Optimized Hybrid Segmentation and Feature-Based Classification for Interpretable, Efficient Automated Medical Diagnosis

-, Narendra Kumar (2025) Optimized Hybrid Segmentation and Feature-Based Classification for Interpretable, Efficient Automated Medical Diagnosis. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (452kB) | Preview

Abstract

This project revisits the concept of automated anomaly detection by doubting the validity of the hybrid segmentation and radiomic classification methodologies. Even though publicly available MRI and X-ray datasets were improved and processed with the help of Sobel-watershed technique, the masks and radiomic features that were obtained did not always reflect the characteristics of the tumour. Classifiers/SVM, XGBoost, and a light-weight neural network exhibited varying accuracy, sensitivity, and ROC-AUC, and none of the models expressed consistent high results. Surprisingly, XGBoost was not consistent, and the effectiveness of SVM did not result in reliable performance. Such results imply that programmable pipelines might have a problem in fulfilling the robustness needed in clinical diagnostic application.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Sahni, Vikas UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence R Medicine > Healthcare Industry
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Ciara O'Brien
Date Deposited:	02 Sep 2026 08:34
Last Modified:	02 Sep 2026 08:34
URI:	https://norma.ncirl.ie/id/eprint/9749

Actions (login required)

View Item