Pulmonary nodule auxiliary diagnosis method based on three-dimensional multi-resolution attention capsule network

Abstract

The invention discloses a pulmonary nodule auxiliary diagnosis method based on a three-dimensional multi-resolution attention capsule network, and belongs to the technical field of medical image processing. The method comprises the following steps: obtaining a pulmonary nodule CT image data set containing a pathological type label; preprocessing samples in the data set; constructing a three-dimensional multi-resolution attention capsule network; and inputting the preprocessed data sample into the three-dimensional multi-resolution attention capsule network for training, and improving the prediction capability of the three-dimensional multi-resolution attention capsule network on multiple pathological types of pulmonary nodules by learning sample distribution. According to the method of the invention, manual features do not need to be designed or auxiliary information such as serum biomarkers does not need to be utilized, and high prediction precision and high robustness can still be kept for small samples difficult to classify and unbalanced and multi-label clinical data sets.

Description

technical field [0001] The invention belongs to the field of medical image processing, and in particular relates to a pulmonary nodule auxiliary diagnosis method based on a three-dimensional multi-resolution attention capsule network. Background technique [0002] Lung cancer is the leading cause of cancer-related deaths worldwide, and the use of CT scans to examine high-risk populations is an effective means of detecting early-stage lung cancer. The large number of such populations has dramatically increased the workload of radiologists, so computer-aided Diagnosis plays a very important role. The application of auxiliary diagnostic methods can reduce the dependence on doctors' personal experience and work status, improve the diagnostic efficiency, and facilitate the realization of early screening, early diagnosis and early treatment of lung cancer. Most of the traditional computer-aided diagnosis methods are based on machine learning algorithms. Such methods need to desig...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the problem that traditional assisted diagnosis methods need to design manual features, and even rely on serum biomarker information, and it is difficult for deep convolutional neural networks to exert their advantages on small-scale clinical data sets confirmed by pathological gold standards, the present invention A Pulmonary Nodule Aided Diagnosis Method Based on 3D Multi-Resolution Attention Capsule Network

Images