Brain image meningioma full-automatic screening and tumor segmentation method and device

Abstract

The invention provides a brain image meningioma full-automatic screening and tumor segmentation method and device. The brain image meningioma full-automatic screening and tumor segmentation method comprises the following steps of S1, performing format conversion on an image file of a meningioma patient; s2, preprocessing the image of the image file after format conversion; s3, carrying out enhancement processing on the image data in the preprocessed image file; s4, screening the enhanced image data to obtain a slice image containing meningioma, and storing the slice image; s5, segmenting the slice image containing the meningioma to obtain a meningioma segmentation result image, and storing the meningioma segmentation result image; and S6, performing post-processing on the slice image containing the meningioma and the meningioma segmentation result image. According to the method, classification and segmentation technologies of deep learning are combined, the method has the advantages of being noninvasive, repeatable, batched and capable of improving clinical diagnosis efficiency of doctors, and assistance can be provided for meningioma imaging omics research and clinical doctor operation scheme making.

Description

technical field [0001] The present invention relates to the technical field of medical image processing, in particular to a method and device for fully automatic screening of brain imaging meningiomas and tumor segmentation, especially to a fully automatic screening and tumor segmentation of brain imaging meningiomas based on deep learning. Segmentation method and device. Background technique [0002] Meningiomas are the most common primary intracranial tumors in adults. The incidence rate of meningioma is 8.03 per 100,000 population, accounting for more than one-third of primary central nervous system (CNS) tumors, making it the most common primary intracranial tumor. Although most meningiomas have an excellent prognosis, they often recur because complete resection remains difficult. Meningiomas appear as widespread dural hemispheric or oval lesions attached to the dura mater. Although meningiomas are generally easy to diagnose, there are also histological variants with ...

AI Technical Summary

Problems solved by technology

[0006] Regarding the above-mentioned related technologies, the inventor believes that in the practical application of the above-mentioned method, the complex clinical environment makes it difficult for many algorithms to achieve good results, there are many processes that require manual intervention, and the workload of doctors is heavy. Various methods are also difficult to adapt to different types of brain images, resulting in large errors in the segmentation results

Images