A Perspective on Improvements in Segmentation Image Processing in Healthcare Datasets
Janjhyam Venkata Naga Ramesh, Rohit Anand, Mohammed Zabeeulla, Abhilash Kumar Saxena, Mukesh Kumar, Nidhi Sindhwani, Ankur Gupta · 2024
The chapter gives an Image Segmentation perspective by UW-Madison which describes a novel method for segmenting organs and cells in medical imaging. The suggested method annotates training data with RLE-encoded masks and works with 16-bit grayscale PNG images. The collection contains several sets of slices of scan, each identified by the date of the scan. Some cases are divided according to time, while others are divided according to case. The purpose is to be able to generalize to both partially and completely unknown scenarios. The publication also contains a wrapper function for easier data exploration and visualization of segmentation masks. The proposed method yields encouraging results and has potential applications in medical picture analysis. Following the acquisition of the mask, post-processing is used to refine and clean the segmentation findings, such as filling gaps in the mask, deleting tiny sections, and smoothing the edges.