Impact of deep learning architectures on accelerated cardiac T-1 mapping using MyoMapNet

The objective of the current study was to investigate the performance of various deep learning (DL) architectures for MyoMapNet, a DL model for T-1 estimation using accelerated cardiac T-1 mapping from four T-1-weighted images collected after a single inversion pulse (Look-Locker 4 [LL4]). We implemented and tested three DL architectures for MyoMapNet: (a) a fully connected neural network (FC), (b) convolutional neural networks (VGG19, ResNet50), and (c) encoder-decoder networks with skip connections (ResUNet, U-Net). Modified Look-Locker inversion recovery (MOLLI) images from 749 patients at 3 T were used for training, validation, and testing. The first four T-1-weighted images from MOLLI5(3)3 and/or MOLLI4(1)3(1)2 protocols were extracted to create accelerated cardiac T-1 mapping data. We also prospectively collected data from 28 subjects using MOLLI and LL4 to further evaluate model performance. Despite rigorous training, conventional VGG19 and ResNet50 models failed to produce anatomically correct T-1 maps, and T-1 values had significant errors. While ResUNet yielded good quality maps, it significantly underestimated T-1. Both FC and U-Net, however, yielded excellent image quality with good T-1 accuracy for both native (FC/U-Net/MOLLI = 1217 +/- 64/1208 +/- 61/1199 +/- 61 ms, all p < 0.05) and postcontrast myocardial T-1 (FC/U-Net/MOLLI = 578 +/- 57/567 +/- 54/574 +/- 55 ms, all p < 0.05). In terms of precision, the U-Net model yielded better T-1 precision compared with the FC architecture (standard deviation of 61 vs. 67 ms for the myocardium for native [p < 0.05], and 31 vs. 38 ms [p < 0.05], for postcontrast). Similar findings were observed in prospectively collected LL4 data. It was concluded that U-Net and FC DL models in MyoMapNet enable fast myocardial T-1 mapping using only four T-1-weighted images collected from a single LL sequence with comparable accuracy. U-Net also provides a slight improvement in precision.