Skip to content

Latest commit

 

History

History
58 lines (41 loc) · 2.96 KB

README.md

File metadata and controls

58 lines (41 loc) · 2.96 KB
Raw

Pulmonary Hypertension Segmentation Comparison

Pulmonary hypertension (PH) is a deadly disease with a worse prognosis than systemic hypertension. In PH, the distal pulmonary vasculature remodels, leading to increased blood pressure within the lungs. Many of the most important changes occur in the distal vasculature that are too small to capture with typical imaging approaches.

To better understand disease development, researchers use ex vivo imaging in animal models. Micro-CT has much higher resolution than traditional CT and can capture the size of vessels that drive long-term complications. These scans contain thousands of vessel segments, making manual measurements challenging.

This repository is part of a Carnegie Mellon University research project that aims to automate the segmentation of the pulmonary arteries from mouse and sheep lungs from MicroCT scans and determine the accuracy of the computer-generated model.

Project Overview

This project investigates an automated segmentation tool to determine its fidelity against manual measurements, focusing on both geometric and morphometric factors that determine patient hemodynamics. We utilize open-source software packages, 3D Slicer and ClearMap, to evaluate images of mouse, sheep, and human lungs with different types of PH. The goal is to quantify the visible changes in the disease.

Features

  • Converts an .nrrd 3D segmentation to a multi-image .tif stack.
  • Overlays the computer-generated .tif stack on top of the 3D Slicer exported segmentation.
  • Compares the accuracy of the computer model.
  • Displays the error per slice and overall error.

Usage

  1. Replace the corresponding file paths in filePaths.py.
  2. Run main.py.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Authors and Acknowledgements

  • Aravind Gunasekaran
  • Dr. Jason Szafron

Technologies Used

  • Python
  • NumPy
  • PILLOW
  • scikit-image
  • nrrd

Tools and Software

  • 3D Slicer: An open-source software platform for medical image informatics, image processing, and three-dimensional visualization.
  • ClearMap: A software for analyzing, visualizing, and exploring large-scale biological image data.

Tools and Software

  • 3D Slicer: An open-source software platform for medical image informatics, image processing, and three-dimensional visualization.
  • ClearMap: A software for analyzing, visualizing, and exploring large-scale biological image data.

TODO

  • Make a region with connected black pixels
  • Find eccentricity index for all regions
  • Ones closer to 0 are probably vessel versus other being cros sections or others
  • Compare to the segmentation (make sure overlap and offset are right)
  • Find areas that have about 75% similarity within the overlap region
  • Those are probably what the tubmap picked up as vessels
  • Go step by step making each a function and then put together
  • histagram of slice errors
  • finding the size fo the vessel