Lung medical image segmentation method and device based on improved U-Net

Abstract

The invention discloses a lung medical image segmentation method and device based on improved U-Net. The method comprises the following steps: carrying out normalization and threshold method binarization processing on a pre-acquired original CT image; adopting a bottleneck residual module to optimize a U-Net down-sampling part so as to construct a U-Net optimization network; adopting a Dice loss function for the U-Net optimization network; and training an original U-Net network and a U-Net optimization network by adopting an NADAM optimization algorithm, and measuring segmentation accuracy by adopting an average intersection-to-union ratio MIoU index. According to the method and device, a residual block structure is adopted to improve a traditional U-Net network structure, the convergence speed is effectively increased, and the accuracy is improved; meanwhile, a Dice loss function is optimized and used to judge the difference degree between the estimated value and the real value; according to the method and device, the image detail quality can be effectively improved, two lung boundaries, small voids and blood vessels around all levels of bronchus in the lung are more accurate, the research of image segmentation can be guided, and the leading-edge technology can be further put into clinical application.

Description

technical field [0001] The invention belongs to the technical field of medical image segmentation, and in particular relates to a lung medical image segmentation method and device based on improved U-Net. Background technique [0002] The novel coronavirus epidemic is spreading all over the world. There are a large number of asymptomatic and mildly infected people with a long latent period. It is difficult to judge suspected cases. The analysis of lung imaging can serve as an important auxiliary means. High-resolution medical images can clearly show the conditions of lung lobes, trachea and other tissue parts, and are the best standard for clinical research, disease diagnosis and functional testing. [0003] Semantic image segmentation, that is, the classification and information labeling of pixel level, so as to interpret the global content. In the medical field, it can provide information on human tissue parts for tumors, fetal brain detection, etc. It has considerable ...

AI Technical Summary

Problems solved by technology

[0007] Purpose of the invention: The present invention proposes a lung medical image segmentation method and device based on improved U-Net for the problems of blurred boundaries and loss of non-connected regions in the current lung CT image segmentation algorithm

Images