Enhancing Medical Image Quality using Neutrosophic Fuzzy Domain and Multi-Level Enhancement Transforms: A Comparative Study for Leukemia Detection and Classification

Abstract

 Medical image processing has become a critical research area due to the vast amounts of 
digital image data available. However, medical images often suffer from poor illumination and low 
visibility of significant structures, requiring image enhancement to improve image quality before 
processing. In this paper, we propose a technique for enhancing medical images by removing noise 
and improving contrast based on three different enhancing transforms. The proposed technique 
embeds the image into a neutrosophic fuzzy domain, where it is mapped into three different levels 
of trueness, falseness, and indeterminacy, and each level is processed individually using the 
enhancement transforms. We compare the proposed technique with four other systems for leukemia 
detection and classification using accuracy and T, I, and F values. The proposed system performs 
the best with an accuracy of 98%, outperforming the other systems in terms of accuracy, degree of 
indeterminacy, and falsity. The proposed system uses different algorithms and filters to process 
images and extract features like color and texture. The system's classification uses k-means for 
segmentation and SVM for classification. The paper highlights the importance of considering T, I, 
and F values in evaluating the performance of different systems for leukemia detection and 
classification, providing a more accurate representation of the uncertainty and ambiguity involved 
in the evaluation process.  

DOI: 10.5281/zenodo.10780568