CT rib segmentation method and device

Abstract

The invention provides a CT rib segmentation method and device, and the method comprises the steps: S1, obtaining training data, and generating two types of labels according to the training data; S2,training a full convolution image semantic segmentation model of two tasks according to the two types of labels to obtain a rib segmentation model; S3, acquiring to-be-segmented CT data, wherein the to-be-segmented CT data comprises all layers in CT; S4, reasoning a two-dimensional segmentation result and an adjacent layer relationship on each layer in the CT data to be segmented by using the trained rib segmentation model to acquire a rib contour of each layer by using a connected domain detection algorithm based on two-dimensional segmentation; S5, combining the rib contours of all layers according to the adjacent layer relationship to obtain a three-dimensional segmentation result; and S6, obtaining a CT rib segmentation result of the to-be-segmented CT data by using a post-processing algorithm. Compared with a traditional algorithm designed in a manual heuristic mode, the method is based on mass data and deep machine learning, and therefore the invention has the advantages in robustness, expandability and the like.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a CT rib segmentation method and device. Background technique [0002] In recent years, deep machine learning has been widely used in the field of image understanding. Among them, the deep full convolutional network proposed for image semantic segmentation has obvious advantages in terms of segmentation accuracy compared with traditional algorithms, and it can better control the time required for inference. In addition, the widespread use of GPUs further improves the inference speed of fully convolutional networks. This makes it possible to apply high-precision fully convolutional networks in medical imaging scenarios. On the other hand, traditional medical imaging diagnosis relies on the empirical and subjective judgment of clinicians, so there are problems such as time-consuming and poor stability, which has gradually become a bottleneck restricting the development of moder...

AI Technical Summary

Problems solved by technology

Such methods often rely on the color or brightness difference between the rib area and the background area, artificially designed features or machine learning models trained with a small amount of data, which is difficult to guarantee robustness and limits its application range

In recent years, although the general-purpose image semantic segmentation method based on deep machine learning has been applied in the field of medical imaging for many times, in terms of rib segmentation, there has been little targeted optimization or improvement, and there are high production costs and poor prediction accuracy. low level problem

Images