Deep Learning-based Analysis of Spinal Cord Regions for Risk Assessment and Clinical Awareness: A Social Worker's Perspective
E. Venkatesan
PG Department of Computer Science, RV Government Arts College, Chengalpattu, India.
V. Thangavel
*
St. Francis Institute of Management and Research, Mumbai, India.
*Author to whom correspondence should be addressed.
Abstract
Accurate detection and segmentation of spinal cord injuries in MRI images are critical for diagnosis and treatment planning. This study investigates the performance of two deep learning algorithms, Artificial Neural Networks (ANN) and Convolutional Neural Networks (CNN), using median and Gaussian filtering as preprocessing techniques to reduce noise and enhance image clarity. The filtered images were analyses to evaluate detection and segmentation accuracy, runtime, and memory usage. Results demonstrate that CNN, particularly with Gaussian-filtered images, achieved the highest accuracy and segmentation precision, outperforming ANN in all cases. Although ANN offered faster processing and lower memory requirements, CNN provided superior performance in injury identification and segmentation. Overall, CNN combined with Gaussian filtering is the most effective approach for accurate and reliable spinal cord injury analysis.
Keywords: Spinal cord injury, MRI, Convolutional Neural Network (CNN), image preprocessing, segmentation, detection accuracy