Pancreatic Tumor Detection Using Image Processing

Abstract

Pancreatic tumor is one of the most life-threatening cancers, often diagnosed at advanced stages due to the absence of early symptoms and the lack of effective screening techniques. This project aims to develop an efficient and accurate computer-aided detection (CAD) system for pancreatic tumor identification using advanced image processing techniques. Utilizing a publicly available dataset from Kaggle consisting of CT and MRI scans, the system employs a 3D Convolutional Neural Network (3D CNN)-based approach to automate tumor detection. Image pre-processing techniques such as resizing for uniform input dimensions, normalization to standardize pixel intensity, and data augmentation to enhance dataset diversity are applied to improve feature extraction and reduce overfitting. The model achieved an accuracy of 99%, demonstrating high efficacy in detecting pancreatic tumors. Additionally, domain adaptation techniques were implemented to enhance generalization across different imaging modalities. The application is developed using Tkinter for its simplicity and efficient integration with the machine learning model, offering a lightweight, local, and user-friendly interface for real-time analysis in clinical settings. This system not only supports early and accurate detection but also aids clinicians in making timely decisions, potentially improving patient outcomes. The application is developed using Tkinter for its simplicity and ease of creating a lightweight, local, and user-friendly interface. This choice allows for seamless integration with the backend machine learning model, providing a quick and efficient way for clinicians to analyse medical scans. Unlike web frameworks like Django, Tkinter eliminates the need for server-side deployment, making the tool more accessible and faster for real-time usage in clinical settings.