Next Article in Journal
Comparison between Single-Dose and Two-Dose Psilocybin Administration in the Treatment of Major Depression: A Systematic Review and Meta-Analysis of Current Clinical Trials
Previous Article in Journal
Epidemiology of Sports-Related Concussion in Japanese University Soccer Players
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Improving the Sensitivity of Task-Based Multi-Echo Functional Magnetic Resonance Imaging via T2* Mapping Using Synthetic Data-Driven Deep Learning

Department of Electronic Science, Xiamen University, Xiamen 361005, China
*
Author to whom correspondence should be addressed.
Brain Sci. 2024, 14(8), 828; https://doi.org/10.3390/brainsci14080828
Submission received: 15 May 2024 / Revised: 30 July 2024 / Accepted: 16 August 2024 / Published: 17 August 2024
(This article belongs to the Section Computational Neuroscience and Neuroinformatics)

Abstract

:
(1) Background: Functional magnetic resonance imaging (fMRI) utilizing multi-echo gradient echo-planar imaging (ME-GE-EPI) has demonstrated higher sensitivity and stability compared to utilizing single-echo gradient echo-planar imaging (SE-GE-EPI). The direct derivation of T2* maps from fitting multi-echo data enables accurate recording of dynamic functional changes in the brain, exhibiting higher sensitivity than echo combination maps. However, the widely employed voxel-wise log-linear fitting is susceptible to inevitable noise accumulation during image acquisition. (2) Methods: This work introduced a synthetic data-driven deep learning (SD-DL) method to obtain T2* maps for multi-echo (ME) fMRI analysis. (3) Results: The experimental results showed the efficient enhancement of the temporal signal-to-noise ratio (tSNR), improved task-based blood oxygen level-dependent (BOLD) percentage signal change, and enhanced performance in multi-echo independent component analysis (MEICA) using the proposed method. (4) Conclusion: T2* maps derived from ME-fMRI data using the proposed SD-DL method exhibit enhanced BOLD sensitivity in comparison to T2* maps derived from the LLF method.

1. Introduction

The gradient echo-planar imaging (GE-EPI) technique [1,2,3], known for its high sensitivity to T2* variation indicative of blood oxygen level-dependent (BOLD) change, has been widely utilized in functional magnetic resonance imaging (fMRI). It is crucial to achieve a balance between signal-to-noise ratio and sensitivity to the BOLD effect across the entire brain range in collecting T2*-weighted images in fMRI. Typically, the optimal BOLD effect is achieved when the echo time (TE) of imaging sequence equals the T2* value [4]. However, finding a global optimal TE value for the best T2* weighting is challenging due to individual and regional brain differences. In these cases, using multi-echo gradient echo-planar imaging (ME-GE-EPI) with different TEs presents a suitable choice, albeit at the expense of spatio-temporal resolution. Multi-echo fMRI (ME-fMRI), typically acquired using ME-GE-EPI, has proven effective in optimizing BOLD contrast by amalgamating time courses of various TEs using optimized weighting schemes [4,5,6], leading to enhanced BOLD-contrast sensitivity and decreased artifacts.
Furthermore, fMRI is susceptible to physical and physiological interferences, such as head motion [7,8,9], scanner instability [10], respiration [11,12], heartbeat [13,14], cerebrospinal fluid [15], and the pooling of blood in veins [16], which can introduce non-BOLD interference signals. ME-fMRI can eliminate non-BOLD noise by leveraging the TE dependency of the BOLD signal. The performance of ME-fMRI can be further improved with multi-echo independent component analysis (MEICA) [17]. By identifying the reliance of independent components on TE, MEICA can differentiate between BOLD and non-BOLD signals. It has been demonstrated that MEICA efficiently reduces the interferences generated by head movements [18,19]. Additionally, MEICA does not affect the analysis of the default mode network [19], and it improves the identification of the brain network in resting-state fMRI of humans at 3T [17] and rats at 11.7T [20].
The T2* map estimated by ME-fMRI can improve the comparability of fMRI studies across multiple scanners and imaging centers. Compared to echo combination methods, the T2* map exhibits higher BOLD sensitivity [21]. Moreover, the T2* map can be incorporated into the weighting coefficient calculation to determine the echo combination strategy [4,17,18] and to aid in anatomical–functional registration [22].
In ME-fMRI, traditional voxel-wise fitting methods (e.g., log-linear fitting, LLF) are usually used to obtain T2* according to the exponential decay model [23,24,25,26,27]. However, the noise acquired during magnetic resonance signal acquisition cannot be accurately estimated or removed via voxel-by-voxel fitting under limited echoes. Furthermore, numerical instability during the fitting may magnify noise [24]. The abovementioned issues result in poor-quality T2* maps, leading to compromised performance of ME-fMRI in brain functional analysis. Consequently, researchers still utilize T2*-weighted images [6,24,28,29,30] rather than quantitative T2* maps in fMRI analysis.
In recent years, fMRI analysis has benefited from the emergence of deep learning. In delving into the application of supervised deep learning in the medical domain, a pivotal challenge lies in its heavy reliance on abundant sample–label pairs. Theoretically, an expansion of the training sample size could amplify the model’s generalization capacity, which is paramount for refining diagnostic precision and predictive accuracy. Nevertheless, in the realm of medical practice, the pursuit of large-scale, precisely annotated sample–label pairs is fraught with obstacles. The vast variability in MRI experimental protocols and the intricacies of extracting representative samples from complex real-world settings conspire to make the scarcity of samples an inherent challenge in medical data analysis. This limitation impedes the effective utilization of traditional supervised learning approaches during data preprocessing. To overcome this challenge, the use of synthetic data-driven deep learning techniques has emerged as a promising avenue of research. These techniques could generate a substantial number of simulated samples, thereby compensating for the scarcity of real-world data. And synthetic data-driven training has been increasingly utilized in deep learning-based MRI reconstruction [31,32,33].
In this work, we propose a synthetic data-driven deep learning method (SD-DL) for ME-fMRI T2* mapping and compare the effects of LLF-fitted and SD-DL-fitted T2* maps on BOLD sensitivity. We generated synthetic training data and corresponding labels using the MR signal model and two-dimensional virtual parametric maps. Then, the model used in SD-DL was subjected to supervised training using the synthetic data and corresponding labels to remove non-BOLD noise and fit the MR parameters. Unlike voxel-by-voxel fitting, SD-DL performs slice-by-slice fitting. On the one hand, a large amount of prior information about the brain T2* map can be learned through synthetic data. On the other hand, spatial correlation can be efficiently utilized to obtain accurate T2* maps. The proposed SD-DL method is validated using a public dataset and a case including in-house data.

