Multilingual VLM-Based Radiographic Analysis System for Automated Abnormality Detection and Clinical Report Generation

View at publisher Summarise with VIRAT
Authors: R Senthilkumar S, Packiaraj R, Sam Varghese George, S. Sumathi
Cited by: 0
References: 30

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Musculoskeletal abnormalities in X-rays are a major global health issue. Fast and accurate diagnosis is crucial. Xrays are commonly used in medical checkups as they help doctors to view the bones clearly and quickly identify abnormalities in the musculoskeletal system. Although Deep Learning has improved automated X-ray analysis, current models still face issues such as misclassification, poor generalization, and image-quality limitations. This project presents a combined Deep Learning Model + Vision–Language Model (VLM) framework for abnormality detection, body-part classification, and automatic clinical text generation. The proposed system works by extracting visual features using a CNN backbone and feeding them into classification heads, while the same features are fused with a Vision–Language Model to generate clinical text. An accuracy of 99.4% was achieved by the proposed model with better test results, confirming the model's capacity to outperform conventional architectures in x-ray image-based analysis.

Fields of Study & Subjects

Medical Imaging and Analysis Multimodal Machine Learning Applications Artificial Intelligence in Healthcare and Education

Publication details

Journal / source
International Conference on Communication, Computing and Electronics Systems (ICCCES)
Pages
1846–1854
Publication date
21 January 2026
Language
EN