CN-117173488-B - Classification model determining method and device, electronic equipment and storage medium

CN117173488BCN 117173488 BCN117173488 BCN 117173488BCN-117173488-B

Abstract

The invention discloses a method, a device, electronic equipment and a storage medium for determining a classification model, wherein the method comprises the steps of determining at least one group of sample images to be trained, generating vectors to be used corresponding to the sample images to be trained based on the sample images to be trained and diagnosis and treatment associated information corresponding to the sample images to be trained aiming at each group of sample images to be trained, and carrying out model training on an initial classification model based on the vectors to be used corresponding to a plurality of groups of sample images to be trained to obtain a target classification model. The method and the device realize that a large number of sample images are obtained based on a small number of pre-diagnosis and post-diagnosis images, and corresponding vectors to be used are generated according to diagnosis and treatment information corresponding to the sample images, so that an original classification model is trained based on the vectors to be used, and the effect of a target classification model with more accurate image classification results is obtained.

Inventors

LIU CHANGDONG
XU WENYI
SHAO TAO
ZHOU ZIJIE

Assignees

联仁健康医疗大数据科技股份有限公司

Dates

Publication Date: 20260505
Application Date: 20230920

Claims (9)

1. A method for determining a classification model, comprising: Determining at least one group of sample images to be trained, wherein the sample images to be trained comprise a breast focus image before diagnosis and treatment, a first breast reference image symmetrical to the breast focus image, a breast focus image to be compared obtained by performing image registration on the breast focus image by using a deformation field after diagnosis and treatment, and a second breast reference image obtained by performing image registration on the first breast reference image by using the deformation field, and determining a first breast reference image in the breast image before diagnosis and treatment according to reference position information corresponding to an image area symmetrical to the breast focus image in the breast image before diagnosis and treatment, wherein the first breast reference image is an image which does not contain a breast focus, and the second breast reference image is an image which does not undergo diagnosis and treatment and does not contain the breast focus; Generating a vector to be used corresponding to the sample images to be trained according to the sample images to be trained and diagnosis and treatment associated information corresponding to the sample images to be trained; Model training is carried out on the initial classification model based on the vectors to be used corresponding to the plurality of groups of sample images to be trained, and a target classification model is obtained; The method comprises the steps of generating a vector to be used corresponding to a sample image to be trained and diagnosis and treatment associated information corresponding to the sample image to be trained based on the sample image to be trained, generating a first vector to be used based on the breast focus image and first random diagnosis and treatment information corresponding to the breast focus image, generating a second vector to be used based on the first breast reference image and second random diagnosis and treatment information corresponding to the first breast reference image, generating a third vector to be used based on the second breast reference image and third random diagnosis and treatment information corresponding to the second breast reference image, generating a fourth vector to be used based on the breast focus image to be compared and actual diagnosis and treatment information corresponding to the breast focus image to be compared, and taking the first vector to be used, the second vector to be used, the third vector to be used and the fourth vector to be used as the vectors to be used.
2. The method of claim 1, wherein the determining at least one set of sample images to be trained comprises: Aiming at least one diagnosis and treatment user to be trained, acquiring a pre-diagnosis and treatment breast image and a post-diagnosis and treatment breast image corresponding to a current diagnosis and treatment user according to a user identification of the current diagnosis and treatment user; And respectively carrying out image segmentation on the pre-diagnosis and post-diagnosis breast image to obtain at least one group of sample images to be trained.
3. The method according to claim 2, wherein the image segmentation of the pre-diagnosis breast image and the post-diagnosis breast image respectively to obtain at least one set of sample images to be trained comprises: Determining a breast focus image in the pre-diagnosis breast image aiming at the pre-diagnosis breast image, and determining image position information of the breast focus image in the pre-diagnosis breast image; Determining reference position information corresponding to the image position information from the pre-diagnosis breast image, and determining a first breast reference image corresponding to the breast focus image from the pre-diagnosis breast image according to the reference position information; and taking the breast focus image and the first breast reference image as the sample image to be trained.
4. The method according to claim 2, wherein the image segmentation of the pre-diagnosis breast image and the post-diagnosis breast image respectively to obtain at least one set of sample images to be trained comprises: Determining deformation fields to be used of the post-diagnosis breast image and the pre-diagnosis breast image according to the post-diagnosis breast image; registering the pre-diagnosis breast image and the post-diagnosis breast image based on the deformation field to be used to obtain registered images; and performing image segmentation processing on the registered images to obtain at least one sample image to be trained.
5. The method of claim 4, wherein performing image segmentation on the registered image to obtain at least one sample image to be trained comprises: Determining a first image area corresponding to the breast focus image in the registered image, and taking the first image area as a breast focus image to be compared; Determining a second image area corresponding to the first mammary gland reference image in the registered image, and taking the second image area as a second mammary gland reference image; and taking the focus image to be compared and the second mammary gland reference image as the sample image to be trained.
6. The method as recited in claim 1, further comprising: Acquiring a breast focus image to be identified corresponding to a target diagnosis and treatment user, and generating a vector to be identified based on the breast focus image to be identified and corresponding actual diagnosis and treatment information, wherein the breast focus image to be identified refers to a breast focus image to be compared after diagnosis and treatment of the target diagnosis and treatment user; And carrying out vector analysis on the vector to be identified based on the target classification model to obtain an image classification result corresponding to the breast focus image to be identified, wherein the image classification result comprises complete restoration or incomplete restoration.
7. A classification model determining apparatus, comprising: The system comprises a sample image determining module, a sample image determining module and a processing module, wherein the sample image determining module is used for determining at least one group of sample images to be trained, the sample images to be trained comprise a breast focus image before diagnosis and treatment, a first breast reference image symmetrical to the breast focus image, a breast focus image to be compared which is obtained by performing image registration on the breast focus image by using a deformation field after diagnosis and treatment, and a second breast reference image which is obtained by performing image registration on the first breast reference image by using the deformation field after diagnosis and treatment, and the first breast reference image in the breast image before diagnosis and treatment is an image which does not contain a breast focus and does not contain the breast focus is determined according to reference position information corresponding to an image area symmetrical to the breast focus image in the breast image before diagnosis and treatment; the vector generation module is used for generating a vector to be used corresponding to the sample image to be trained according to the sample image to be trained and diagnosis and treatment associated information corresponding to the sample image to be trained; the model determining module is used for carrying out model training on the initial classification model based on the vectors to be used corresponding to the plurality of groups of sample images to be trained to obtain a target classification model; the vector generation module comprises a first vector determination submodule, a second vector determination submodule, a third vector determination submodule, a fourth vector determination submodule and a vector determination submodule, wherein the first vector determination submodule is used for generating a first vector to be used based on the breast focus image and first random diagnosis and treatment information corresponding to the breast focus image, the second vector determination submodule is used for generating a second vector to be used based on the first breast reference image and second random diagnosis and treatment information corresponding to the first breast reference image, the third vector determination submodule is used for generating a third vector to be used based on the second breast reference image and third random diagnosis and treatment information corresponding to the second breast reference image, the fourth vector determination submodule is used for generating a fourth vector to be used based on the breast focus image to be compared and actual diagnosis and treatment information corresponding to the breast focus image to be compared, and the vector determination submodule is used for taking the first vector to be used, the second vector to be used, the third vector to be used and the fourth vector to be used as the vector to be used.
8. An electronic device, the electronic device comprising: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the method of determining a classification model according to any one of claims 1-6.
9. A computer readable storage medium storing computer instructions for causing a processor to perform the method of determining a classification model according to any one of claims 1-6.

