3d Reconstruction from 2d Corpus Callosum Segmentation

Corpus Callosum (CC) surgery is required to be performed with thorough and careful preparations. The reason is that the part of the corpus callosum is considered to be crucial. Doctors need to conduct preparations, one of which is simulation. Doctors require 3D visualization to prepare for an operation. There are four classifications of the corpus callosum, namely trunk, splenium, genu, and rostrum. In this case, reconstruction methods with a high level of accuracy are needed to describe the right size and the desired pattern of the corpus callosum. The objective of this research is to develop an automatic method for 3D Reconstruction from 2D Corpus Callosum segmentation of MRI images. We explore and comparative the performance of three approaches that are 3DR2N2, 3DUNet, and 3DMarchingCubes. There are two stages of our proposal that are 2D Corpus Callosum segmentation of MRI image and 3D Reconstruction from segmented 2D Corpus Callosum. In summary, 3D Reconstruction with Marching Cube Algorithm has proven to be successful with an IoU of 0.592 and a loss of 0.091 and has a smaller size than the other two methods.