computed Tomography) image, or a multi-dimensional image generated in real time by other medical tools or reconstructed later.

[0057] Optionally, the step of performing feature extraction on the medical image in each dimension to obtain the first feature of the medical image includes: determining the irradiation range of each field source based on the target field range of the medical image and the delineation information of the target area; determining the overlapping area of ​​each organ at risk and the target area under each field source based on the irradiation range and the delineation information of each organ at risk; determining the sub-features of each organ at risk under each field source based on the relationship between the overlapping area and the distance range of the organ at risk from each field source; and constructing the first feature of the medical image under the target field range based on the sub-features of each organ at risk under each field source.

[0058] In the embodiment of the present invention, the target radiation field range (including radiation field angle and radiation field number) of the medical image can be determined according to the type of cancer suffered by the target object, and the target radiation field range can be multiple. For example, the commonly used radiation field angle and radiation field number are obtained based on the historical cases of patients with the same cancer type.

[0059] In the embodiment of the present invention, the target area and the organ at risk may be delineated in advance on each dimensional plane according to the medical image to obtain the delineation information of the target area and the delineation information of the organ at risk.

[0060] In an embodiment of the present invention, the irradiation range of each field source can be determined first according to the target field range of the medical image and the delineation information of the target area. Specifically, since it is known that the original wheelbase is fixed (that is, the distance from the radiation source to the target area is a preset fixed value), the position of each field source can be located according to each field angle, and then according to the position of each field source and the delineation information of the target area, the range (that is, the irradiation range) of the field actually generated by each radiation source in the sampling point (that is, the target area sampling point indicated by the delineation information of the target area) can be calculated by a vertical line.

[0061] Exemplarily, the straight line from the radiation source to the center of the target area is obtained through the field angle and the fixed distance value from the radiation source to the target area, and the coordinates of the connection point are calculated based on the perpendicular line corresponding to the straight line and the edge of the target area under ideal conditions. The area irradiated by the radiation source to the target area (i.e., the irradiation range) can be approximately calculated.

[0062] In the embodiment of the present invention, according to the irradiation range of each radiation source and the delineation information of each organ at risk, the overlapping area between each organ at risk and the target area under each field source can be determined. Specifically, the irradiation range of the field actually generated by each radiation source in the sampling point of the target area and the delineation range of the organ at risk (determined by the delineation information) are multiplied to obtain the sampling points on the overlapping area of ​​the two, that is, the actual irradiation situation of the beam on the OAR, thereby obtaining the overlapping area between each organ at risk and the target area under each field source.

[0063] Figure 2 is a schematic diagram of an optional overlapping area of ​​an organ at risk and a target area under a radiation field source according to an embodiment of the present invention, such as Figure 2 As shown in the figure, the target area PTV, organs at risk 1 (OAR1) and organs at risk 2 (OAR2) are outlined in the current dimension medical image. According to the field angle corresponding to the medical image, the position of the field source source can be located, and then according to the position of the source, combined with the PTV outline information, the irradiation range of the field actually generated by the source in the PTV can be calculated by the vertical line, and then the irradiation range is overlapped with the outline range of the organs at risk OAR (OAR1 and OAR2), and the overlapping area of ​​the organs at risk and the target area under the field source can be calculated.

[0064] In the embodiment of the present invention, for each organ at risk under each radiation source, according to the relationship between its overlapping area and the distance range (i.e., the distance range of the organ at risk from the radiation source), an overlapping volume histogram of each organ at risk under each radiation source can be constructed, and the overlapping volume histogram can be characterized as a sub-feature of each organ at risk under each radiation source. Then, according to the sub-feature of each organ at risk under each radiation source, the first feature of the medical image under the target radiation range can be constructed (including the overlapping volume histogram of each organ at risk on the medical image of the current dimension under each radiation source).

[0065] In some optional embodiments, the plane where the same dimensional medical image is located may have multiple field sources, and the OARs near the field sources may intersect with the fields. For each OAR associated with the field, only the intersection portion is considered as the delineation area of ​​the OAR.

