Mitochondrial Network Segmentation from Fluorescence Microscopy Images using VGG-UNet

Automatic medical image examination plays a vital part in disease screening and decision making processes. This research aims to create deep-learning (DL) assisted Mitochondrial Network (MN) segmentation from the chosen Fluorescence Microscopy Images (FMI) with better accuracy. Several parts of this arrangement encompasses the following; (i) Image collection and resizing, (ii) extraction of the MN section using a chosen scheme, (iii) comparing the segmented section against the ground-truth (GT) and computing the performance metrics, and (iv) performance evaluation and validation of the proposed technique. In this work, a comparative analysis is presented using the UNet and its variants. The experimental findings affirm that the VGG-UNet-supported scheme outperforms alternatives, yielding an overall segmentation accuracy of >97%. This underscores the efficacy of the proposed methodology in advancing the automated examination of MN for the chosen FMI database.