Lightweight CNN-Enabled Framework for Automated Detection of Scoliosis and Spondylosis Using Spine X-rays

Gobalakrishnan Natesan; Anbarasan Murugesan; Ramshankar Nagarajan

doi:10.47852/bonviewJCCE52026933

Authors

Gobalakrishnan Natesan Department of Computer Science and Engineering, Saveetha Engineering College, India https://orcid.org/0000-0003-3820-6744
Anbarasan Murugesan Department of Computer Science and Engineering, Saveetha Engineering College, India https://orcid.org/0000-0002-2828-9016
Ramshankar Nagarajan Department of Computer Science and Engineering, Saveetha Engineering College, India https://orcid.org/0000-0003-4170-0401

DOI:

https://doi.org/10.47852/bonviewJCCE52026933

Keywords:

spine abnormalities, convolutional neural networks, AlexNet, ResNet

Abstract

The spine serves as a fundamental structure that supports bodily functions, movements, protection of vital neural pathways, and overall well-being. Any abnormalities or injuries to the spine can significantly impact a person's quality of life. Early detection of these conditions is crucial for effective treatment and management. Hence, this study focuses on the detection and classification of spine abnormalities using advanced deep learning algorithms. In recent years, deep learning methods, particularly convolutional neural networks (CNNs), have demonstrated promising performance in medical image processing applications. This study investigates the effectiveness of advanced CNN architectures, specifically AlexNet and ResNet, in detecting spine abnormalities and distinguishing scoliosis, spondylosis, and normal spine. Leveraging a dataset comprising 3341 diverse spine X-rays images, this study aims to not only compare the effectiveness of AlexNet and ResNet but also determine the most accurate model for deployment in clinical settings. The results demonstrate the comparable performance of AlexNet and ResNet in detecting spine abnormalities. Insights gained from this comparative analysis can inform healthcare practitioners and researchers on the optimal choice of deep learning architecture for spine abnormality detection, ultimately contributing to improved diagnostic accuracy and patient outcomes.

Received: 25 July 2025 | Revised: 28 September 2025 | Accepted: 15 November 2025

Conflicts of Interest

The authors declare that they have no conflicts of interest to this work.

Data Availability Statement

The data that support the findings of this study are openly available in Mendeley Data at http://doi.org/10.17632/xkt857dsxk.1, reference number [23].

Author Contribution Statement

Gobalakrishnan Natesan: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Data curation, Writing – original draft, Project administration. Anbarasan Murugesan: Software, Formal analysis, Investigation, Resources, Data curation, Writing – review & editing, Visualization, Supervision. Ramshankar Nagarajan: Software, Formal analysis, Investigation, Resources, Data curation, Writing – review & editing, Visualization.