[0066] Figure 3 FIG. 1 is a schematic diagram of overlapping areas of organs at risk and target areas under multiple radiation field sources according to an optional embodiment of the present invention, such as Figure 3 As shown in the figure, it shows the selection method of the overlapping area between OAR (organ at risk 1 (OAR1), organ at risk 2 (OAR2)) and the target PTV when there are three radiation sources (source 1, source 2, source 3). In the AB-OVH calculation, in the radiation field generated by radiation source 1 (source 1), only the part where OAR1 and OAR2 overlap with the radiation field is considered for feature extraction. Figure 3 In the radiation field generated by source 2, only the overlapped part of OAR2 and the radiation field is considered for feature extraction; in radiation source 3, since there is no OAR coverage, the OAR features in this case are not extracted.

[0067] In an embodiment of the present invention, because beam irradiation is a multi-dimensional space (e.g., three-dimensional space), the relationship between the target area and the OAR (i.e., the extracted features) not only exists on one horizontal plane, but also needs to be extracted on other planes (e.g., cross-sectional plane, sagittal plane, coronal plane, etc.) in a multi-dimensional coordinate system to complete the relationship between the target area, OAR and radiation field in multiple dimensions.

[0068] In the embodiment of the present invention, the extraction of image features is improved by supplementing the angular relationship information between the organ and the target area, and the actual principle of the field and the process of its effect on the dose distribution can be effectively simulated in a way that the field and the organ overlap, and the field information can be fused with high interpretability, thereby improving the subsequent retrieval accuracy.

[0069] Step S102, based on the first feature of the medical image in each dimension, determine a reference case set of the target object from the feature plan library, wherein the feature plan library includes: multiple historical cases; the reference case set includes: at least one reference case; the reference case is used to provide a reference when determining a radiotherapy plan for the target object.

[0070] In step S102, by calculating the similarity between the first feature of the medical image in each dimension and the feature corresponding to each historical case in the feature planning library, a reference case set of the target object can be screened from the feature planning library, and the reference case set includes: at least one reference case.

[0071] In this embodiment, the screened reference cases are historical cases that are closest to the cases of the target object. These reference cases can provide references when determining the radiotherapy plan for the target object. For example, the doses used in these reference cases can be used as effective dose references when manually optimizing the radiotherapy plan, and can also be used as initial constraints when automatically optimizing the radiotherapy plan.

[0072] In some optional embodiments, for the multiple reference cases screened out, multiple types of visualizations can be performed on the reference cases as required.

[0073] Optionally, based on the first feature of the medical image in each dimension, the step of determining a reference case set of the target object from the feature plan library includes: screening an initial historical case set belonging to the target cancer type from the feature plan library, and determining a sub-historical case set corresponding to each target field range from the initial historical case set, wherein the target cancer type is the cancer type determined for the target object; the target field range is the field range corresponding to the target cancer type; calculating the feature similarity between the first feature under the same target field range and the second feature corresponding to each sub-historical case; and determining the reference case set based on all feature similarities.

[0074] In an embodiment of the present invention, an initial historical case set belonging to the target cancer type (the target cancer type is the cancer type determined by the target object) can be first screened from the feature plan library, which can reduce the number of historical cases retrieved subsequently, thereby effectively improving the retrieval efficiency. Then, a sub-historical case set corresponding to each target field range (i.e., the field range corresponding to the target cancer type, and each type of cancer may correspond to multiple field ranges) is determined from the initial historical case set, so that under the same target field range, the closest historical cases can be further screened in the corresponding sub-historical case set, which is conducive to further improving the retrieval efficiency. Afterwards, the feature similarity between the first feature and the second feature corresponding to each sub-historical case under the same target field range can be calculated to screen out the closest reference case set based on all feature similarities.

[0075] Exemplarily, the AB-OVH algorithm may be used to calculate the AB-OVH similarity (ie, feature similarity) between the first feature and each second feature.

[0076] Optionally, the step of calculating the feature similarity between the first feature under the same target field range and the second feature corresponding to each sub-historical case includes: determining the weight of each organ at risk under each target field range based on the overlapping area and distance between the organ at risk and the target area on the medical image; calculating the first similarity of each organ at risk under the target field range based on the first feature corresponding to the same target field range and the second feature of each sub-historical case; weighting the first similarities of all organs at risk according to the weight of each organ at risk to obtain the second similarity under the target field range; accumulating the second similarity of each target field range on the medical image in each dimension to obtain the feature similarity under each target field range.

