Temporal lobe sketching method and sketching system based on multi-modal image, computing device and storage medium

Abstract

The invention discloses a temporal lobe sketching method and sketching system based on a multi-modal image, a computing device and a storage medium, and the method comprises the steps: collecting a preset number of CT brain images and T1 MRI brain images, and making a temporal lobe segmentation label of each T1 MRI brain image; performing preprocessing, registration transformation and image cutting on the CT brain image, the T1 MRI brain image and the temporal lobe segmentation label; taking the cut CT brain image and the T1 MRI brain image as network input, taking the cut temporal lobe segmentation tag as network target output, training in a semantic segmentation convolutional neural network until the model is stably converged, stopping training, and obtaining an optimal temporal lobe segmentation neural network model; and inputting an image to be segmented into the trained temporal lobe segmentation neural network model to obtain a temporal lobe automatic segmentation result of the corresponding CT image, and performing post-processing and edge detection on the temporal lobe automatic segmentation result to obtain a temporal lobe sketching result.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a temporal lobe delineation method based on multimodal images, a delineation system, a computing device and a storage medium. Background technique [0002] The function of the temporal lobe is mainly responsible for processing auditory information, and the temporal lobe also has a certain relationship with memory and emotion. Therefore, in the clinical diagnosis of radiotherapy and other departments, it is extremely important to accurately segment the temporal lobe. However, in clinical practice based on CT brain images, there is no clear boundary between the temporal lobe and other brain regions on CT brain images, and it is not easy to manually delineate temporal lobe regions on CT brain images or to draw temporal lobe regions based on edge detection algorithms. automatic sketching. In contrast, due to the high resolution of soft tissue imaging in MRI imaging, multi...

AI Technical Summary

Problems solved by technology

[0003] In addition, traditional image segmentation algorithms include segmentation methods based on threshold segmentation, adaptive edge detection, and active contour models. These methods have problems such as poor noise resistance and poor data adaptability for brain organ segmentation. Low performance and robustness in

Images