Heterogeneous Virtual Population of Simulated CMR Images for Improving the Generalization of Cardiac Segmentation Algorithms

Simulating a large set of medical images with variability in anatomical representation and image appearance has the potential to provide solutions for addressing the scarcity of properly annotated data in medical image analysis research. However, due to the complexity of modeling the imaging procedure and lack of accuracy and flexibility in anatomical models, available solutions in this area are limited. In this paper, we investigate the feasibility of simulating diversified cardiac magnetic resonance (CMR) images on virtual male and female subjects of the eXtended Cardiac and Torso phantoms (XCAT) with variable anatomical representation. Taking advantage of the flexibility of the XCAT phantoms, we create virtual subjects comprising different body sizes, heart volumes, and orientations to account for natural variability among patients. To resemble inherent image quality and contrast variability in data, we vary acquisition parameters together with MR tissue properties to simulate diverse-looking images. The database includes 3240 CMR images of 30 male and 30 female subjects. To assess the usefulness of such data, we train a segmentation model with the simulated images and fine-tune it on a small subset of real data. Our experiment results show that we can reduce the number of real data by almost 80% while retaining the accuracy of the prediction using models pre-trained on simulated images, as well as achieve a better performance in terms of generalization to varying contrast. Thus, our simulated database serves as a promising solution to address the current challenges in medical imaging and could aid the inclusion of automated solutions in clinical routines.