[0077] In an embodiment of the present invention, the weight of each organ at risk in each target field range can be determined according to the overlapping area and distance between the organ at risk and the target area on the medical image. For example, the weight can be set as follows: the weight ratio of organs overlapping with the target area, organs near the target area, and organs far from the target area is 4:2:1. Then, according to the first feature and the second feature under the same target field range, the first similarity of each organ at risk in the target field range is calculated, and then according to the weight of each organ at risk, the first similarities of all organs at risk are weighted to obtain the second similarity in the target field range. After that, the second similarity in the medical image in each dimension can be accumulated to obtain the feature similarity in the target field range.

[0078] For example, under different field classifications (i.e., different target field ranges), the AB-OVH algorithm can be used to calculate the AB-OVH similarity (i.e., the first similarity) of each OAR under the same target field range, and then the AB-OVH similarities are weighted and accumulated according to the distance of the OAR from the target area and the overlapping area with the target area to obtain the similarity under the target field range (i.e., the second similarity). After that, the similarities under the target field range in all dimensions are accumulated to obtain the feature similarity under the target field range.

[0079] Optionally, based on the first feature corresponding to the same target field range and the second feature of each sub-historical case, the step of calculating the first similarity of each organ at risk within the target field range includes: converting the sub-feature of each organ at risk in the first feature into a first-type curve within the target field range; converting the sub-feature of each organ at risk in the second feature of each sub-historical case into a second-type curve within the target field range; for the same organ at risk, calculating the difference between the first-type curve and the second-type curve located on the same coordinate axis at each same coordinate value; and accumulating the difference values ​​at all coordinate values ​​to obtain the first similarity of each organ at risk within the target field range.

[0080] In an embodiment of the present invention, similarity calculation is achieved by calculating the difference value between the curves (i.e., the value obtained by accumulating the numerical difference when the coordinates are the same). Specifically, the sub-feature of each organ at risk in the first feature can be converted into a first-type curve under the target field range, and the sub-feature of each organ at risk in the second feature of each sub-historical case can be converted into a second-type curve under the target field range (i.e., the extracted overlapping volume histogram is represented in the form of a curve). Then, for the same organ at risk, the difference value between the first-type curve and the second-type curve located on the same coordinate axis at each same coordinate value is calculated, and then the difference values ​​at all coordinate values ​​are accumulated to obtain the first similarity of each organ at risk under the target field range.

[0081] Exemplarily, the AB-OVH curves corresponding to the historical case set preliminarily screened in the feature plan library (i.e., the sub-historical case set corresponding to each target field range) and the AB-OVH curves corresponding to the medical images are matched one by one, and then the numerical difference at each coordinate point is calculated and accumulated to obtain the similarity, wherein the smaller the difference value, the higher the similarity.

[0082] Optionally, the step of determining a reference case set based on all feature similarities includes: sorting the feature similarities, and based on the sorting results, determining a feature similarity set that is within a preset feature similarity threshold range; characterizing the second feature corresponding to each feature similarity in the feature similarity set as a reference feature; and adding historical cases corresponding to each reference feature to the reference case set.

[0083] In an embodiment of the present invention, the feature similarities of multiple target radiation fields can be combined to screen out the closest historical cases. Specifically, the feature similarities can be sorted, and then based on the sorting results, a feature similarity set within a preset feature similarity threshold range (which can be set in advance according to actual conditions, for example, greater than 90) is determined, and then the second feature corresponding to each feature similarity in the feature similarity set is characterized as a reference feature, and the historical cases corresponding to each reference feature are added to the reference case set, so that historical cases with a high similarity to the target object's case can be screened out, which is convenient for generating or optimizing the radiotherapy plan for the target object.

[0084] Optionally, the determination method also includes: obtaining multiple historical cases and determining the field range corresponding to each historical case; extracting features of the historical cases based on the field range of the historical cases, the delineation information of the target area, and the delineation information of the organs at risk, to obtain the second feature of the historical cases; and building a feature plan library based on all historical cases and the second feature of each historical case.