2. Materials and Methods

2.1. Multi-Echo fMRI Signal Model

The fMRI signal obeys the mono-exponential decay model, in which the acquired fMRI signal of a voxel can be described as:
S t = S 0 · e t T 2 * + ε = S 0 · e t · R 2 * + ε S 0 · e t · R 2 *
where S(t) is the acquired fMRI signal at time t, S0 is the non-BOLD signal related to proton density, T2* is the transverse relaxation time, R2* is the inverse of T2*, and ε is noise.
In ME-fMRI, T2* and S0 can be solved by LLF:
log S 0 R 2 * = p i n v 1 , T E 1 1 , T E 2 1 , T E n × log S T E 1 log S T E 2 log S T E n
where p i n v ( · ) is a pseudo-inverse operator.
Meanwhile, the voxel-wise linear combination of each echo can be calculated. In this study, we use the T2*-weighted combination [4,17] to investigate the differences in performance between LLF and the proposed SD-DL in T2* fitting. The T2*-weighted combination is referred to as echo combination (EC) in the following. The weighting factor of EC is as follows:
w n E C = T E n * exp ( T E n T ^ 2 * ) i = 1 N T E i * exp ( T E i T ^ 2 * )
where n indicates the nth echo, N refers to the total number of echoes, and T ^ 2 * refers to the voxel-wise estimation of T2*.

2.2. Data Processing Pipeline

Figure 1 shows the processing pipeline of ME-fMRI data. The preprocessing was carried out on Matlab (R2021, MathWorks, Natick, MA, USA) and through using the Statistical Parametric Mapping (SPM) 12 toolbox [34]. The flowchart in Figure 1a gives an overview of ME-fMRI data preprocessing. The steps were as follows: In the data preprocessing stage (Figure 1b), realignment parameters were first calculated using ME-fMRI data, and then, slice timing correction and realignment correction were carried out. Subsequently, six time series were calculated based on the realignment-corrected time series (Figure 1c). Two four-dimensional (4D) T2* fMRI time series were fitted by LLF and SD-DL, respectively. EC and MEICA-denoised EC time series were calculated with T2* estimated by LLF, and SD-DL-EC and MEICA-denoised SD-DL-EC time series were calculated with T2* estimated by SD-DL. Finally, indicator calculation and statistical analysis were conducted on six different time series to compare LLF with SD-DL in reflecting brain function (Figure 1d). After the multi-echo combination was completed, spatial smoothing (Gaussian kernel with 7 mm full-width at the half of the maximum, FWHM) and first-level Generalized Linear Model (GLM) analysis were applied to all the time series to extract the BOLD signals. For the rest of this paper, the T2* fMRI time series fitted by LLF and SD-DL are referred to as SLLF and SSD-DL, respectively. The EC and MEICA-denoised EC time series are referred to as SEC and SMEICA, respectively. The SD-DL-EC and MEICA-denoised SD-DL-EC time series are referred to as SSD-DL-EC and SSD-DL-MEICA, respectively.

2.3. T2* Mapping and Multi-Echo Combination

Supervised learning was used for SD-DL T2* mapping with a 4-layer U-net model. The overall framework is shown in Figure 2. First, two-dimensional (2D) parametric templates (M0 and T2*) were produced from the public database IXI (https://brain-development.org/ixi-dataset/ (accessed on 15 August 2024)) by the method described by Yang [32] and were down-sampled to a matrix of 64 × 64 to match the matrix size of real data. Then, synthetic multi-echo MR data with TE equal to the real data were generated through Equation (1). Randomized Rician noise was added to the signal magnitude of synthetic data with a standard deviation of 0–5% to mimic a real situation. Additionally, data augmentation was carried out to augment the number of training samples (i.e., random rotation and flip [35]).
The synthetic data were fed into the U-net for model training, and the corresponding parametric templates were used as training labels. Subsequently, the T2* maps of real data were estimated on the trained U-net model. Overall, 5400 synthesized samples were randomly divided into a training set (4500 samples) and a validation set (900 samples). The training and validation sets were derived from different subjects to avoid data leakage. The random cropping technique was used during model training to improve the model’s generalization ability.
For comparison, SMEICA and SEC were calculated with the Tedana toolbox [17,36,37,38]. To calculate SSD-DL-EC, the preprocessed time series was firstly averaged over the time dimension, and then the 3D T2* maps were subjected to slice-by-slice estimation with SD-DL; secondly, the 3D SD-DL-estimated T2* maps were used voxel by voxel as T ^ 2 * in Equation (3) to calculate the weighting factor of SSD-DL-EC.
For SD-DL, the hyperparameters of the model were set as follows: batch size = 16; random crop size = 32 × 32; optimizer = Adam; and learning rate = 10 4 . Upon reaching the lowest loss on the test set, the model was considered fully trained. Synthetic data generation, model training, and pipeline testing were carried out using Python 3.6 in the Pycharm library. All processes were performed with an NVIDIA GeForce RTX 2080Ti GPU.

2.4. Data Collection

The study was approved by the institutional research ethics committees, and written informed consent was obtained from all individual participants before the experiments. We conducted experiments based on the real-time multi-echo fMRI (rt-ME-fMRI) dataset [39]. The rt-ME-fMRI dataset, which contains data from 28 participants (20 male and 8 female; mean age = 24.9 (years), SD = 4.7), was accessed via a data use agreement. After excluding corrupted data, the fMRI and physiological data of 27 participants were used in the study. Task-based data were collected using an ON/OFF design, starting from OFF, and each condition lasted for 10 volumes. For each participant, three-echo fMRI data and cardiac and respiratory fluctuations were used for analysis. Cardiac and respiratory fluctuations were sampled at 500 Hz. In GLM analysis, a total of 20 regression variable parameters were used, including 6 realignment parameters and 14 RETROICOR [14] regressors. The RETROICOR regressors were estimated with the TAPAS PhysIO toolbox [40] (both cardiac and respiratory to the 2nd order, excluding interaction regressors). We also collected in-house ME-fMRI data from a healthy volunteer (male, 27-year-old) to further evaluate the performance of the proposed method.

