Multi-scale feature extraction-based segmentation model establishment method and device, and segmentation method and device

Abstract

The invention belongs to the technical field of semantic segmentation, and discloses amulti-scale feature extraction-based segmentation model establishment method and device and a segmentation method and device. The device comprises: a data acquisition and preprocessing module used for acquiring an original abdomen 3D CT image set, preprocessing the original abdomen 3D CT image set to obtain an abdomen 3D CT image set, and labeling to obtain a label set; the model building module that is used for building a 3D U-Net model, wherein the 3D U-Net model comprises an encoder and a decoder, each of the encoder and the decoder comprises a plurality of levels from low to high, jumping connection exists between the encoder of each layer and the same level and between the decoder of all high levels, and jumping connection exists between the decoder of each layer and the decoder of all high levels; the model training module that is used for training and taking a trained model as a segmentation model; the segmentation module that is used for obtaining a segmentation result of the original abdomen 3D CT image to be segmented. According to the invention, jump connection is introduced between encoders and decoders of different levels, so that information transmission paths in the network are increased, the problem that features of different scales are difficult to extract from the original U-Net at the same time is solved, and the segmentation accuracy is improved.

Description

technical field [0001] The invention belongs to the technical field of semantic segmentation, and in particular relates to a segmentation model establishment and segmentation method and device based on multi-scale feature extraction. Background technique [0002] In recent years, with the continuous maturity of image processing and deep learning technology, computer-aided diagnosis based on artificial intelligence technology can help pathologists make more objective and effective diagnoses. Deep convolutional neural network can implicitly and automatically learn medical image features directly from data samples, and its learning process is essentially a solution process of an optimization problem. Through learning, the model selects the correct features from the training data, enabling it to make correct decisions when tested on new data. [0003] Although it is common to apply deep learning to medical imaging problems, research on gastrointestinal tumors is still scarce. ...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a segmentation model establishment, segmentation method and device based on multi-scale feature extraction to solve the problem of poor segmentation effect on small tumors in the prior art

Images