A qualitative method for fast segmentation of meningioma based on deep neural network

Abstract

The present invention relates to the field of meningioma grade judgment before operation, and discloses a method for fast segmentation and qualitative meningioma based on deep neural network, comprising the following steps: preparing magnetic resonance brain images; establishing a meningioma segmentation model, and performing meningioma segmentation The model screens effective images containing meningioma regions from magnetic resonance brain images; establishes a meningioma grading model, and uses the meningioma grading model to perform grading detection on effective images, and outputs the meningioma grading detection results. Through the above-mentioned method proposed by the present invention, it is only necessary to input a series of magnetic resonance brain images generated by a certain patient into the network, and after comprehensive analysis and calculation of all magnetic resonance brain images, the patient's meningioma is quickly given. The grading results can help doctors to diagnose, and the whole process is automatic, which reduces a lot of repetitive work for doctors, saves time, and enables patients to receive treatment faster.

Description

technical field [0001] The present invention relates to the field of meningioma grade judgment before operation, in particular to a method for fast segmentation and qualitative meningioma based on deep neural network. Background technique [0002] Meningioma originates from the meninges and the derivatives of the meningeal space, and is the second most common intracranial tumor, accounting for 13% to 26% of intracranial tumors, and its incidence has been on the rise in recent years. According to the 2016 WHO classification of tumors of the central nervous system, meningiomas are divided into three grades. Most meningiomas are classified as WHO grade I lesions, that is, benign lesions with slow growth and less recurrence after surgery. A small number of lesions are classified as WHO grade II or III lesions according to local invasiveness and atypical cell characteristics, that is, malignant lesions. Patients may suffer from blindness, hemiplegia, epilepsy and other symptoms. ...

AI Technical Summary

Problems solved by technology

[0004] Although most hospitals have MRI scanning machines, what is more important is the diagnosis by professional doctors. For relatively remote rural hospitals, it is difficult to obtain clinical opinions from experienced doctors in a timely manner.

At the same time, due to the very small intervals of tomographic scans, there are usually 200-300 brain magnetic tomographic scan slices of a patient, which is a huge number. It is a huge workload, time-consuming, and efficient to rely entirely on doctors' manual image reading and diagnosis. lower

Images