HEPATOSCAN:AI BASED LIVER TUMOR SEGMENTATION SYSTEM

Abstract: Liver cancer is one of the leading causes of cancer-related deaths worldwide, and early diagnosis plays a crucial role in improving treatment outcomes and patient survival. Traditional methods of analyzing CT scan images rely heavily on manual interpretation by radiologists, which can be time-consuming, labor-intensive, and prone to human error. To address these challenges, this project presents HepatoScan, an AI-based web application designed for automated liver and tumor segmentation using deep learning techniques.

The proposed system utilizes a U-Net–based convolutional neural network implemented using TensorFlow to perform accurate pixel-level segmentation of liver and tumor regions from CT scan images. Prior to segmentation, the images undergo preprocessing steps such as grayscale conversion, resizing, and normalization to ensure consistent input quality for the model. The trained model generates liver and tumor masks, enabling automated identification of affected regions. In addition to segmentation, the system performs tumor size estimation using pixel-based area calculation and provides a user-friendly web interface developed using Flask and web technologies.

The application also includes automated PDF medical report generation containing the uploaded CT image, segmentation outputs, and tumor size details. By integrating deep learning, medical image processing, and automated reporting into a single platform, HepatoScan aims to improve diagnostic efficiency, reduce manual effort, and support radiologists in clinical decision-making.

Keywords: Artificial Intelligence (AI), Deep Learning, Medical Image Segmentation, Liver Tumor Detection, U-Net Architecture, Convolutional Neural Network (CNN), TensorFlow, CT Scan Analysis, Flask Web Application, Tumor Size Estimation, Image Preprocessing, Medical Imaging, Computer Vision, Automated PDF Report Generation, Healthcare AI, Liver Cancer Diagnosis, Supervised Learning, Biomedical Image Analysis.