anonymousdog12/left-ventricle-segmentation

Computer Vision Project - LV Segmentation using Deep Learning, Active Contour, and Principle Component Analysis

Cardiac MR Left Ventricle Segmentation

UCLA-CS269 Project -- Cardiac MR Left Ventricle Segmentation Challenge.

Abstract

Image segmentation of the left ventricle from cardiac magnetic resonance imaging is a crucial but tedious step for clinical cardiac health diagnosis. In this project, we proposed to use convolutional neural network combined with deformable model to conduct medical image segmentation. A three-step approach is proposed to deal with the low-contrast nature of medical image and relative small size of available data. Finally, the performance of the segmentation algorithm is evaluated from both quantitative and qualitative aspects.

Members

Qi Qu
Jingxi Yu
Changyu Yan
Sha Liu

Dataset

Data could be downloaded on this site after registering http://smial.sri.utoronto.ca/LV_Challenge/Home.html. All data is expected to be released to ../Data/

Codes

ROI_detection.ipynb shows the process of ROI detection and also how to loads and prepare the data in this challenge.
StackedAE.ipynb && PCA_autoencoder.ipynb shows the process of shape prior inference.
SAE+ActiveContour.ipynb shows the last step of computing the contour.