2.5. MRI Protocol

Furthermore, rt-ME-fMRI dataset images were acquired using a 3T Philips Achieva scanner with a 32-channel head coil [39]. For each volunteer, anatomical images were acquired using a 3D gradient echo sequence, while fMRI scans were conducted using a multi-echo EPI sequence. The parameters used were as follows:
Anatomical MRI: time of repetition (TR) = 8.2 ms; TE = 3.75 ms; pulse flip angle = 8°; field of view (FOV) = 240 × 240 × 180 mm3; and resolution = 1 × 1 × 1 mm3. Functional MRI: TR = 2000 ms; TEs = 14, 28, 42 ms (3 echoes); number of volumes = 210 (excluding 5 dummy volumes discarded by the scanner); total scan time = 7 min (excluding 5 dummy volumes); pulse flip angle = 90°; FOV = 224 × 224 × 119 mm3; resolution = 3.5 × 3.5 × 3.5 mm3; in-plane matrix size = 64 × 64; number of slices = 34; interslice gap = 0 mm; slice orientation = oblique; slice order/direction = sequential/ascending; and SENSE acceleration factor = 2.5.
In-house fMRI data were acquired using a whole-body MRI scanner at 3T (Prisma; Siemens Healthcare, Erlangen, Germany) with a 20-channel head coil. The acquisition parameters were as follows: resolution = 3.5 × 3.5 × 3.5 mm3; TR = 2500 ms; TEs = 17, 42, 67 ms (3 echoes); slice number = 32; GRAPPA acceleration factor = 2; and number of volumes = 120 (task-based) and 160 (resting-state). The other imaging parameters were the same as those used for the rt-ME-fMRI dataset. A conventional block-design visual stimulation task with “ON-OFF” blocks (30 s “checkerboard”, followed by a 30 s black field) was performed, in which the “ON” block was used to introduce visual stress. The participant was asked to lay down in a relaxed position in the MRI scanner to minimize noise signals. For comparison, a conventional 2D multi-echo gradient echo sequence was also acquired as the ground truth (GT) of resting state for in-house ME-fMRI data (TR = 3500 ms; TEs = 10, 23, 36, 50, 57.5, and 65 ms; and resolution = 3.5 × 3.5 × 3.5 mm3).

2.6. Evaluation Indicators

To evaluate our results, mean square error (MSE) analysis was performed on simulation samples under different levels of Rician noise. A temporal signal-to-noise ratio (tSNR) map was calculated by dividing the voxel-wise time series mean by the voxel-wise time series standard deviation (SD) following the elimination of the linear and quadratic trends with the fMRwhy toolbox [41]. To evaluate the sensitivity of SLLF and SSD-DL, activation maps (T-value maps) and BOLD response curves of the maximum T-value voxel were illustrated. To evaluate the rt-ME-fMRI data, in addition to the tSNR, the mean of percentage signal change (PSC) [42], T-value, and functional contrast (FC) were illustrated in a target region of interest (ROI).
PSC was used to quantify BOLD sensitivity, which was computed as follows:
P S C = 100 × S F × β ^ c o n d i t i o n   β ^ c o n s t a n t
where β ^ c o n d i t i o n and β ^ c o n s t a n t are parameter maps of condition and constant from GLM analysis, respectively, and the scaling factor (SF) is the maximum value of a reference trial at the resolution of the super-sampled design matrix. The average T-values were used as a proxy for the contrast-to-noise ratio and computed within significantly active voxels inside target ROIs. FC was used as a proxy for signal sensitivity, and for the preprocessed time series, the removal of linear and quadratic trends was first performed, followed by spatial smoothing and GLM analysis and, lastly, spatial averaging within the target ROIs. Subsequently, the mean of the time series in the “ON” condition was subtracted from the mean of the time series in the “OFF” condition to obtain FC.
To avoid scan-specific bias, a binary mask with p < 0.001, i.e., the union of family-wise error (UFWE) regions of six time series, was used for statistical analysis. This binary mask will hereafter be referred to as the UFWE ROI. A two-tailed paired t-test was applied to the results, and Cohen’s d was calculated to identify significant differences, as well as the effect size, between different pipelines.

3. Results

3.1. Results of Simulation Experiments

Figure 3 illustrates the comparative results of the MSEs for the LLF method and SD-DL method across five distinct levels of Rician noise contamination. In the training phase of the SD-DL neural network, a stochastic Rician noise intensity varying from 0% to 5% standard deviation was introduced into the synthesized dataset. For the MSE evaluation, synthetic samples embodying Rician noise at five discrete levels, ranging from 1% to 5%, were produced. The progression depicted in Figure 3a–e revealed a shared trend wherein the MSE values for both SD-DL and LLF escalated in conjunction with the increase in noise magnitude in the synthetic data. Notably, the average rise in MSE for LLF surpassed that of SD-DL. Furthermore, SD-DL consistently achieved significantly reduced MSEs relative to LLF across all examined noise intensity tiers, highlighting its superior denoising efficiency.

3.2. Results of Visual Stimulation Task Data

An individual subject’s resting-state T2* maps, derived from the in-house data, are shown in Figure 4a, with a T2* range of 0–200 ms. After averaging the 4D LLF and SD-DL-fitting resting-state T2* maps in the time dimension, 3D LLF and SD-DL-fitting resting-state T2* maps were obtained. In contrast to LLF, SD-DL was able to eliminate the noise in the T2* maps while preserving the textural features. tSNR maps of resting-state data are shown in Figure 4b. Compared to SLLF (42.96), the whole brain tSNR of SSD-DL (46.11) was 7.33% higher. Activation maps (T-value maps) and BOLD curves of SLLF and SSD-DL are displayed in Figure 4c. The activation slice with the greatest T-value voxel is displayed on the left side of Figure 4c. Compared to SLLF, SSD-DL exhibited a higher mean T-value. The voxel with the highest T-value yielded the BOLD response curve on the right. For both LLF and SD-DL, the BOLD response curves were in line with the expected curves. The SD-DL result showed a larger amplitude fluctuation (i.e., higher sensitivity) compared to that for the LLF method.

