GAN and dual‐input two‐compartment model‐based training of a neural network for robust quantification of contrast uptake rate in gadoxetic acid‐enhanced MRI

Abstract

Purpose: Gadoxetic acid uptake rate (k 1 ) obtained from dynamic, contrast-enhanced (DCE) magnetic resonance imaging (MRI) is a promising measure of regional liver function. Clinical exams are typically poorly temporally characterized, as seen in a low temporal resolution (LTR) compared to high temporal resolution (HTR) experimental acquisitions. Meanwhile, clinical demands incentivize shortening these exams. This study develops a neural network-based approach to quantitation of k 1 , for increased robustness over current models such as the linearized single-input, two-compartment (LSITC) model. Methods: Thirty Liver HTR DCE MRI exams were acquired in 22 patients with at least 16 min of postcontrast data sampled at least every 13 s. A simple neural network (NN) with four hidden layers was trained on voxel-wise LTR data to predict k 1 . Low temporal resolution data were created by subsampling HTR data to contain six time points, replicating the characteristics of clinical LTR data. Both the total length and the placement of points in the training data were varied considerably to encourage robustness to variation. A generative adversarial network (GAN) was used to generate arterial and portal venous inputs for use in data augmentation based on the dual-input, two-compartment, pharmacokinetic model of gadoxetic acid in the liver. The performance of the NN was compared to direct application of LSITC on both LTR and HTR data. The error was assessed when subsampling lengths from 16 to 4 min, enabling assessment of robustness to acquisition length. Results: For acquisition lengths of 16 min NRMSE (Normalized Root-Mean-Squared Error) in k 1 was 0.60, 1.77, and 1.21, for LSITC applied to HTR data, LSITC applied to LTR data, and GAN-augmented NN applied to LTR data, respectively. As the acquisition length was shortened, errors greatly increased for LSITC approaches by several folds. For acquisitions shorter than 12 min the GAN-augmented NN approach outperformed the LSITC approach to a statistically significant extent, even with HTR data. Conclusions: The study indicates that data length is significant for LSITC analysis as applied to DCE data for standard temporal sampling, and that machine learning methods, such as the implemented NN, have potential for much greater resilience to shortened acquisition time than directly fitting to the LSITC model.

Estimation of pharmacokinetic parameters from DCE‐MRI by extracting long and short time‐dependent features using an LSTM network

“…Another path to improvement generalizability is to incorporate temporal sampling intervals into the network input as another channel. 29 The performance of the current vanilla LSTM architecture could be further improved by using a bidirectional LSTM with attention at the expense of longer training and inference time. A further improved model could use more realistic synthetic data that takes motion artifacts and other factors into account to improve the robustness of performance of the LSTM on in vivo DCE-MRI datasets.…”

“…Further manipulation of the training dataset distribution and/or a modification of the loss function or a weighting training data in different parameter ranges could improve performance of the algorithm. Another path to improvement generalizability is to incorporate temporal sampling intervals into the network input as another channel 29 . The performance of the current vanilla LSTM architecture could be further improved by using a bidirectional LSTM with attention at the expense of longer training and inference time.…”