1Ph.D Research Scholar, Erode Arts and Science College, Erode, Tamilnadu, India. Email: sudhayuva29@gmail.com
2Associate Professor, Erode Arts and Science College, Erode, Tamilnadu, India
Background: Breast cancer is one of the most significant causes of death among women in the world and this creates the need to develop effective computer-aided diagnosis systems that will increase early detection and prognosis. The proposed research will show an integrated methodology that introduces Modified Fuzzy C-Means (M-FCM) segmentation strategy that will be combined with Mathematically Modified Dolphin Swarm Optimization (MDSO) as the best feature selection in the context of breast cancer prediction.
Methodology: M-FCM algorithm includes spatial, entropy and contrast based weight properties to enhance the segmentation accuracy by differentiating tumor areas successfully in medical images. To select the features MDSO employs a reinforced version of the classic Dolphin Swarm Optimization, merging Lévy refined global exploration, chaotic perturbation to local refinement, and dynamic convergence control to provide an exploration/exploitation balance. A binary encoding method that allows effective reduction of dimensions with a maintained diagnostic efficiency. The algorithm features extracted such as intensity (texture, GLCM, LBP, shape, and intensity) descriptors are processed with a transformation, conducted using a sigmoid-based transformation, after which selection is made.
Results: The comparison shows experimentally an improvement over accuracy, redundancy and also computational complexity metrics with improved results in Support Vector Machine and Random Forest classifiers.
Conclusion: The suggested M-FCM + MDSO algorithm is a more reliable solution to automated breast cancer diagnosis delivering higher classification accuracy.
Keywords: Breast Cancer Prediction, Modified Fuzzy C-Means, Dolphin Swarm Optimization, Feature Selection, Image Segmentation, Lévy Flight, Chaotic Perturbation.
How to cite this article: Sudha SA, Sankarasubramanian R. Mathematically Modified Optimisation Technique for Segmentation and Feature Selection. Int J Drug Deliv Technol. 2026;16(13s): 336-347. DOI: 10.25258/ijddt.16.13s.36
Source of support: Nil.
Conflict of interest: None