3.3. Results of rt-ME-fMRI Dataset

3.3.1. tSNR

With the goal of eliminating individual bias and providing a more quantitative view than the individual subject tSNR (Figure 4b), whole-brain mean tSNR was calculated directly from the 4D preprocessed time series without spatial smoothing using rest_run-1 data in the rt-ME-fMRI dataset. Figure 5 illustrates the whole-brain mean tSNR distribution of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. In all the time series, the whole-brain mean tSNR of SSD-DL was higher than that of SLLF. The whole-brain mean tSNR for SSD-DL (38.68) was 1.47% higher than that for SLLF (38.12), with p < 0.0001 and d = 2.11, and that for SSD-DL-EC (116.9) was 9.77% higher than SEC (106.5), with p < 0.0001 and d = 3.40, while the whole-brain mean tSNR for SSD-DL-MEICA (140.6) exceeded that for SMEICA (123.5) by 13.85%, with p < 0.0001 and d = 2.43.

3.3.2. Percentage Signal Change

The whole-brain PSC was calculated for each subject based on the fingerTapping task according to Equation (4). To quantify BOLD sensitivity equitably, the mean PSC distributions of all the time series within the UFWE were calculated. Figure 6 illustrates the mean PSC distributions within the UFWE of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA.
As shown in Figure 6, the T2* time series exhibited higher PSCs, while the mean PSCs of SLLF and SSD-DL were much higher than those of the other four methods. Specifically, for the fingerTapping task (Figure 6a), SSD-DL (1.863) had a 2.59% higher PSC than SLLF (1.816), with p < 0.01 and d = 0.65, while that of SSD-DL-EC (0.8117) exceeded that of SEC (0.7682) by 5.66%, with p < 0.0001 and d = 1.56, and that of SSD-DL-MEICA (0.8142) exceeded that of SMEICA (0.7677) by 6.06%, with p < 0.01 and d = 0.69. For the fingerTappingImagined task (Figure 6b), SSD-DL (1.173) had a 5.01% higher PSC than SLLF (1.117), with p < 0.001 and d = 0.78, while that of SSD-DL-EC (0.586) exceeded that of SEC (0.558) by 5.02%, with p < 0.01 and d = 0.65, and that of SSD-DL-MEICA (0.566) exceeded that of SMEICA (0.535) by 5.79%, with p < 0.05 and d = 0.48. For the emotionProcessing task (Figure 6c), the t-test of PSCs between SLLF (1.069) and SSD-DL (1.020), SEC (0.704) and SSD-DL-EC (0.706), SMEICA (0.713) and SSD-DL-MEICA (0.722) were not significantly different. For the emotionProcessingImagined task (Figure 6d), SSD-DL (1.230) had a 4.06% higher PSC than SLLF (1.182), with p < 0.0001 and d = 1.00, while that of SSD-DL-EC (0.683) exceeded that of SEC (0.666) by 2.55%, with p < 0.0001 and d = 1.10, and that of SSD-DL-MEICA (0.642) exceeded that of SMEICA (0.614) by 4.56%, with p < 0.0001 and d = 0.92.

3.3.3. T-Values

T-values were calculated after GLM analysis within the UFWE using fingerTapping data in rt-ME-fMRI. Figure 7 illustrates the distribution of mean T-values (over all participants) of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA within the UFWE.
As shown in Figure 7, for the fingerTapping task (Figure 7a), the average of mean T-value of SSD-DL (7.065) was 5.62% higher than that of SLLF (6.689), with p < 0.0001 and d = 1.92, while the t-test of the averages of the mean T-values between SEC (7.724) and SSD-DL-EC (7.711), SMEICA (8.063), and SSD-DL-MEICA (8.000) were not significantly different. For the fingerTappingImagined task (Figure 7b), the average of the mean T-values of SSD-DL (5.516) was 6.90% higher than that of SLLF (5.160), with p < 0.0001 and d = 1.38, while, according to t-test, the averages of the mean T-values between SEC (6.349) and SSD-DL-EC (6.313), SMEICA (6.778), and SSD-DL-MEICA (6.861) were not significantly different. For the emotionProcessing task (Figure 7c), the average of the mean T-values of SSD-DL (5.789) was 8.88% higher than that of SLLF (5.317), with p < 0.0001 and d = 1.43, and SEC (7.221) had a 1.66% higher average of mean T-values than SSD-DL = EC (7.104), with p < 0.001 and d = 0.76, while, according to t-test, the averages of the mean T-values between SMEICA (7.622) and SSD-DL-MEICA (7.553) were not significantly different. For the emotionProcessingImagined task (Figure 7d), the average of the mean T-values of SSD-DL (5.265) was 6.62% higher than that of SLLF (4.938), with p < 0.0001 and d = 2.18, and SSD-DL-MEICA (6.974) exceeded SMEICA (6.754) by 3.26%, with p < 0.05 and d = 0.51, while the results of the t-test of the averages of the mean T-values between SEC (6.459) and SSD-DL-EC (6.454) were not significantly different.

