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

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 that 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 data sets on which machine learning can successfully be trained. 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 ground truth 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. The results reveal that generalization performances of a segmentation neural network trained on non-expert ground truth data is, to all practical purposes, as good as that trained on expert ground truth data, particularly when the non-expert receives a decent level of training, highlighting an opportunity for the efficient and cost-effective creation of annotations for cardiac data sets.

CoNeS: Conditional neural fields with shift modulation for multi-sequence MRI translation

“…Although plenty of work has shown great progress in medical image translation, most previous works have been evaluated based on image similarity metrics and only a few papers have evaluated the benefits of using synthesized images for downstream analysis. Amirrajab et al (2023) fine-tuned a segmentation model with synthesized cardiac images to improve the performance of different modalities; Skandarani et al (2020) introduced a variational autoencoder (VAE) based network for image translation based data augmentation to improve the generalization capabilities of a segmentation model. In practice, however, it would be more straightforward and beneficial to use the synthesized images directly without finetuning or training a new network.…”

Automatic Myocardium Segmentation in Delayed-Enhancement MRI with Pathology-Specific Data Augmentation and Deep Learning Architectures

“…New shapes are generated from an orthonormal feature basis that is computed based on eigenvector analysis on the correlation matrix of shape features. Skandarani et al [14] proposed the use of variational autoencoders and generative adversarial networks (GANs) [15] for generating realistic synthetic MRI. Their results show that training convolutional neural networks with the data generated by their model achieved competitive performance with respect to other traditional techniques.…”