Thermal Interception Strike Optimization (TISO) for Binary Feature Selection in Otoscopic Image Classification: A Duplicate-Corrected Controlled Evaluation

This project investigates how the Thermal Interception Strike Optimization (TISO) algorithm—a metaheuristic method—can be used for binary feature selection in classifying otoscopic images. In medical imaging, the feature space is often very large and contains many redundant or irrelevant variables. Those extra features hurt classifier performance and slow down computation. To tackle this, the study combines two types of feature extraction: * Hand‑crafted Histogram of Oriented Gradients (HOG) descriptors. * Deep features taken from a frozen MobileNetV2 backbone. After extracting these features, TISO acts as a binary selector that picks the most useful subset. The goal is to boost classification accuracy while cutting down the number of features. The experiments

follow a duplicate‑corrected, controlled protocol, which helps keep the performance numbers unbiased. Because the full PDF was not available during review, the exact TISO settings and the precise accuracy gains are not listed—only the metadata from the source is known. Nevertheless, the framework offers a solid starting point for M.Tech students who want to try metaheuristic‑based feature reduction in computer‑aided diagnostic tools. It also provides a clear way to compare this approach with more traditional feature‑selection methods.

Message: Hi FE, I need help with “Thermal Interception Strike Optimization (TISO) for Binary Feature Selection in Otoscopic Image Classification: A Duplicate-Corrected Controlled Evaluation” in “Electronics & Communication Engineering”