3.3.4. Functional Contrast

FC was calculated within the UFWE ROI using fingerTapping data in rt-me-fMRI. Figure 8 illustrates the distribution of FC of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA within the UFWE.
As shown in Figure 8, both SLLF and SSD-DL demonstrated a higher average FC than SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. Specifically, for the fingerTapping task (Figure 8a), SLLF (1.330) exceeded SSD-DL (1.300) by 2.31%, with p < 0.05 and d = 0.52; SSD-DL-EC (0.591) exceeded SEC (0.569) by 3.86%, with p < 0.0001 and d = 1.12. Meanwhile, the t-test results for SMEICA (0.532) and SSD-DL-MEICA (0.538) were not significantly different. For the fingerTappingImagined task (Figure 8b), SSD-DL-EC (0.472) exceeded SEC (0.448) by 5.36%, with p < 0.05 and d = 0.47, while the t-test results for SLLF (1.023) and SSD-DL (0.982), SMEICA (0.401), and SSD-DL-MEICA (0.415) were not significantly different. For the emotionProcessing task (Figure 8c), SLLF (1.175) exceeded SSD-DL (1.086) by 8.20%, with p < 0.05 and d = 0.46. The t-test results for SEC (0.549) and SSD-DL-EC (0.547), SMEICA (0.517), and SSD-DL-MEICA (0.515) were not significantly different. For the emotionProcessingImagined task (Figure 8d), SLLF (1.247) exceeded SSD-DL (1.200) by 3.92%, with p < 0.01 and d = 0.59; SSD-DL-EC (0.614) exceeded SEC (0.600) by 2.33%, with p < 0.0001 and d = 1.15. The t-test results between SMEICA (0.505) and SSD-DL-MEICA (0.503) were not significantly different.

4. Discussion

In this study, we have presented an SD-DL method for T2* mapping from multi-echo functional magnetic resonance imaging (ME-fMRI) datasets. The efficacy of SD-DL was examined. Human brain templates derived from publicly available datasets were used to synthesize training data. To enhance the neural network model’s ability to recognize noise in real ME-fMRI data, a controlled level of noise was introduced to the synthetic data. The BOLD sensitivity and image quality between SD-DL and traditional voxel-by-voxel LLF methods were compared.
Theoretically, LLF represents the gold standard for T2* estimation in fMRI devoid of noise contamination. However, simulation experiments have revealed that the MSE of LLF is significantly greater than that observed with the SD-DL when performing T2* mapping based on signals that added noises. This disparity suggests that SD-DL exhibits superior noise resilience over LLF, thereby enabling more effective handling of fMRI datasets characterized by diverse noise profiles. This enhanced noise robustness of SD-DL is further corroborated by a noticeable improvement in tSNR.
Regarding BOLD sensitivity, the T2* time series exhibits higher PSC and FC values compared to EC and MEICA, indicating that the T2*-based time series possesses higher sensitivity. Notably, the SD-DL T2* time series achieves higher PSC and comparable FC levels to those derived from LLF, implying an augmented sensitivity through SD-DL application. Furthermore, SD-DL-EC outperforms EC in both PSC and FC, and SD-DL-MEICA outperforms MEICA in PSC and displays a similar level of FC, suggesting SD-DL is capable of improving the performance of EC and MEICA. Additionally, SD-DL improves the mean T-value compared to LLF, indicating enhanced BOLD sensitivity without compromising specificity. Fundamentally, the BOLD effect arises from neuronal activity-induced fluctuations in the local magnetic field, subsequently altering the fMRI signal. Given that T2* is more responsive to such magnetic field changes, it inherently has a higher BOLD sensitivity. Nonetheless, in routine fMRI investigations, T2* mapping is challenged by noise contamination, limiting its broad applicability. SD-DL, on the other hand, can effectively mitigate the noise interference, consequently improving the BOLD sensitivity.
While T2* time series, SEC, and SMEICA have distinct advantages or disadvantages in ME-fMRI across various indicators, SD-DL consistently shows superior performance in most indicators for different types of time series. Moreover, even if the data have different matrix sizes, resolutions, or sets of echo times, simulation samples can be generated and a neural network can be trained through the SD-DL framework. Although deep learning is sensitive to the training set and training method, we use synthetic data to minimize its dependence on real data and the risk of overfitting, and at the same time, we use a simpler network structure involving fewer hyperparameters, and these manipulations make SD-DL higher applicable and reproducible. Moreover, the introduction of a known level of noise to the synthetic samples enables the generation of data that more closely resemble actual fMRI data. Additionally, U-net possesses a large receptive field, allowing it to capture a lot of information in the image, while jump connections ensure that the convolutional neural network can maximize the preservation of the original information during deep processing and effectively realize the denoising function, which further enhances the accuracy and reliability of T2* mapping. In contrast, the LLF method is unable to account for the presence of noise when performing T2* mapping, which may result in amplified fitting errors due to numerical uncertainties.
Nevertheless, our proposed method for ME-fMRI has some limitations. The noise level introduced to SD-DL in the generation of simulation samples is an a priori knowledge that may result in suboptimal model performance if the added noise level is significantly different from the true noise level. The SD-DL T2* time series may not significantly improve tSNR due to the limited acquisition of three echoes, hindering high-accuracy T2* map reconstruction. Future studies should explore methods for obtaining more reliable T2* quantification in MR images under various means of accelerating imaging, potentially through the introduction of novel T2* quantitative imaging sequences [43].
Real-time fMRI-based studies have higher BOLD sensitivity requirements, and increasing the whole-brain mean tSNR is beneficial for real-time brain wide connectivity measurements [21]. Compared to that derived from the LLF method, the SD-DL-fitted T2* map has higher BOLD sensitivity and could be developed for adaptive paradigms and neurofeedback research.
The quality of synthetic data plays a crucial role in T2* map reconstruction. More precise modeling of non-ideal factors such as head movements in training data synthesis can enhance the generation of higher-quality T2* maps. This work highlights that SD-DL effectively reduces data noise while maintaining sensitivity similar to the LLF method, offering a new perspective for accurate T2* fitting in ME-fMRI data.
SD-DL could be applied to T2* mapping based on ME-fMRI data by adhering to the following steps: Firstly, generate synthetic samples using M0 and T2* parameter templates based on the parameters of real-world data and the training model. Then, feed the preprocessed ME-fMRI data into the trained model to obtain T2* maps. Since SD-DL does not require the use of real-world training samples, it is possible to train the network so that it can complete T2* mapping on its own by only knowing the parameters of the acquired real-world data. Thus, SD-DL can be better integrated into the traditional multi-echo fMRI preprocessing process.

