Three-dimensional medical image segmentation model and training method and application thereof

Abstract

The invention discloses a three-dimensional medical image segmentation model and a training method and application thereof, the model is based on a deep learning convolutional neural network, and the network model is composed of a Prototype calculation module, an encoder decoder module, a task controller module, a dynamic convolution kernel generator module and a dynamic segmentation head module. According to the model, a task controller is designed and a Prototype feature vector is introduced to guide the network to carry out segmentation task discrimination, and different segmentation tasks are flexibly processed by using dynamic convolution, so that the problem of partial marking in the field of medical image segmentation is solved, and the aim of segmenting organs and tumors of multi-category three-dimensional medical images is fulfilled; the problems of excessive parameters and inflexible task expansion in the current mainstream method are solved, and meanwhile, the segmentation result is improved to a certain extent.

Description

technical field [0001] The invention relates to a three-dimensional medical image segmentation model, a training method and application thereof, and belongs to the technical field of image segmentation. Background technique [0002] Automatic segmentation of organs and tumors using computed tomography (CT) is one of the most fundamental but also the most challenging tasks in medical image analysis, and it plays a pivotal role in various computer-aided diagnosis tasks, such as analyzing lesions tasks such as contouring, developing surgical plans, and performing 3D reconstructions. However, due to the limitation of labor cost and expertise, it is difficult to perform voxel-level annotation of multiple organs and tumors in large datasets, so most of the current basic datasets are only used for one type of organ or tumor For the segmentation task, all task-irrelevant organs and tumors are annotated as background. For example, the LiTS dataset only has annotations for liver and...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to propose a three-dimensional medical image segmentation model. By designing a task controller module and introducing the concept of Prototype to guide the network to perform segmentation task discrimination, dynamic convolution is used to flexibly process different segmentation tasks, thereby achieving The purpose of multi-category 3D medical image organ and tumor segmentation solves the problem of too many parameters and inflexible task expansion in the current mainstream methods

Images