Heart Segmentation

Preprocessing Pipeline

(Preprocessing.ipynb) for preprocessing heart MRI datasets for heart segmentation tasks. The workflow is built around the Medical Segmentation Decathlon (MSD) - Task02_Heart dataset.

📋 Overview

1. Data Extraction: Extracts MRI image and label data from a .tar archive.
2. Path Management: pathlib to organize and access image (imagesTr) and label (labelsTr) directories.
3. Data Loading: Loads NIfTI (.nii.gz) files using nibabel for 3D medical imaging.
4. Visualization: Visualizes MRI slices and segmentation masks with matplotlib.
5. Preprocessing: Includes helper functions to:
   • Map image files to corresponding label files
   • Normalize and prepare data arrays
   • Optionally resize or crop image volumes for model input

Dataset

• Source: Medical Segmentation Decathlon (MSD) - Heart dataset
• Type: 3D Semantic Segmentation
• Format: NIfTI

Task02_Heart/
├── imagesTr/
│   ├── la_001.nii.gz
│   ├── la_002.nii.gz
│   └── ...
└── labelsTr/
    ├── la_001.nii.gz
    ├── la_002.nii.gz
    └── ...