Few-Shot Thoracic Disease Classification using Prototypical Networks

Amir, Ramsha (2024) Few-Shot Thoracic Disease Classification using Prototypical Networks. Masters thesis, Dublin, National College of Ireland.

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

Abstract

Thoracic diseases are among the most common health challenges worldwide, requiring accurate diagnostic tools to ensure timely intervention and better treatment outcomes. These diseases encompass a wide range of conditions affecting the lungs, heart, and other key organs within a thoracic cavity, which poses significant challenges for early diagnosis and effective treatment. For instance, diseases such as pneumonia, cardiomegaly, and COVID-19 lead to several complications ranging from respiratory failure and in some cases even death, hence the need for timely diagnosis to improve patient outcomes. However, the lack of labeled medical image datasets for these diseases makes it difficult to create reliable diagnostic models. This study investigates few-shot learning models, particularly Prototypical Networks into practice, for thoracic disease detection using minimal annotated chest X-ray images. Few-shot learning has proved significant for medical imaging, where large annotated datasets are not available. The approach proposed makes use of pre-trained models like VGG19, ResNet50, or DenseNet121 for feature extraction will subsequently classify diseases well with very few samples per category.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Anant, Aaloka UNSPECIFIED
Uncontrolled Keywords:	X-ray Image; Few-Shot Learning; VGG19; ResNet50; DenseNet121; Prototypical Networks; Thoracic Disease Detection
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science R Medicine > Healthcare Industry
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	01 Sep 2025 13:59
Last Modified:	01 Sep 2025 13:59
URI:	https://norma.ncirl.ie/id/eprint/8671

Actions (login required)

View Item