Guide wire segmentation method and device based on video time sequence, and readable medium

Abstract

The invention relates to a guide wire segmentation method and device based on a video time sequence, and a readable medium. The method comprises the following steps: obtaining image frames with different time steps of a time sequence image; carrying out feature extraction on the image frames of different time steps, and fusing features; determining time sequence information of the image frames through the fused features, and generating a first feature map; performing up-sampling processing on the first feature map, and fusing information of different depths in the previous step to obtain a second feature map; and activating the second feature map to obtain a segmentation result. The method has the beneficial effects that the operation speed of the guide wire segmentation network is realized, and meanwhile, the guide wire segmentation accuracy is ensured.

Description

technical field [0001] The invention relates to the field of computer images, in particular to a video time series-based guide wire segmentation method, device and readable medium. Background technique [0002] The development of computer technology has promoted the development of robotic intervention systems. In recent years, cardiovascular robotic interventional systems, as a research hotspot in the field of robotic interventional systems, have attracted extensive attention from scholars at home and abroad. The segmentation of the guide wire is a key step in the cardiovascular robotic interventional system, so the accurate segmentation of the guide wire is very important for a cardiovascular robotic interventional system. [0003] Guidewire segmentation is a challenging task for several reasons: [0004] (1) Extreme foreground-background class imbalance caused by low ratio of guidewire pixels to background pixels; [0005] (2) The existence of linear structures such as ...

AI Technical Summary

Problems solved by technology

[0006] (3) The signal-to-noise ratio of X-ray images is low, and the background noise greatly interferes with the segmentation of catheters

The currently proposed guidewire segmentation model is mainly based on the guidewire information of a single frame, and does not make good use of the timing information in the guidewire sequence

Images