5. Conclusions

In this study, we proposed a ME-fMRI T2* mapping technique called SD-DL for accurate T2* map reconstruction from ME-fMRI data and compared the effects of LLF-fitted and SD-DL-fitted T2* maps on BOLD sensitivity. SD-DL utilizes deep learning to model the mapping relationship between multi-echo fMRI signals and T2* maps. By leveraging the powerful feature extraction and nonlinear mapping capabilities of deep neural networks, SD-DL is able to effectively capture the fundamental physical properties of the MRI acquisition process while accomplishing denoising, thereby improving robustness and generalizability. Compared to LLF-fitted T2* maps, SD-DL-fitted T2* maps have higher BOLD sensitivity, which helps to investigate the neural mechanisms of cognitive processes, brain diseases, and therapeutic interventions in more detail, meaning that the proposed method has a promising future in adaptive paradigms and neurofeedback research.

Author Contributions

Conceptualization, Y.Z. and Q.Y.; methodology, Y.Z. and Q.Y.; software, Y.Z.; validation, Y.Z. and Q.Y.; formal analysis, Y.Z.; investigation, Y.Z. and Q.Y.; resources, Y.Z. and Q.Y.; data curation, Y.Z.; writing—original draft preparation, Y.Z.; writing—review and editing, Q.Y., S.Q., J.D., S.C. and C.C.; visualization, Y.Z.; supervision, S.C. and C.C.; project administration, S.C. and C.C.; funding acquisition, S.C., Z.C. and C.C. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the National Natural Science Foundation of China via grant numbers 82071913, 12375291, and 62331021.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the institutional research ethics committees.

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The rt-ME-fMRI dataset (https://dataverse.nl/dataverse/rt-me-fmri (accessed on 15 August 2024)) [39] used in this study can be accessed via a data use agreement. The code is publicly available on https://github.com/yinghezhao/SD-DL (accessed on 15 August 2024).

