Abstract

Spinal cord deformities, including scoliosis, kyphosis, and lordosis, significantly impact the quality of life and often require early and precise diagnosis to prevent further complications. Traditional diagnostic methods such as X-ray interpretation and manual measurements are time-consuming and prone to subjective errors. To address these challenges, this work proposes a deep learning-based approach leveraging Region-Based Convolutional Neural Networks (RCNN) for automatic spinal cord deformity detection and classification. The method processes medical imaging data, extracts critical spinal features, and accurately identifies deformities. RCNN''s capability to localize regions of interest allows it to detect deformities with high accuracy, overcoming limitations in prior approaches like feature extraction constraints and limited generalization. The method was trained and evaluated using a curated dataset of spinal X-ray images, ensuring robustness across varying deformity severities. Experimental results demonstrate superior performance compared to three existing methods, achieving significant improvements in accuracy, precision, recall, and F1-score. This approach provides a reliable and efficient tool for clinicians, reducing diagnostic time and enhancing the consistency of deformity detection.

Authors

C. Rupesh, T. Madhuri
Savitribai Phule Pune University, India

Keywords

Spinal cord deformity, Region-Based Convolutional Neural Network, Deep learning, Medical imaging, Automated diagnosis