Description

Classification model determining method and device, electronic equipment and storage medium Technical Field The present invention relates to the field of data processing technologies, and in particular, to a method and apparatus for determining a classification model, an electronic device, and a storage medium. Background Currently, when determining the recovery state of a breast cancer area of a user who is in a visit, it is generally necessary to perform a result judgment on the recovery state of the breast by a pathological detection means. However, such a procedure is time consuming and laborious for the receiving user and requires invasive detection of the receiving user. Or the breast cancer area of the user can be detected based on the trained neural network model, but a large number of training samples are needed in the model training process, and in the actual training process, the training samples are not easy to obtain. In order to solve the above problems, the present technical solution proposes a classification model for lesion detection of breast images. Disclosure of Invention The invention provides a method, a device, electronic equipment and a storage medium for determining a classification model, which are used for solving the problems that when the classification model for classifying focus images is used for model training, sample image acquisition is not easy, and the obtained classification model has inaccurate image classification result due to single data dimension of model training. In a first aspect, an embodiment of the present invention provides a method for determining a classification model, including: Determining at least one group of sample images to be trained, wherein the sample images to be trained comprise a breast focus image before diagnosis and treatment, a first breast reference image corresponding to the breast focus image, a breast focus image to be compared corresponding to the breast focus image after diagnosis and treatment, and a second breast reference image corresponding to the breast focus image to be compared, wherein the first breast reference image is an image which does not contain a breast focus, and the second breast reference image is an image which does not go through diagnosis and treatment and does not contain the breast focus; Generating a vector to be used corresponding to the sample images to be trained according to the sample images to be trained and diagnosis and treatment associated information corresponding to the sample images to be trained; And carrying out model training on the initial classification model based on the vectors to be used corresponding to the plurality of groups of sample images to be trained, and obtaining a target classification model. In a second aspect, an embodiment of the present invention further provides a device for determining a classification model, including: The sample image determining module is used for determining at least one group of sample images to be trained, wherein the sample images to be trained comprise a breast focus image before diagnosis and treatment, a first breast reference image corresponding to the breast focus image, a breast focus image to be compared corresponding to the breast focus image after diagnosis and treatment and a second breast reference image corresponding to the breast focus image to be compared, the first breast reference image is an image which does not contain a breast focus, and the second breast reference image is an image which does not go through diagnosis and treatment and does not contain the breast focus; the vector generation module is used for generating a vector to be used corresponding to the sample image to be trained according to the sample image to be trained and diagnosis and treatment associated information corresponding to the sample image to be trained; the model determining module is used for carrying out model training on the initial classification model based on the vectors to be used corresponding to the plurality of groups of sample images to be trained, and obtaining a target classification model. In a third aspect, an embodiment of the present invention further provides an electronic device, including: at least one processor, and A memory communicatively coupled to the at least one processor, wherein, The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the method of determining a classification model according to any of the embodiments of the invention. In a fourth aspect, an embodiment of the present invention further provides a computer readable storage medium, where computer instructions are stored, where the computer instructions are configured to cause a processor to implement a method for determining a classification model according to any embodiment of the present invention. According to the technical scheme, at least one group of sample images to be trained is determi