Accelerated multicontrast reconstruction for synthetic MRI using joint parallel imaging and variable splitting networks

Abstract

Purpose Synthetic magnetic resonance imaging (MRI) requires the acquisition of multicontrast images to estimate quantitative parameter maps, such as T1, T2, and proton density (PD). The study aims to develop a multicontrast reconstruction method based on joint parallel imaging (JPI) and joint deep learning (JDL) to enable further acceleration of synthetic MRI. Methods The JPI and JDL methods are extended and combined to improve reconstruction for better‐quality, synthesized images. JPI is performed as a first step to estimate the missing k‐space lines, and JDL is then performed to correct and refine the previous estimate with a trained neural network. For the JDL architecture, the original variable splitting network (VS‐Net) is modified and extended to form a joint variable splitting network (JVS‐Net) to apply to multicontrast reconstructions. The proposed method is designed and tested for multidynamic multiecho (MDME) images with Cartesian uniform under‐sampling using acceleration factors between 4 and 8. Results It is demonstrated that the normalized root‐mean‐square error (nRMSE) is lower and the structural similarity index measure (SSIM) values are higher with the proposed method compared to both the JPI and JDL methods individually. The method also demonstrates the potential to produce a set of synthesized contrast‐weighted images that closely resemble those from the fully sampled acquisition without erroneous artifacts. Conclusion Combining JPI and JDL enables the reconstruction of highly accelerated synthetic MRIs.

Accelerated 3D myelin water imaging using joint spatio‐temporal reconstruction

“…12,[22][23][24] Recently, it has been shown that integrating PI methods into deep neural networks has the potential to improve the reconstruction quality of accelerated MRI acquisitions. [24][25][26][27][28][29][30][31] Additionally, several studies have implemented various forms of spatiotemporal convolutions for deep neural networks to exploit temporal information. [32][33][34][35] Inspired by these approaches, we propose a reconstruction method that combines joint parallel imaging (JPI) and joint deep learning (JDL) with spatiotemporal convolutions.…”

Improving high frequency image features of deep learning reconstructions via k‐space refinement with null‐space kernel

“…Previously, it was shown that prior to DL reconstruction, using parallel imaging methods resulted in improved reconstruction. 35,36 We believe that these types of methods might have the potential to improve the current state of the DL architecture for MRI reconstruction.…”

Synthetic MR: Physical principles, clinical implementation, and new developments

“…By applying a deep learning‐based reconstruction to the base images before parameter fitting, the resulting parameter maps and synthetic images also exhibit less noise without any additional bias 74 . Networks have also been trained to jointly reconstruct the multiple contrast base images acquired by MDME, 75 which can reduce artifacts and improve SNR over standard parallel imaging techniques. These advanced reconstruction techniques increase the effective SNR efficiency of the MDME sequence, potentially reducing scan time and enabling new applications that were previously SNR challenged.…”

“…After an inversion pulse, four acquisitions are acquired in a similar manner to the Look-Locker technique. Cardiac imaging is possible when the five acquisitions are triggered, resulting in equally spaced acquisition times acquired by MDME,75 which can reduce artifacts and improve SNR over standard parallel imaging techniques. These advanced reconstruction techniques increase the effective SNR efficiency of the MDME sequence, potentially reducing scan time and enabling new applications that were previously SNR challenged.…”