Parameter estimation for WMTI‐Watson model of white matter using encoder–decoder recurrent neural network

Abstract

Purpose Biophysical modeling of the diffusion MRI (dMRI) signal provides estimates of specific microstructural tissue properties. Although non‐linear least squares (NLLS) is the most widespread fitting method, it suffers from local minima and high computational cost. Deep learning approaches are steadily replacing NLLS, but come with the limitation that the model needs to be retrained for each acquisition protocol and noise level. In this study, a novel fitting approach was proposed based on the encoder–decoder recurrent neural network (RNN) to accelerate model estimation with good generalization to various datasets. Methods The white matter tract integrity (WMTI)‐Watson model as an implementation of the Standard Model of diffusion in white matter derives its parameters indirectly from the diffusion and kurtosis tensors (DKI). The RNN‐based solver, which estimates the WMTI‐Watson model from DKI, is therefore more readily translatable to various data, irrespective of acquisition protocols as long as the DKI was pre‐computed from the signal. An embedding approach was also used to render the model insensitive to potential differences in distributions between training data and experimental data. The analytical solution, NLLS, RNN‐, and a multilayer perceptron (MLP)‐based methods were evaluated on synthetic and in vivo datasets of rat and human brain. Results The proposed RNN solver showed highly reduced computation time over the analytical solution and NLLS, with similar accuracy but improved robustness, and superior generalizability over MLP. Conclusion The RNN estimator can be easily applied to various datasets without retraining, which shows great potential for a widespread use.

Profiling specific cell populations within the inflammatory tumor microenvironment by oscillating-gradient diffusion-weighted MRI

“…17 24 Recently, also deep learning-based methods have been applied for DWI-based biophysical modeling, 47 which may further improve accuracy of IMPULSED-derived microstructural parameters, combined with highly reduced computation time. 48 Difficulties of translation to human studies have already been discussed by Jiang et al, 24 mainly revolving around hardware limitations with lower gradient strengths of clinical MR scanners compared with small animal MRI setups. However, Xu et al introduced a clinically feasible IMPULSED approach with modified gradient waveforms.…”

Nomograms for Predicting the Risk and Prognosis of Liver Metastases in Pancreatic Cancer: A Population-Based Analysis

“…Deep learning evolved from the study of artificial neural networks; however, it is not identical to conventional neural networks. Nevertheless, in terms of vocabulary, the many deep learning algorithms, including deep reinforcement learning, generative adversarial networks, recurrent neural networks, and convolutional neural networks, use the phrase “neural network” [ 47 , 48 , 49 , 50 ]. Deep learning can be thought of as a semi-theoretical, semi-empirical modelling approach that employs human understanding of mathematics and computer algorithms, along with as much training information as is possible, to construct an architectural framework, utilizing the massive computing power of computers to tune the internal criteria to approximate the issue’s objectives as closely as possible [ 51 ].…”

Subject classification and cross-time prediction based on functional connectivity and white matter microstructure features in a rat model of Alzheimer’s using machine learning

“…Going forward, the acquisition of multi-shell diffusion MRI data (at least two non-zero b -values, e.g. b =1000 and 2500 s/mm 2 ) in clinical studies of dementia or other brain diseases is highly recommended to enable the estimation of DKI metrics brain-wide, and of WM microstructure features using the WMTI-Watson model, for which analysis code is readily available (Diao and Jelescu, 2023).…”