Classification, Regression and Segmentation directly from k-Space in Cardiac MRI

TL;DR

This study investigates diagnosing cardiac diseases directly from raw k-space data, leveraging phase information that is lost in conventional magnitude images. It introduces KMAE, a Transformer-based framework with a pre-trained K-GIN encoder that processes undersampled k-space and supports classification, phenotype regression, and myocardial segmentation through task-specific decoders. KMAE demonstrates competitive regression and classification performance relative to image-domain MAEs and achieves a myocardial segmentation Dice of $0.884$, while maintaining robustness up to undersampling factors of $R=8$. The results advocate end-to-end k-space–based diagnosis in cardiac MRI and point to future extensions to multi-coil acquisitions to further enhance accuracy and clinical impact.

Abstract

Cardiac Magnetic Resonance Imaging (CMR) is the gold standard for diagnosing cardiovascular diseases. Clinical diagnoses predominantly rely on magnitude-only Digital Imaging and Communications in Medicine (DICOM) images, omitting crucial phase information that might provide additional diagnostic benefits. In contrast, k-space is complex-valued and encompasses both magnitude and phase information, while humans cannot directly perceive. In this work, we propose KMAE, a Transformer-based model specifically designed to process k-space data directly, eliminating conventional intermediary conversion steps to the image domain. KMAE can handle critical cardiac disease classification, relevant phenotype regression, and cardiac morphology segmentation tasks. We utilize this model to investigate the potential of k-space-based diagnosis in cardiac MRI. Notably, this model achieves competitive classification and regression performance compared to image-domain methods e.g. Masked Autoencoders (MAEs) and delivers satisfactory segmentation performance with a myocardium dice score of 0.884. Last but not least, our model exhibits robust performance with consistent results even when the k-space is 8* undersampled. We encourage the MR community to explore the untapped potential of k-space and pursue end-to-end, automated diagnosis with reduced human intervention.

Figures (2)

• Figure 1: An overview of KMAE and MAEs with downstream tasks. The upper section depicts KMAE, a modification of the K-GIN model. The lower section illustrates the modification of MAEs. (a) The pre-training of KMAE and MAEs for MRI reconstruction. KMAE processes under-sampled k-space data, while MAEs handle in the image domain with masked-out patches. (b) In downstream task fine-tuning, we freeze encoders of KMAE and MAEs while their decoders are modified for regression and classification. (c) We adapt decoders of KMAE and MAEs for segmentation tasks, with the upper section highlighting our newly proposed k-space segmentation method.
• Figure 2: Comparison of segmentation methods for delineating myocardial regions