An MLHOps-Driven Vision Transformer Approach for Pneumonia Classification in Chest X-Rays

Nagothi, Dharma Teja Venkatesh (2024) An MLHOps-Driven Vision Transformer Approach for Pneumonia Classification in Chest X-Rays. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (3MB) | Preview

Abstract

Diagnostic innovations are needed to enhance early detection and treatment of pneumonia, a global public health crisis. This work improves pneumonia detection from chest X-rays using Machine Learning Operations (MLOps) and Vision Transformers (ViT), a sophisticated deep learning (DL) model used in many computer vision applications. Pre-processed chest X-rays are given to a ViT model for feature extraction. The ViT encoder extracts hierarchical visual cues, whereas the classifier predicts pneumonia. The proposed method is tested using 5863 pneumonia-labeled NIH Chest X-ray pictures. The experiment compares the ViT model to a CNN classifier model for pneumonia classification based on accuracy, sensitivity, specificity, and AUC score criteria and it was found that ViT performed in terms of validation accuracy of 95.22% better than CNN accuracy of 94.84%. The ViT model execution was completed much faster than CNN execution (3018 seconds vs 7816 seconds). Based on these results, ViT was chosen to be implemented using MLOps practices for model training, evaluation, and deployment on Microsoft AzureML cloud. The suggested pneumonia detection on ML Health Operations (MLHOps) infrastructure using integrated ML pipelines allows rapid iteration and model optimization and ensures reproducibility for additional medical image analysis applications.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Sahni, Vikas UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Cloud computing R Medicine > Healthcare Industry Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning H Social Sciences > HM Sociology > Information Science > Communication > Medical Informatics
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	03 Jun 2025 13:56
Last Modified:	03 Jun 2025 13:56
URI:	https://norma.ncirl.ie/id/eprint/7727

Actions (login required)

View Item