Gorilla Troops Optimization with Deep Learning-based Intracranial Hemorrhage Diagnosis on computed tomography images

Intracranial haemorrhage (ICH) is a decisive healthcare emergency that needs initial diagnoses and precise evaluation. Due to the higher mortality rate of nearly 40 %, the initial classification and recognition of diseases utilizing computed tomography (CT) scans were essential to ensure a favourable estimation and confine the presence of neurological disorders. As the manual diagnosis method consumes more time, automatic ICH classification and detection methods utilizing Deep Learning (DL), Artificial Intelligence (AI), and Machine Learning (ML) techniques are employed to scale up medical diagnosis platforms. Though numerous DL-based ICH diagnosis models are available in the literature, they suffer from several challenges, such as gradient vanishing, over-fitting, hyperparameter tuning, and extensive computation. This study developed a Gorilla Troops Optimization with Deep Learning-based Intracranial Hemorrhage Diagnosis (GTODL-ICHD) approach to resolve these issues on CT images. The major aim of the GTODL-ICHD approach is to identify and classify ICH on CT imageries. To achieve this, the presented GTODL-ICHD approach utilizes the Wiener filtering (WF) technique at the primary level to remove the noise. Also, the GTODL-ICHD approach employs an enhanced capsule network (ECN) model for feature extraction, and the GTO method is utilized for hyperparameter tuning. Finally, the deep belief network (DBN) method is implemented for the ICH detection process. A series of experimental analyses were conducted to demonstrate the improved performance of the GTODL-ICHD technique. The performance validation of the GTODL-ICHD technique portrayed a superior accuracy value of 97.79 % over existing models.