Identification and Classification of Cancer from Images using CNN and Vision Transformers

Rane, Aniket Vivekanand (2025) Identification and Classification of Cancer from Images using CNN and Vision Transformers. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (6MB) | Preview
Preview	PDF (Configuration Manual) Download (770kB) | Preview

Abstract

Traditional methods of detecting cancer rely on invasive methods and tend to be often time-consuming, and early detection is crucial to increase the chances of survival for the patient. Deep learning (DL) is a subset of machine learning that utilizes neural networks that have a structure similar to the neurons inside a human brain and have the ability to learn complex patterns from data and this ability of deep learning models makes them suitable for tasks like image recognition, natural language processing, and speech recognition. In this study, the aim is to train deep learning models, specifically Convolutional Neural Networks (CNN’s) and Transformer-based models. In this study, a comparison between these two models has been done on the basis of different metrics like accuracy, precision, recall and training time. The dataset that was used for the study is a multi-cancer dataset which was formed after combining a few different cancer datasets and is available on Kaggle. The dataset consists of images for eight different types of cancer (cervical, lung, colon, oral, kidney, breast, lymphoma and brain) and 26 subclasses. Our research focuses on lung, colon, oral, lymphoma, and breast cancer and their related subclasses and few CNN and Transformer models including the vision transformer (ViT) were trained on this dataset in order to test which type of models are most suitable and perform the best at accurately recognizing different cancer types. The dataset consists of 60,002 cancer images and the findings of this study might help future researchers in improving the patient outcomes and provide a direction for further research in this domain.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Subhnil, Shubham UNSPECIFIED
Uncontrolled Keywords:	Deep learning; ResNet; CNN; cancer; cancer classification; Lung cancer; Lymphoma; Oral cancer; accuracy; Vision Transformer; zero-shot; few-shots
Subjects:	Q Science > Life sciences > Medical sciences > Pathology > Tumors > Cancer R Medicine > Healthcare Industry Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	08 Sep 2026 11:48
Last Modified:	08 Sep 2026 11:48
URI:	https://norma.ncirl.ie/id/eprint/9897

Actions (login required)

View Item