[0085] In an embodiment of the present invention, it is necessary to first build a feature plan library, specifically: some historical cases can be first obtained, and the field range corresponding to each historical case can be determined, and then the irradiation range of all beams on the image (i.e., the medical image included in the historical case) is determined according to the field angle and the target area delineation information, and then based on the irradiation range, the overlapping volume histogram of the OAR and the target area under different beams is calculated, and the overlapping volume histogram and the beam are simultaneously used as the middle feature (i.e., the second feature) of the feature plan library, thereby completing the construction of the feature plan library.

[0086] In the embodiment of the present invention, the constructed feature plan library can be continuously updated, and the feature plan library does not need to repeatedly use the data training model when continuously updating the data, which is more suitable for use in dynamic scenarios. In addition, this embodiment covers more information at the feature extraction level. In dynamic scenarios, only the new data needs to be feature extracted and added to the library to achieve efficient and accurate historical case retrieval without relying on training data, which can save model training time and resources.

[0087] In an embodiment of the present invention, considering the objective differences between images and doses in actual use, an evaluation method is proposed, which can use AB-OVH and DVH (Dose-Volume Histogram) to calculate respectively to return the most similar preset number (such as 10) results, and obtain the dose band range of the corresponding case DVH based on the returned results, and then calculate the similarity of the dose band as an evaluation criterion. In this embodiment, the calculation method can use the Dice coefficient (i.e., a similarity measurement indicator), and the calculation formula is as follows:

[0088] ;

[0089] Wherein, S is the calculated Dice coefficient, X and Y are the DVH dose bands corresponding to the cases returned by the AB-OVH and DVH searches, respectively. In this embodiment, the higher the Dice coefficient, the higher the accuracy of the AB-OVH feature search.

[0090] In some optional embodiments, it can be set that when the Dice coefficient is greater than 80%, AB-OVH is accepted as a feature, when the Dice coefficient is between 70% and 80%, AB-OVH is determined to have reference value as a feature, and when the Dice coefficient is lower than 70%, AB-OVH cannot be used as a feature.

[0091] In some optional embodiments, the dose bands corresponding to the retrieved cases can be visualized for validity verification.

[0092] In an embodiment of the present invention, a cervical cancer dataset may be used for testing, so as to extract OVH features and DVH features based on the cervical cancer dataset, and then compare the DVH dose band (i.e., doseband_OVH) corresponding to the case returned by the OVH feature retrieval with the DVH dose band (i.e., doseband_DVH) corresponding to the case returned by the DVH feature retrieval to obtain the Dice coefficient (e.g., bladder_Dice (bladder coefficient), bone marrow_Dice (bone marrow coefficient), rectum_Dice (rectum coefficient), femoral head l_Dice (left femoral head coefficient), femoral head r_Dice (right femoral head coefficient)).

[0093] Figure 4 is a schematic diagram of an optional bladder_Dice according to an embodiment of the present invention, such as Figure 4 As shown in the figure, with dose Dose as the horizontal axis and volume vol (volume) as the vertical axis, the DVH dose band corresponding to the cases returned by OVH feature retrieval (i.e., the area formed by the two curves of the overlapping volume histogram (doseband_OVH) representing the dose range) and the DVH dose band corresponding to the cases returned by DVH feature retrieval (i.e., the area formed by the two curves of the dose-volume histogram (doseband_DVH) representing the dose range) are respectively shown, and the bladder coefficient (bladder_Dice) is calculated based on these two DVH dose bands to be 0.8815.

[0094] Figure 5 is a schematic diagram of an optional bone marrow_Dice according to an embodiment of the present invention, such as Figure 5As shown in the figure, with dose Dose as the horizontal axis and volume vol (volume) as the vertical axis, the DVH dose band corresponding to the cases returned by OVH feature retrieval (i.e., the area formed by the two curves of the overlapping volume histogram (doseband_OVH) representing the dose range) and the DVH dose band corresponding to the cases returned by DVH feature retrieval (i.e., the area formed by the two curves of the dose-volume histogram (doseband_DVH) representing the dose range) are respectively shown, and the bone marrow coefficient (bone marrow_Dice) is calculated based on these two DVH dose bands to be 0.8276.

