The aims of this work are (1) to explore deep learning (DL) architectures, spectroscopic input types, and learning designs toward optimal quantification in MR spectroscopy of simulated pathological spectra; and (2) to demonstrate accuracy and precision of DL predictions in view of inherent bias toward the training distribution. Methods: Simulated 1D spectra and 2D spectrograms that mimic an extensive range of pathological in vivo conditions are used to train and test 24 different DL architectures. Active learning through altered training and testing data distributions is probed to optimize quantification performance. Ensembles of networks are explored to improve DL robustness and reduce the variance of estimates. A set of scores compares performances of DL predictions and traditional model fitting (MF).Results: Ensembles of heterogeneous networks that combine 1D frequency-domain and 2D time-frequency domain spectrograms as input perform best. Dataset augmentation with active learning can improve performance, but gains are limited. MF is more accurate, although DL appears to be more precise at low SNR. However, this overall improved precision originates from a strong bias for cases with high uncertainty toward the dataset the network has been trained with, tending toward its average value. Conclusion:MF mostly performs better compared to the faster DL approach. Potential intrinsic biases on training sets are dangerous in a clinical context that requires the algorithm to be unbiased to outliers (i.e., pathological data). Active learning and ensemble of networks are good strategies to improve prediction performances. However, data quality (sufficient SNR) has proven as a bottleneck for adequate unbiased performance-like in the case of MF.

No abstract

PurposeThe inherently poor SNR of MRS measurements presents a significant hurdle to its clinical application. Denoising by machine or deep learning (DL) was proposed as a remedy. It is investigated whether such denoising leads to lower estimate uncertainties or whether it essentially reduces noise in signal‐free areas only.MethodsNoise removal based on supervised DL with U‐nets was implemented using simulated 1H MR spectra of human brain in two approaches: (1) via time‐frequency domain spectrograms and (2) using 1D spectra as input. Quality of denoising was evaluated in three ways: (1) by an adapted fit quality score, (2) by traditional model fitting, and (3) by quantification via neural networks.ResultsVisually appealing spectra were obtained; hinting that denoising is well‐suited for MRS. However, an adapted denoising score showed that noise removal is inhomogeneous and more efficient for signal‐free areas. This was confirmed by quantitative analysis of traditional fit results as well as DL quantitation following DL denoising. DL denoising, although apparently successful as judged by mean squared errors, led to substantially biased estimates in both implementations.ConclusionThe implemented DL‐based denoising techniques may be useful for display purposes, but do not help quantitative evaluations, confirming expectations based on estimation theory: Cramér Rao lower bounds defined by the original data and the appropriate fitting model cannot be circumvented in an unbiased way for single data sets, unless additional prior knowledge can be incurred in the form of parameter restrictions/relations or applicable substates.

Deep Learning has introduced the possibility to speed up quantitation in Magnetic Resonance Spectroscopy. However, questions arise about how to access and relate to prediction uncertainties. Distributions of predictions and Monte-Carlo dropout are here used to investigate data and model related uncertainties, exploiting ground truth knowledge (in-silico set up). It is confirmed that DL is a dataset-biased technique, showing higher uncertainties toward the edges of its training set. Surprisingly, metabolites present in high concentrations suffer from comparable high uncertainties as when present in low concentrations. Evaluating and respecting fitting uncertainties is equally crucial for DL and traditional approaches.

Multi-Echo Single-Shot (MESS) spectroscopy is tested in-vivo aiming at simultaneous determination of metabolite content and T2 times through simultaneous multi-parametric model fitting of partially sampled echoes. Cramer-Rao Lower Bounds (CRLBs) are used as measure of performances. The novel scheme was compared with the traditional Multi-Echo Multi-Shot (MEMS) method. Results confirmed former in-silico studies and indicate that MESS outperforms MEMS for simultaneous determinations of T2s and concentrations, with improvements ranging from 5-20% for T2s and 10-50% for concentrations.