Optimized Multi-Layered Feature-Based Multi-Label Classification Framework for Detecting Hemorrhagic Lesions in Traumatic Brain Injury

Traumatic Brain Injury is considered one of the most relevant global health problems since the correct diagnosis and discrimination of hemorrhagic lesions are only possible by adopting timely intervention. This study proposes an optimal feature-based, multi-layer classification system based on advanced machine learning and image processing techniques to attain even higher accuracy in the detection of lesion in TBI. Dimensionality reduction-multi-label classification-driven feature extraction paradigm offers robustness in handling variability of datasets. Its 98.05% accuracy, 99.04% precision, 99.01% recall, and 1.05% MSE show high capability of classification and low capability of prediction error of the proposed system. Its comparative performance on state-of-the-art algorithms is well supported in terms of improvement on every index. The proposed approach is stable and scalable with computer-aided framework for diagnosis of hemorrhagic lesions, paving the way to improved clinical outcomes.