[0095] Figure 6 is a schematic diagram of an optional rectum_Dice according to an embodiment of the present invention, such as Figure 6 As shown in the figure, with dose Dose as the horizontal axis and volume vol (volume) as the vertical axis, the DVH dose band corresponding to the cases returned by OVH feature retrieval (i.e. the area formed by the two curves of the overlapping volume histogram (doseband_OVH) representing the dose range) and the DVH dose band corresponding to the cases returned by DVH feature retrieval (i.e. the area formed by the two curves of the dose-volume histogram (doseband_DVH) representing the dose range) are respectively shown, and the rectum coefficient (rectum_Dice) is calculated based on these two DVH dose bands to be 0.9036.

[0096] Figure 7 is a schematic diagram of an optional femoral head l_Dice according to an embodiment of the present invention Figure 1 ,like Figure 7 As shown in the figure, with dose Dose as the horizontal axis and volume vol (volume) as the vertical axis, the DVH dose band corresponding to the cases returned by OVH feature retrieval (i.e., the area formed by the two curves of the overlapping volume histogram (doseband_OVH) representing the dose range) and the DVH dose band corresponding to the cases returned by DVH feature retrieval (i.e., the area formed by the two curves of the dose-volume histogram (doseband_DVH) representing the dose range) are respectively shown, and based on the calculation of these two DVH dose bands, the left femoral head coefficient (femoral head l_Dice) is 0.763.

[0097] Figure 8 is a schematic diagram of an optional femoral head r_Dice according to an embodiment of the present invention Figure 2 ,like Figure 8As shown in the figure, with dose Dose as the horizontal axis and volume vol (volume) as the vertical axis, the DVH dose band corresponding to the case returned by OVH feature retrieval (i.e. the area formed by the two curves of the overlapping volume histogram (doseband_OVH) representing the dose range) and the DVH dose band corresponding to the case returned by DVH feature retrieval (i.e. the area formed by the two curves of the dose-volume histogram (doseband_DVH) representing the dose range) are respectively shown, and based on the calculation of these two DVH dose bands, the right femoral head coefficient (femoral head r_Dice) is 0.717.

[0098] based on Figures 4 to 8 It can be seen that AB-OVH, as a new feature calculation method, has a very good effect in retrieving and returning DVH of historical cases, and can have excellent performance on most OARs.

[0099] Another optional specific implementation is described in detail below.

[0100] In an embodiment of the present invention, a method for optimizing radiotherapy plans based on a knowledge plan library of field and image features is proposed, which supplements the angle relationship information between the organ and the target area, and can be accurately located in a low-dimensional form after being combined with the distance information. At the same time, it is also possible to effectively simulate the actual principle of the field and the process of acting on the dose distribution in a way that the field and the organ overlap, and the field information is highly interpretably integrated, thereby improving the retrieval accuracy and facilitating the optimization of the radiotherapy plan.

[0101] Fig. 9 is a schematic diagram of an optional most similar historical case retrieval process according to an embodiment of the present invention, such as Fig. 9 As shown, AB-OVH can be generated for each historical case in the historical case dataset, and then the historical case and the corresponding AB-OVH can be recorded in the historical case AB-OVH dataset. After that, AB-OVH can be generated for each dimensional image in the target case, and AB-OVHs can be extracted after fusion of AB-OVHs of multiple dimensional images. After that, the similarity between the AB-OVHs of the target case and each AB-OVH in the historical case AB-OVH dataset can be calculated (i.e., the statistical difference is calculated), and finally the most similar historical case is returned to optimize the radiotherapy plan.

[0102] exist Fig. 9In , the specific process of generating AB-OVH is as follows: the field information (including the number and angle of fields) can be read first, and then the part of field 1 irradiating the OAR can be determined according to the mask corresponding to field 1 (that is, the irradiation range formed when field 1 irradiates the target area mask), and then the detailed distance information of the irradiated OAR part is extracted, and the distance distribution is converted into an AB-OVH statistic, and the field needs to be looped to the three coordinate systems to complete the calculation to obtain the AB-OVH of the field on the three-dimensional image; similarly, the above calculation can also be performed according to the mask corresponding to field 1, ..., and the mask corresponding to field n to generate the AB-OVH of all fields on the three-dimensional image.

[0103] In the embodiment of the present invention, by fusing the radiation field information and the multi-dimensional angle information, more accurate features can be extracted, thereby effectively improving the accuracy of case retrieval, and more accurate historical cases can be obtained for radiotherapy plan optimization.

