Deep Learning Based Cardiac MRI Segmentation: Do We Need Experts?

TL;DR

This study investigates whether expert annotations are necessary for training deep learning models in cardiac MRI segmentation, finding that non-expert annotations can achieve comparable performance to expert labels when properly trained.

Contribution

The paper demonstrates that non-expert annotated data can be effectively used to train deep learning models for cardiac MRI segmentation, reducing reliance on costly expert annotations.

Findings

Non-expert groundtruth yields comparable segmentation performance to expert groundtruth.

Deep learning models trained on non-expert data perform well in clinical metrics.

Training with non-expert annotations offers a cost-effective alternative for dataset creation.

Abstract

Deep learning methods are the de-facto solutions to a multitude of medical image analysis tasks. Cardiac MRI segmentation is one such application which, like many others, requires a large number of annotated data so a trained network can generalize well. Unfortunately, the process of having a large number of manually curated images by medical experts is both slow and utterly expensive. In this paper, we set out to explore whether expert knowledge is a strict requirement for the creation of annotated datasets that machine learning can successfully train on. To do so, we gauged the performance of three segmentation models, namely U-Net, Attention U-Net, and ENet, trained with different loss functions on expert and non-expert groundtruth for cardiac cine-MRI segmentation. Evaluation was done with classic segmentation metrics (Dice index and Hausdorff distance) as well as clinical measurements, such as the ventricular ejection fractions and the myocardial mass. Results reveal that generalization performances of a segmentation neural network trained on non-expert groundtruth data is, to all practical purposes, as good as on expert groundtruth data, in particular when the non-expert gets a decent level of training, highlighting an opportunity for the efficient and cheap creation of annotations for cardiac datasets.