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Vaidya, M.V.; Lazar, M.; Deniz, C.M.; Haemer, G.G.; Chen, G.; Bruno, M.; Sodickson, D.K.; Lattanzi, R.; Collins, C.M. Improved detection of fMRI activation in the cerebellum at 7T with dielectric pads extending the imaging region of a commercial head coil. J. Magn. Reson. Imaging 2018, 48, 431–440. [Google Scholar] [CrossRef] [PubMed]
  2. Gao, J.H.; Miller, I.; Lai, S.; Xiong, J.; Fox, P.T. Quantitative assessment of blood inflow effects in functional MRI signals. Magn. Reson. Med. 1996, 36, 314–319. [Google Scholar] [CrossRef]
  3. Logothetis, N.K. MR imaging in the non-human primate: Studies of function and of dynamic connectivity. Curr. Opin. Neurobiol. 2003, 13, 630–642. [Google Scholar] [CrossRef]
  4. Posse, S.; Wiese, S.; Gembris, D.; Mathiak, K.; Kessler, C.; Grosse-Ruyken, M.-L.; Elghahwagi, B.; Richards, T.; Dager, S.R.; Kiselev, V.G. Enhancement of BOLD-contrast sensitivity by single-shot multi-echo functional MR imaging. Magn. Reson. Med. 1999, 42, 87–97. [Google Scholar] [CrossRef]
  5. Poser, B.A.; Versluis, M.J.; Hoogduin, J.M.; Norris, D.G. BOLD contrast sensitivity enhancement and artifact reduction with multiecho EPI: Parallel-acquired inhomogeneity-desensitized fMRI. Magn. Reson. Med. 2006, 55, 1227–1235. [Google Scholar] [CrossRef]
  6. Poser, B.A.; Norris, D.G. Investigating the benefits of multi-echo EPI for fMRI at 7 T. Neuroimage 2009, 45, 1162–1172. [Google Scholar] [CrossRef] [PubMed]
  7. Power, J.D.; Barnes, K.A.; Snyder, A.Z.; Schlaggar, B.L.; Petersen, S.E. Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. Neuroimage 2012, 59, 2142–2154. [Google Scholar] [CrossRef]
  8. Satterthwaite, T.D.; Wolf, D.H.; Loughead, J.; Ruparel, K.; Elliott, M.A.; Hakonarson, H.; Gur, R.C.; Gur, R.E. Impact of in-scanner head motion on multiple measures of functional connectivity: Relevance for studies of neurodevelopment in youth. Neuroimage 2012, 60, 623–632. [Google Scholar] [CrossRef] [PubMed]
  9. Van Dijk, K.R.A.; Sabuncu, M.R.; Buckner, R.L. The influence of head motion on intrinsic functional connectivity MRI. Neuroimage 2012, 59, 431–438. [Google Scholar] [CrossRef]
  10. Smith, A.M.; Lewis, B.K.; Ruttimann, U.E.; Ye, F.Q.; Sinnwell, T.M.; Yang, Y.H.; Duyn, J.H.; Frank, J.A. Investigation of low frequency drift in fMRI signal. Neuroimage 1999, 9, 526–533. [Google Scholar] [CrossRef]
  11. Pais-Roldán, P.; Biswal, B.; Scheffler, K.; Yu, X. Identifying Respiration-Related Aliasing Artifacts in the Rodent Resting-State fMRI. Front. Neurosci. 2018, 12, 788. [Google Scholar] [CrossRef]
  12. Birn, R.M.; Diamond, J.B.; Smith, M.A.; Bandettini, P.A. Separating respiratory-variation-related neuronal-activity-related fluctuations in fluctuations from fMRI. Neuroimage 2006, 31, 1536–1548. [Google Scholar] [CrossRef] [PubMed]
  13. Bianciardi, M.; Fukunaga, M.; van Gelderen, P.; Horovitz, S.G.; de Zwart, J.A.; Shmueli, K.; Duyn, J.H. Sources of functional magnetic resonance imaging signal fluctuations in the human brain at rest: A 7 T study. Magn. Reson. Imaging 2009, 27, 1019–1029. [Google Scholar] [CrossRef] [PubMed]
  14. Glover, G.H.; Li, T.Q.; Ress, D. Image-based method for retrospective correction of physiological motion effects in fMRI: RETROICOR. Magn. Reson. Med. 2000, 44, 162–167. [Google Scholar] [CrossRef] [PubMed]
  15. Barton, M.; Marecek, R.; Krajcovicová, L.; Slavícek, T.; Kaspárek, T.; Zemánková, P.; Ríha, P.; Mikl, M. Evaluation of different cerebrospinal fluid and white matter fMRI filtering strategies-Quantifying noise removal and neural signal preservation. Hum. Brain Mapp. 2019, 40, 1114–1138. [Google Scholar] [CrossRef] [PubMed]
  16. Lee, A.T.; Glover, G.H.; Meyer, C.H. Discrimination of Large Venous Vessels in Time-Course Spiral Blood-Oxygen-Level-Dependent Magnetic-Resonance Functional Neuroimaging. Magn. Reson. Med. 1995, 33, 745–754. [Google Scholar] [CrossRef] [PubMed]
  17. Kundu, P.; Inati, S.J.; Evans, J.W.; Luh, W.M.; Bandettini, P.A. Differentiating BOLD and non-BOLD signals in fMRI time series using multi-echo EPI. Neuroimage 2012, 60, 1759–1770. [Google Scholar] [CrossRef] [PubMed]
  18. Power, J.D.; Plitt, M.; Gotts, S.J.; Kundu, P.; Voon, V.; Bandettini, P.A.; Martin, A. Ridding fMRI data of motion-related influences: Removal of signals with distinct spatial and physical bases in multiecho data. Proc. Natl. Acad. Sci. USA 2018, 115, E2105–E2114. [Google Scholar] [CrossRef] [PubMed]
  19. Dipasquale, O.; Sethi, A.; Lagana, M.M.; Baglio, F.; Baselli, G.; Kundu, P.; Harrison, N.A.; Cercignani, M. Comparing resting state fMRI de-noising approaches using multi- and single-echo acquisitions. PLoS ONE 2017, 12, e0173289. [Google Scholar] [CrossRef]
  20. Kundu, P.; Santin, M.D.; Bandettini, P.A.; Bullmore, E.T.; Petiet, A. Differentiating BOLD and non-BOLD signals in fMRI time series from anesthetized rats using multi-echo EPI at 11.7 T. Neuroimage 2014, 102, 861–874. [Google Scholar] [CrossRef]
  21. Heunis, S.; Breeuwer, M.; Caballero-Gaudes, C.; Hellrung, L.; Huijbers, W.; Jansen, J.F.; Lamerichs, R.; Zinger, S.; Aldenkamp, A.P. The effects of multi-echo fMRI combination and rapid T2*-mapping on offline and real-time BOLD sensitivity. Neuroimage 2021, 238, 118244. [Google Scholar] [CrossRef]
  22. Kundu, P.; Voon, V.; Balchandani, P.; Lombardo, M.V.; Poser, B.A.; Bandettini, P.A. Multi-echo fMRI: A review of applications in fMRI denoising and analysis of BOLD signals. Neuroimage 2017, 154, 59–80. [Google Scholar] [CrossRef] [PubMed]
  23. Beissner, F.; Baudrexel, S.; Volz, S.; Deichmann, R. Dual-echo EPI for non-equilibrium fMRI-Implications of different echo combinations and masking procedures. Neuroimage 2010, 52, 524–531. [Google Scholar] [CrossRef] [PubMed]
  24. Gonzalez-Castillo, J.; Panwar, P.; Buchanan, L.C.; Caballero-Gaudes, C.; Handwerker, D.A.; Jangraw, D.C.; Zachariou, V.; Inati, S.; Roopchansingh, V.; Derbyshire, J.A.; et al. Evaluation of multi-echo ICA denoising for task based fMRI studies: Block designs, rapid event-related designs, and cardiac-gated fMRI. Neuroimage 2016, 141, 452–468. [Google Scholar] [CrossRef] [PubMed]
  25. Glover, G.H.; Lemieux, S.K.; Drangova, M.; Pauly, J.M. Decomposition of inflow and blood oxygen level-dependent (BOLD) effects with dual-echo spiral gradient-recalled echo (GRE) fMRI. Magn. Reson. Med. 1996, 35, 299–308. [Google Scholar] [CrossRef]
