Enhanced 3D segmentation techniques for reconstructed 3D medical volumes: Robust and Accurate Intelligent System

TLDR

Medical imaging from modalities such as PET, CT, and MRI is essential for diagnosis, and automated 2D segmentation methods like FCM, k‑means, MRFM, and neural networks are commonly used to extract regions of interest slice by slice. This study proposes a 3D FCM segmentation model to improve the segmentation process by treating the entire 3D volume as a single testing dataset. The proposed method applies fuzzy c‑means clustering directly to the 3D volume, enabling simultaneous segmentation across all slices rather than independent 2D processing.

Abstract

Medical images play an important role in treating a large number of ailments as they are integral and even indispensable to the diagnosis process of such ailments. Medical images come from different acquisition systems (such as PET, CT, MRI) and, in many situations, automated processing of these images can greatly aid physicians and make their jobs easier. In medical imaging and its applications, 2D segmentation (with its different approaches such as FCM, k-means, MRFM and NN) is the first step which is used to extract ROI. This helps in extracting ROI in each slice (2D medical image) separately regardless of its relation to the next and the previous slices. In this paper, a 3D model of FCM segmentation techniques is proposed to enhance the segmentation process and take in mind the overall 3D-Volume as one testing data.