NORMA eResearch @NCI Library

NeuroPCOS: Detection of Polycystic Ovary Syndrome in Ultrasound Images Using Filter and Transfer Learning model

Dutta, Saheli (2023) NeuroPCOS: Detection of Polycystic Ovary Syndrome in Ultrasound Images Using Filter and Transfer Learning model. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
Preview
PDF (Master of Science)
Download (6MB) | Preview
[thumbnail of Configuration Manual]
Preview
PDF (Configuration Manual)
Download (4MB) | Preview

Abstract

In recent years, polycystic ovary syndrome (PCOS) has been seen a lot in women, because of their lifestyle and food habits. This particular syndrome is affecting their health to a large extent. Keeping these diseases undiagnosed can aggravate the possibility of cancer and damage the reproductive system at any stage of age. There have been a lot of studies conducted related to this topic where several kinds of deep neural networks have been employed to diagnose and identify PCOS-infected ultrasound images because the ultrasound images help to see the condition of the ovary, like a CT scan for the brain or MRI. Unfortunately, because of the capturing devices, the images are exposed to noise, which can lead to the wrong diagnosis of the disease. That is why introducing the filtration method is very necessary. This proposed research utilized the efficacy of a median filter to cancel the noises and fed the images to transfer learning models such as VGG16 and ResNet50 because of their deep hierarchical architecture and skip connection, respectively. Moreover, due to data limitation issues, the ImageDataGenerator data augmentation method is utilized as well. Among these models, ResNet50 outperformed VGG16 by showing balanced specificity(46%) and sensitivity(43%), indicating that identifying both positive and negative classes. Although VGG16’s (81%) accuracy is better than the ResNet50 model(52%) VGG16 model fails to identify true positive cases with a low sensitivity of 12% which indicates biases in the model and makes the model unreliable for PCOS diagnosis.

Item Type: Thesis (Masters)
Supervisors:
Name
Email
Palaniswamy, Sasirekha
UNSPECIFIED
Uncontrolled Keywords: Medical Diagnosis; Median Filter; Data Augmentation; Transfer Learning Model
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
R Medicine > RA Public aspects of medicine > RA0421 Public health. Hygiene. Preventive Medicine
R Medicine > RG Gynecology and obstetrics
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Ciara O'Brien
Date Deposited: 08 May 2025 09:48
Last Modified: 08 May 2025 09:48
URI: https://norma.ncirl.ie/id/eprint/7509

Actions (login required)

View Item View Item