Fusion-Based Deep Learning with Nature-Inspired Algorithm for Intracerebral Haemorrhage Diagnosis.

Area of Science:

• Computational intelligence
• Medical imaging analysis
• Natural computing applications

Background:

• Intracerebral hemorrhage (ICH) is a critical cause of stroke, necessitating accurate and timely diagnosis.
• Manual segmentation of CT scans for ICH is labor-intensive and time-consuming.
• Deep learning (DL) offers potential for automated and efficient ICH diagnosis.

Purpose of the Study:

• To develop and evaluate an automated intracerebral hemorrhage diagnosis system using a novel fusion-based deep learning with swarm intelligence (AICH-FDLSI) algorithm.
• To address the challenges of data fusion and improve diagnostic performance in healthcare.
• To automate the time-consuming process of CT scan segmentation and analysis for ICH.

Main Methods:

• The AICH-FDLSI algorithm incorporates preprocessing (median filtering), image segmentation (Otsu multilevel thresholding with seagull optimization algorithm), feature extraction (Capsule Network and EfficientNet fusion), and classification (fuzzy support vector machine).
• Hyperparameter optimization for deep learning models was performed using the deer hunting optimization algorithm.
• The model was trained and validated on a benchmark intracranial hemorrhage dataset.

Main Results:

• The AICH-FDLSI model demonstrated proficient diagnostic performance on the intracranial hemorrhage dataset.
• Experimental results showed a significant improvement in accuracy compared to existing methods.
• The automated approach streamlines the diagnosis of intracerebral hemorrhage.

Conclusions:

• The proposed AICH-FDLSI algorithm provides an effective and automated solution for intracerebral hemorrhage diagnosis.
• Fusion-based deep learning combined with swarm intelligence enhances diagnostic accuracy in medical imaging.
• This approach has the potential to significantly aid radiologists in faster and more accurate ICH detection.