Dual path parallel hierarchical diagnosis model for intracranial tumors based on multi-feature entropy weight

Lingling Fang, Yumeng Jiang

Abstract: Highlights•According to the pathological features of intracranial tumors, the features of the solid tumor mass and intratumoral necrosis are extracted.•Multi-feature entropy weight is used to analyze different feature weights, optimize MRI data sets, and maximize feature learning.•ME-DPNet is used the dual path structure to achieve multi-modal MRI image input. The information on different modalities is fully used to analyze tumor features from multiple angles. Besides, multiple features are superimposed and integrated to achieve multi-feature fusion.•The proposed model is trained with the intracranial tumor images obtained from hospital clinics to minimize the medical cost and simulate the clinical diagnosis environment to the great extent. Accurate automatic grading of intracranial tumors is achieved using multi-modal MRI images.