Comparison of iterative parametric and indirect deep learning-based reconstruction methods in highly undersampled DCE-MR Imaging of the breast Aditya Rastogi and Phaneendra K. Yalavarthy a) Department of Computational and Data Sciences, Indian Institute of Science, Bangalore 560012, India (Received 19 April 2020; revised 24 July 2020; accepted for publication 3 August 2020; published 6 September 2020) Purpose: To compare the performance of iterative direct and indirect parametric reconstruction methods with indirect deep learning-based reconstruction methods in estimating tracer-kinetic param- eters from highly undersampled DCE-MR Imaging breast data and provide a systematic comparison of the same. Methods: Estimation of tracer-kinetic parameters using indirect methods from undersampled data requires to reconstruct the anatomical images initially by solving an inverse problem. This recon- structed images gets utilized in turn to estimate the tracer-kinetic parameters. In direct estimation, the parameters are estimated without reconstructing the anatomical images. Both problems are ill-posed and are typically solved using prior-based regularization or using deep learning. In this study, for indirect estimation, two deep learning-based reconstruction frameworks namely, ISTA-Net + and MODL, were utilized. For direct and indirect parametric estimation, sparsity inducing priors (L1 and Total-Variation) and limited memory Broyden-Fletcher-Goldfarb-Shanno algorithm as solver was deployed. The performance of these techniques were compared systematically in estimation of vascu- lar permeability (K trans ) from undersampled DCE-MRI breast data using Patlak as pharmaco-kinetic model. The experiments involved retrospective undersampling of the data 209, 509, and 1009 and compared the results using PSNR, nRMSE, SSIM, and Xydeas metrics. The K trans maps estimated from fully sampled data were utilized as ground truth. The developed code was made available as https://github.com/Medical-Imaging-Group/DCE-MRI-Compare open-source for enthusiastic users. Results: The reconstruction methods performance was evaluated using ten patients breast data (five patients each for training and testing). Consistent with other studies, the results indicate that direct parametric reconstruction methods provide improved performance compared to the indirect parame- teric reconstruction methods. The results also indicate that for 209 undersampling, deep learning- based methods performs better or at par with direct estimation in terms of PSNR, SSIM, and nRMSE. However, for higher undersampling rates (509 and 1009) direct estimation performs better in all metrics. For all undersampling rates, direct reconstruction performed better in terms of Xydeas metric, which indicated fidelity in magnitude and orientation of edges. Conclusion: Deep learning-based indirect techniques perform at par with direct estimation tech- niques for lower undersampling rates in the breast DCE-MR imaging. At higher undersampling rates, they are not able to provide much needed generalization. Direct estimation techniques are able to pro- vide more accurate results than both deep learning- and parametric-based indirect methods in these high undersampling scenarios. © 2020 American Association of Physicists in Medicine [https:// doi.org/10.1002/mp.14447] Key words: AIF, breast imaging, DCE-MRI, deep learning, fast imaging, ISTA-Net, perfusion imaging, under-sampling 1. INTRODUCTION Early detection of pathologies is vital for reducing mortality and morbidity and with the development of various medical imaging techniques, clinicians are able to provide early and accurate diagnosis on the basis of anatomical manifestation of these pathologies. In the past few years, techniques have been developed, which can show both physical and physio- logical manifestation of the diseases. One such technique is dynamic contrast enhanced (DCE) magnetic resonance imaging (MRI) 1 in which a T 1 shortening contrast agent 2 (CA) gets injected into the bloodstream and T 1 weighted scans of the organ of interest are taken after that. This tech- nique of acquiring three-dimensional (3D) data with time results in Dynamic MR Imaging and based on the collected dynamic data one can measure the physiological characteris- tics of both healthy and unhealthy tissues. In DCE-MRI, one takes advantage of the fact that the vasculature near the unhealthy tissue will behave differently than that of its coun- terpart’ s, for example, the vasculature formed by malignant 4838 Med. Phys. 47 (10), October 2020 0094-2405/2020/47(10)/4838/24 © 2020 American Association of Physicists in Medicine 4838