Master’s Project: Reproducible Pulmonary Nodule Segmentation on LIDCIDRI

This project focuses on building a reproducible deep learning pipeline for pulmonary nodule segmentation using the LIDC-IDRI dataset. The goal is to accurately detect and segment lung nodules from CT scans, which is essential for early lung cancer diagnosis.

The system uses advanced image segmentation models (such as U-Net or its variants) to identify the exact location and shape of nodules within the lungs. Special attention is given to reproducibility by ensuring consistent data preprocessing, model training, and evaluation steps.

The final outcome is a reliable and repeatable framework that can be used by researchers and medical professionals to achieve consistent segmentation results and support clinical analysis