[0104] The following is a detailed description in conjunction with another embodiment.

[0105] Embodiment 2

[0106] A historical case determination device provided in this embodiment includes multiple implementation units, each implementation unit corresponds to each implementation step in the above-mentioned embodiment 1.

[0107] Fig.10 is a schematic diagram of an optional historical case determination device according to an embodiment of the present invention, such as Fig.10 As shown, the determining device may include: an extracting unit 1000 and a determining unit 1001.

[0108] The extraction unit 1000 is used to obtain an N-dimensional medical image of the target object, and perform feature extraction on the medical image in each dimension to obtain a first feature of the medical image, wherein the first feature of the medical image includes a relationship between an overlapping area and a distance range; the overlapping area is an overlapping area between an organ at risk on the medical image and a target area on the medical image under each radiation source; the distance range is a distance range composed of the distance between the organ at risk and the radiation source; N is a positive integer;

[0109] Determination unit 1001 is used to determine a reference case set of the target object from a feature plan library based on the first feature of the medical image in each dimension, wherein the feature plan library includes: multiple historical cases; the reference case set includes: at least one reference case; the reference case is used to provide a reference when determining a radiotherapy plan for the target object.

[0110] The above-mentioned determination device can obtain the N-dimensional medical image of the target object through the extraction unit 1000, and perform feature extraction on the medical image in each dimension to obtain the first feature of the medical image, and determine the reference case set of the target object from the feature plan library based on the first feature of the medical image in each dimension through the determination unit 1001. In an embodiment of the present invention, the features of the multidimensional medical image of the target object are extracted by combining the field angle and the number of fields, and then the closest reference case set is retrieved from the feature plan library based on the extracted features for subsequent radiotherapy plan optimization. In this way, the extracted features integrate the field information and the dimensional information, so that the most similar historical cases can be searched efficiently and accurately according to the extracted features, thereby solving the technical problem that the most similar historical cases cannot be accurately retrieved in the related technology, resulting in the low accuracy of the generated radiotherapy plan, and achieving the technical effect of accurately retrieving the most similar historical cases to facilitate the subsequent optimization of the radiotherapy plan.

[0111] Optionally, the extraction unit includes: a first determination module, used to determine the irradiation range of each field source based on the target field range of the medical image and the delineation information of the target area; a second determination module, used to determine the overlapping area of ​​each organ at risk and the target area under each field source based on the irradiation range and the delineation information of each organ at risk; a third determination module, used to determine the sub-features of each organ at risk under each field source based on the relationship between the overlapping area and the distance range of the organ at risk from each field source; and a first construction module, used to construct the first feature of the medical image under the target field range based on the sub-features of each organ at risk under each field source.

[0112] Optionally, the determination unit includes: a fourth determination module, used to screen an initial historical case set belonging to the target cancer type from the feature plan library, and determine a sub-historical case set corresponding to each target field range from the initial historical case set, wherein the target cancer type is the cancer type determined for the target object; the target field range is the field range corresponding to the target cancer type; a first calculation module, used to calculate the feature similarity between the first feature and the second feature corresponding to each sub-historical case under the same target field range; and a fifth determination module, used to determine a reference case set based on all feature similarities.

[0113] Optionally, the first calculation module includes: a first determination submodule, used to determine the weight of each organ at risk under each target field range based on the overlapping area and distance between the organ at risk and the target area on the medical image; a first calculation submodule, used to calculate the first similarity of each organ at risk under the target field range based on the first feature corresponding to the same target field range and the second feature of each sub-historical case; a first processing submodule, used to perform weighted processing on the first similarities of all organs at risk according to the weight of each organ at risk to obtain the second similarity under the target field range; a first accumulation submodule, used to accumulate the second similarity of each target field range on the medical image in each dimension to obtain the feature similarity under each target field range.

[0114] Optionally, the first calculation submodule includes: a first conversion submodule, used to convert the sub-feature of each organ at risk in the first feature into a first type of curve within the target field range; a second conversion submodule, used to convert the sub-feature of each organ at risk in the second feature of each sub-historical case into a second type of curve within the target field range; a second calculation submodule, used to calculate, for the same organ at risk, the difference between the first type of curve and the second type of curve located on the same coordinate axis at each same coordinate value; and a second accumulation submodule, used to accumulate the difference values ​​at all coordinate values ​​to obtain the first similarity of each organ at risk within the target field range.

