PancreaScan : Pancreatic Cancer Detection

Abstract

Pancreatic cancer is one of the most dangerous forms of cancer, primarily because it is often diagnosed at a very advanced stage. By the time symptoms appear, the disease has usually progressed significantly, which contributes to its extremely high mortality rate. Early diagnosis is critical to improving the chances of successful treatment and survival. In this project, we developed a machine learning-based system to assist in the early detection of pancreatic cancer. The system uses two types of data: structured medical records in CSV format and medical imaging data such as CT or MRI scans.

We trained an XGBoost model on the medical record data, achieving an accuracy of 80%. In parallel, we used a Convolutional Neural Network (CNN) to analyze the image data, which reached an accuracy of 96%. These two models were designed to capture different but complementary indicators of the disease. We further enhanced the system by combining the predictions of both models through ensemble learning, making the overall system more robust and reliable. By integrating structured medical data with imaging data, our system provides a more comprehensive analysis, increasing the chances of early and accurate detection. This project demonstrates how advanced technologies like machine learning can support medical professionals in making faster and more precise diagnoses. Ultimately, this approach can reduce treatment delays and improve patient outcomes.