3D reconstruction of brain tumors from 2D MRI scans: An improved marching cube algorithm

Abstract

MRI is a popular imaging technique in medicine, particularly for the diagnosis of brain malignancies. Like any other acquisition approach, it has several drawbacks in addition to its many advantages. One of these is the problem of noise, which can provide extraneous and incorrect information that can fool the human eye. Methods such as MRI and CT, produce 2D images of the body's internal organs. This research aims with the 3D reconstruction of brain tumor, as it is very crucial for surgical planning, and biological study because 2D images can never capture a tumor's true appearance. However, it is extremely difficult to visualize the tumors in MRIs due to the different complexity of tumors. This work develops an Improved Marching Cube Algorithm (IMCA) based 3D reconstruction of brain tumors under 2 phases. Phase 1 will include MF-based preprocessing, segmentation based on Modified BIRCH (M-BIRCH) and hybrid classification combining optimized CNN and optimized Bi-LSTM. The suggested Self Adaptive Chimp Optimization Algorithm (SA-CHOA) will optimize the weight of CNN and Bi-LSTM. Phase 2 involves IMC (Improved Marching Cube) based 3D reconstruction. The validation outcomes demonstrate the suggested model's superiority to other established models for 3D reconstruction.