Medical Imaging Analysis

Compact MATLAB portfolio project covering classical medical image analysis workflows across multimodal registration, level-set segmentation, quantitative morphometry, and 3D reconstruction.

The repository brings together a small set of focused studies on MRI, CT, and PET data, emphasizing interpretable classical methods, visual validation, and physically meaningful measurements.

Overview

The project is organized around four complementary tasks:

• Multimodal registration: align T2 MRI, DWI MRI, PET, and rotated MRI slices using SSD, NCC, and mutual information.
• Classical segmentation: segment renal, cardiac, and breast structures with Malladi-Sethian and Chan-Vese level-set methods.
• Quantitative morphometry: compute structure-specific areas in mm^2 and volumetric estimates in mm^3.
• 3D reconstruction: reconstruct the left atrium from slice-wise MRI segmentation.

Included studies:

• T2/DWI/PET registration with translation and rotation assessment
• Kidney, cardiac ventricle, left atrium, and breast lesion segmentation
• Area and volume estimation in physical units
• 3D left-atrium surface reconstruction

Main Components

• Medical_Image_Registration/ implements resolution harmonization, translation and rotation search, similarity heatmaps, and checkerboard-based registration assessment.
• Medical_Image_Segmentation/Kidney_Segmentation/ combines ROI-based registration, Chan-Vese segmentation, and kidney/medulla/cortex area estimation.
• Medical_Image_Segmentation/Cardiac_Ventricle_Segmentation/ segments left and right ventricular cavities from short-axis cardiac MRI.
• Medical_Image_Segmentation/Left_Atrium_Segmentation/ performs slice-by-slice atrial segmentation, volume estimation, and 3D surface reconstruction.
• Medical_Image_Segmentation/MR_Breast_Segmentation/ applies an edge-based level-set workflow to breast lesion delineation.

Key Outputs

• Registration heatmaps, joint histograms, checkerboard overlays, and summary figures for multimodal alignment
• Segmentation overlays and intermediate contour-evolution visualizations
• Quantitative area and volume estimates reported in physical units
• A 3D left-atrium surface reconstruction from segmented MRI slices

Representative Methods

• Similarity metrics: SSD, NCC, MI
• Segmentation models: Malladi-Sethian, Chan-Vese
• Preprocessing: intensity normalization, sigmoid contrast enhancement, Gaussian smoothing, anisotropic diffusion
• Post-processing: connected-component filtering, morphological cleanup, surface extraction

Repository Structure

Medical-Imaging-Analysis/
|-- Medical_Image_Registration/
|   |-- Medical_Image_Registration.m
|   |-- figures/
|   `-- report.md
|-- Medical_Image_Segmentation/
|   |-- Cardiac_Ventricle_Segmentation/
|   |-- Kidney_Segmentation/
|   |-- Left_Atrium_Segmentation/
|   `-- MR_Breast_Segmentation/
|-- data/
|-- lib/
`-- README.md

Requirements

• MATLAB with Image Processing Toolbox and DICOM support
• Repository helper functions in lib/
• Input datasets stored in data/
• Optional: Iso2Mesh on the MATLAB path for the 3D reconstruction step in the left-atrium workflow

The MATLAB scripts resolve data/ and lib/ relative to the repository root so the checked-in structure remains portable across machines.

Detailed Reports

• Registration: Medical_Image_Registration/report.md
• Kidney segmentation: Medical_Image_Segmentation/Kidney_Segmentation/report.md
• Cardiac ventricle segmentation: Medical_Image_Segmentation/Cardiac_Ventricle_Segmentation/report.md
• Left atrium segmentation and 3D reconstruction: Medical_Image_Segmentation/Left_Atrium_Segmentation/report.md
• Breast lesion segmentation: Medical_Image_Segmentation/MR_Breast_Segmentation/report.md