[0115] Optionally, the fifth determination module includes: a first sorting submodule, used to sort the feature similarities, and based on the sorting results, determine a feature similarity set within a preset feature similarity threshold range; a first characterization submodule, used to characterize the second feature corresponding to each feature similarity in the feature similarity set as a reference feature; and a first adding submodule, used to add historical cases corresponding to each reference feature to the reference case set.

[0116] Optionally, the determination device also includes: a sixth determination module, used to obtain multiple historical cases and determine the field range corresponding to each historical case; a first extraction module, used to extract features of historical cases based on the field range of historical cases, target area delineation information, and endangered organ delineation information to obtain a second feature of the historical cases; a first building module, used to build a feature plan library based on all historical cases and the second feature of each historical case.

[0117] The above-mentioned determination device may also include a processor and a memory. The above-mentioned extraction unit 1000, determination unit 1001, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to implement corresponding functions.

[0118] The processor includes a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the reference case set of the target object is determined from the feature plan library based on the first feature of the medical image in each dimension by adjusting the kernel parameters.

[0119] The above-mentioned memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one storage chip.

[0120] The present invention also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining an N-dimensional medical image of a target object, and performing feature extraction on the medical image in each dimension to obtain a first feature of the medical image, and based on the first feature of the medical image in each dimension, determining a reference case set of the target object from a feature plan library.

[0121] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, the computer-readable storage medium including a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned method for determining historical cases.

[0122] According to another aspect of an embodiment of the present invention, there is also provided an electronic device, comprising one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by one or more processors, the one or more processors implement the above-mentioned method for determining historical cases.

[0123] Fig.11 is a hardware structure block diagram of an electronic device (or mobile device) for a method for determining historical cases according to an embodiment of the present invention. Fig.11 As shown, the electronic device may include one or more processors (e.g., Fig.11 The processor 1102a, processor 1102b, ..., processor 1102n, etc., which may include but are not limited to a microprocessor MCU or a programmable logic device FPGA, etc.), a memory 1104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a keyboard, a power supply and / or a camera. It can be understood by those skilled in the art that Fig.11 The structure shown is only for illustration and does not limit the structure of the above electronic device. Fig.11More or fewer components as shown, or with Fig.11 Different configurations are shown.

[0124] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0125] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0126] In the several embodiments provided by the present invention, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0127] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0128] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0129] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc. Various media that can store program codes.

[0130] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for determining historical cases, characterized in that: include: Acquire an N-dimensional medical image of the target object, and perform feature extraction on the medical image in each dimension to obtain the first feature of the medical image, wherein the first feature of the medical image includes the relationship between the overlapping area and the distance range; the overlapping area is the overlapping area between the organ at risk on the medical image and the target area on the medical image under each field source; the distance range is the distance range composed of the distance between the organ at risk and the field source; N is a positive integer; the steps of performing feature extraction on the medical image in each dimension to obtain the first feature of the medical image include: determining the irradiation range of each field source based on the target field range of the medical image and the delineation information of the target area; determining the overlapping area between each organ at risk and the target area under each field source based on the irradiation range and the delineation information of each organ at risk; determining the sub-feature of each organ at risk under each field source based on the relationship between the overlapping area and the distance range between the organ at risk and each field source; constructing the first feature of the medical image under the target field range based on the sub-feature of each organ at risk under each field source; Based on the first feature of the medical image in each dimension, determining a reference case set of the target object from a feature plan library, wherein the feature plan library includes: a plurality of historical cases; the reference case set includes: at least one reference case; the reference case is used to provide a reference when determining a radiotherapy plan for the target object; Determining a reference case set for the target object from a feature plan library includes: screening an initial historical case set belonging to a target cancer type from the feature plan library, and determining a sub-historical case set corresponding to each target field range from the initial historical case set, wherein the target cancer type is the cancer type determined for the target object; the target field range is the field range corresponding to the target cancer type; calculating the feature similarity between the first feature and the second feature corresponding to each sub-historical case under the same target field range, wherein the second feature is obtained based on the field range of the historical case, the delineation information of the target area, and the delineation information of the organs at risk; and determining the reference case set based on all the feature similarities.

