Deep Learning for Liver Segmentation
Authors:-Amarnath Chigurupati. Madhuri Sirasanagandla. Ankit Kommalapati. Siddique Ibrahim Peer Mohammed, Madhuri Sirasanagandla
Abstract-Liver cancer is becoming a huge threat to global health health, where early detection and accurate diagnosis are crucial for effective treatment [1]. Our research on deep learning- A based learning system for automatic segmentation of the liver and The tumor from computed tomography (CT) images is highlighted, using a U-Net model integrated with ResNet-34, which acts as a backbone [2]. This model is trained on the Liver Tumor Segmentation Challenge (LiTS) dataset, which is a standard for This type of problem [2]. Training a high-performance model, The project itself differentiates with the development of a user- friendly GUI with the help of the Python package PyQt5, making It is possible to achieve real- time visualization and user-friendly interaction for the end users like radiologists, students, and researchers [10] . This interface helps in taking input as an image in the form of a JPG, predicts segmentation tasks, and compares the results With the grayscale liver anatomy structures. Our model delivers high accuracy in segmentation, obtaining a high accuracy Dice coefficient of 98.20% with an extraordinary precision, recall, and f-score up to 99.89%, making it usable for real-time scenarios like clinical and research purposes Index Terms—Liver Segmentation, Deep Learning, U-Net, ResNet-34, FastAI, PyQt5.
