Medical Image Analysis Project (MVA - ENS Paris Saclay)

Paper review on using Deep-Learning and Deformable-Model Approach to Fully Automatic Segmentation of the Left Ventricle in Cardiac MRI[1].

Abstract

Manual segmentation of the Left Ventricle (LV) is a tedious and meticulous task that can vary depending on the patient, the Magnetic Resonance Images (MRI) cuts and the experts. Still today, we consider manual delineation done by experts as being the ground truth for cardiac diagnosticians. Thus, we are reviewing the paper - written by Avendi and al. - who presents a combined approach with Convolutional Neural Networks, Stacked Auto-Encoders and Deformable Models, to try and automate the segmentation while performing more accurately. Furthermore, we have implemented parts of the paper (around three quarts) and experimented both the original method and slightly modified versions when changing the architecture and the parameters.

Table of contents

1. Convolutional Neural Network
2. Stacked Auto Encoder
3. Deformable Models
4. Test and metrics

Paper review

For the course of Introduction to Medical Image Analysis (MVA 2017-2018), we (Sharone Dayan and Alexandre Attia) have written a NIPS format paper review and published it on arXiv [2] and presented its content.

References

[1] M. R. Avendi, A. Kheradvar and H. Jafarkhani. A Combined Deep-Learning and Deformable-Model Approach to Fully Automatic Segmentation of the Left Ventricle in Cardiac MRI. 2015
[2] A. Attia and S. Dayan. Detection and segmentation of the Left Ventricle in Cardiac MRI using Deep Learning. 2017.