2. The determination method according to claim 1, characterized in that: The step of calculating the feature similarity between the first feature and the second feature corresponding to each sub-historical case within the same target field range comprises: Determining the weight of each of the organs at risk within each of the target radiation fields based on the overlapping area and distance between the organs at risk and the target area on the medical image; Calculate a first similarity of each of the organs at risk within the target radiation field based on the first feature corresponding to the same target radiation field and the second feature of each of the sub-historical cases; According to the weight of each of the organs at risk, weighted processing is performed on the first similarities of all the organs at risk to obtain a second similarity within the target radiation field; The second similarities of each target field range on the medical image in each dimension are accumulated to obtain the feature similarity of each target field range.

3. The determination method according to claim 2, characterized in that: The step of calculating the first similarity of each of the organs at risk within the target radiation field based on the first feature corresponding to the same target radiation field and the second feature of each of the sub-historical cases comprises: Converting each of the sub-features of the organ at risk in the first feature into a first type of curve within the target field range; converting each of the sub-features of the organ at risk in the second feature of each of the sub-historical cases into a second type of curve within the target field range; For the same organ at risk, calculating the difference value between the first type of curve and the second type of curve located on the same coordinate axis at each same coordinate value; The difference values ​​under all the coordinate values ​​are accumulated to obtain the first similarity of each of the organs at risk within the target field range.

4. The determination method according to claim 1, characterized in that: The step of determining the reference case set based on all the feature similarities comprises: Sorting the feature similarities, and determining a feature similarity set within a preset feature similarity threshold range based on the sorting result; Characterize the second feature corresponding to each feature similarity in the feature similarity set as a reference feature; The historical cases corresponding to each of the reference features are added to the reference case set.

5. The determination method according to claim 1, characterized in that: The determination method further comprises: Acquire multiple historical cases, and determine the radiation field range corresponding to each of the historical cases; Based on the radiation field range, target area delineation information, and organ-at-risk delineation information of the historical case, feature extraction is performed on the historical case to obtain a second feature of the historical case; The feature planning library is constructed based on all the historical cases and the second feature of each of the historical cases.

6. A device for determining historical cases, characterized in that: include: An extraction unit is used to obtain an N-dimensional medical image of a target object, and perform feature extraction on the medical image in each dimension to obtain a first feature of the medical image, wherein the first feature of the medical image includes a relationship between an overlapping area and a distance range; the overlapping area is an overlapping area between an organ at risk on the medical image and a target area on the medical image under each radiation source; the distance range is a distance range composed of the distance between the organ at risk and the radiation source; N is a positive integer; A determination unit, configured to determine a reference case set of the target object from a feature plan library based on the first feature of the medical image in each dimension, wherein the feature plan library includes: a plurality of historical cases; the reference case set includes: at least one reference case; the reference case is used to provide a reference when determining a radiotherapy plan for the target object; The extraction unit includes: a first determination module, used to determine the irradiation range of each of the field sources based on the target field range of the medical image and the delineation information of the target area; a second determination module, used to determine the overlapping area of ​​each of the organs at risk and the target area under each of the field sources based on the irradiation range and the delineation information of each of the organs at risk; a third determination module, used to determine the sub-feature of each of the organs at risk under each of the field sources based on the relationship between the overlapping area and the distance range of the organs at risk from each of the field sources; a first construction module, used to construct the first feature of the medical image under the target field range based on the sub-feature of each of the organs at risk under each of the field sources; The determination unit includes: a fourth determination module, used to screen an initial historical case set belonging to the target cancer type from the feature plan library, and determine a sub-historical case set corresponding to each target field range from the initial historical case set, wherein the target cancer type is the cancer type determined for the target object; the target field range is the field range corresponding to the target cancer type; a first calculation module, used to calculate the feature similarity between the first feature and the second feature corresponding to each sub-historical case under the same target field range, wherein the second feature is obtained based on the field range of the historical case, the delineation information of the target area and the delineation information of the organs at risk; a fifth determination module, used to determine the reference case set based on all the feature similarities.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for determining historical cases described in any one of claims 1 to 5.

8. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the method for determining historical cases described in any one of claims 1 to 5.

Images