  26. Speck, O.; Ernst, T.; Clang, L. Biexponential modeling of multigradient-echo MRI data of the brain. Magn. Reson. Med. 2001, 45, 1116–1121. [Google Scholar] [CrossRef]
  27. Marciani, L.; Pfeiffer, J.C.; Hort, J.; Head, K.; Bush, D.; Taylor, A.J.; Spiller, R.C.; Francis, S.; Gowland, P.A. Improved methods for fMRI studies of combined taste and aroma stimuli. J. Neurosci. Methods 2006, 158, 186–194. [Google Scholar] [CrossRef]
  28. Saad, Z.S.; Glen, D.R.; Chen, G.; Beauchamp, M.S.; Desai, R.; Cox, R.W. A new method for improving functional-to-structural MRI alignment using local Pearson correlation. Neuroimage 2009, 44, 839–848. [Google Scholar] [CrossRef]
  29. Lombardo, M.V.; Auyeung, B.; Holt, R.J.; Waldman, J.; Ruigrok, A.N.V.; Mooney, N.; Bullmore, E.T.; Baron-Cohen, S.; Kundu, P. Improving effect size estimation and statistical power with multi-echo fMRI and its impact on understanding the neural systems supporting mentalizing. Neuroimage 2016, 142, 55–66. [Google Scholar] [CrossRef]
  30. Olafsson, V.; Kundu, P.; Wong, E.C.; Bandettini, P.A.; Liu, T.T. Enhanced identification of BOLD-like components with multi-echo simultaneous multi-slice (MESMS) fMRI and multi-echo ICA. Neuroimage 2015, 112, 43–51. [Google Scholar] [CrossRef]
  31. Dai, L.; Yang, Q.; Lin, J.; Zhou, Z.; Zhang, P.; Cai, S.; Chen, Z.; Wu, Z.; Kang, T.; Cai, C. Learning from synthetic data for reference-free Nyquist ghost correction and parallel imaging reconstruction of echo planar imaging. Med. Phys. 2023, 50, 2135–2147. [Google Scholar] [CrossRef] [PubMed]
  32. Yang, Q.Q.; Lin, Y.H.; Wang, J.C.; Bao, J.F.; Wang, X.Y.; Ma, L.C.; Zhou, Z.H.; Yang, Q.Z.; Cai, S.H.; He, H.J.; et al. Model-Based SyntheTic Data-Driven Learning (MOST-DL): Application in Single-Shot T-2 Mapping with Severe Head Motion Using Overlapping-Echo Acquisition. IEEE Trans. Med. Imaging 2022, 41, 3167–3181. [Google Scholar] [CrossRef]
  33. Zhang, C.; Yang, Q.; Fan, L.; Yu, S.; Sun, L.; Cai, C.; Ding, X. Towards Better Generalization Using Synthetic Data: A Domain Adaptation Framework for T2 Mapping via Multiple Overlapping-Echo Acquisition. IEEE Trans. Med. Imaging 2024, 1. [Google Scholar] [CrossRef]
  34. Ashburner, J. Statistical Parametric Mapping. Available online: https://github.com/spm/spm12 (accessed on 7 July 2024).
  35. Shorten, C.; Khoshgoftaar, T.M. A survey on Image Data Augmentation for Deep Learning. J. Big Data 2019, 6, 60. [Google Scholar] [CrossRef]
  36. Ahmed, Z.; Bandettini, P.A.; Bottenhorn, K.L.; Caballero-Gaudes, C.; Dowdle, L.T.; DuPre, E.; Gonzalez-Castillo, J.; Handwerker, D.; Heunis, S.; Kundu, P.; et al. ME-ICA/Tedana: 0.0.12. Available online: https://zenodo.org/records/6461353 (accessed on 14 May 2024).
  37. Kundu, P.; Brenowitz, N.D.; Voon, V.; Worbe, Y.; Vertes, P.E.; Inati, S.J.; Saad, Z.S.; Bandettini, P.A.; Bullmore, E.T. Integrated strategy for improving functional connectivity mapping using multiecho fMRI. Proc. Natl. Acad. Sci. USA 2013, 110, 16187–16192. [Google Scholar] [CrossRef] [PubMed]
  38. DuPre, E.; Salo, T.; Ahmed, Z.; Bandettini, P.A.; Bottenhorn, K.L.; Caballero-Gaudes, C.; Dowdle, L.T.; Gonzalez-Castillo, J.; Heunis, S.; Kundu, P. TE-dependent analysis of multi-echo fMRI with tedana. J. Open Source Softw. 2021, 6, 3669. [Google Scholar] [CrossRef]
  39. Heunis, S.; Breeuwer, M.; Caballero-Gaudes, C.; Hellrung, L.; Huijbers, W.; Jansen, J.F.A.; Lamerichs, R.; Zinger, S.; Aldenkamp, A.P. rt-me-fMRI: A task and resting state dataset for real-time, multi-echo fMRI methods development and validation. F1000Research 2021, 10, 70. [Google Scholar] [CrossRef]
  40. Kasper, L.; Bollmann, S.; Diaconescu, A.O.; Hutton, C.; Heinzle, J.; Iglesias, S.; Hauser, T.U.; Sebold, M.; Manjaly, Z.M.; Pruessmann, K.P.; et al. The PhysIO Toolbox for Modeling Physiological Noise in fMRI Data. J. Neurosci. Methods 2017, 276, 56–72. [Google Scholar] [CrossRef]
  41. Stephan Henuis, R.G. fMRwhy. Available online: https://github.com/jsheunis/fMRwhy (accessed on 7 July 2024).
  42. Pernet, C.R. Misconceptions in the use of the General Linear Model applied to functional MRI: A tutorial for junior neuro-imagers. Front. Neurosci. 2014, 8, 1. [Google Scholar] [CrossRef]
  43. Yang, Q.; Ma, L.; Zhou, Z.; Bao, J.; Yang, Q.; Huang, H.; Cai, S.; He, H.; Chen, Z.; Zhong, J.; et al. Rapid high-fidelity T2* mapping using single-shot overlapping-echo acquisition and deep learning reconstruction. Magn. Reson. Med. 2023, 89, 2157–2170. [Google Scholar] [CrossRef]
Figure 1. Overview of ME-fMRI processing pipeline. (a) Overview of ME-fMRI data preprocessing. (b) Preprocessing step. (c) Six time series calculated after preprocessing for comparison: SLLF indicates 4-dimensional T2* map estimated by log-linear fitting (LLF), SEC indicates echo combination (EC) time series, SMEICA indicates MEICA-denoised EC time series, SSD-DL indicates 4-dimensional T2* map estimated by SD-DL, SSD-DL-EC indicates SD-DL-EC time series, and SSD-DL-MEICA indicates MEICA-denoised SD-DL-EC time series. (d) Analysis pipeline.
Figure 1. Overview of ME-fMRI processing pipeline. (a) Overview of ME-fMRI data preprocessing. (b) Preprocessing step. (c) Six time series calculated after preprocessing for comparison: SLLF indicates 4-dimensional T2* map estimated by log-linear fitting (LLF), SEC indicates echo combination (EC) time series, SMEICA indicates MEICA-denoised EC time series, SSD-DL indicates 4-dimensional T2* map estimated by SD-DL, SSD-DL-EC indicates SD-DL-EC time series, and SSD-DL-MEICA indicates MEICA-denoised SD-DL-EC time series. (d) Analysis pipeline.
Brainsci 14 00828 g001
Figure 2. A schematic demonstration of the synthetic data-driven multi-echo deep learning (SD-DL) framework. Parametric templates were transformed from a public multi-contrast image database. Synthetic data were generated from parametric templates through using the fMRI signal relaxation model.
Figure 2. A schematic demonstration of the synthetic data-driven multi-echo deep learning (SD-DL) framework. Parametric templates were transformed from a public multi-contrast image database. Synthetic data were generated from parametric templates through using the fMRI signal relaxation model.
Brainsci 14 00828 g002
Figure 3. Comparison of mean square error (MSE) between LLF-fitted T2*, SD-DL-fitted T2*, and T2* labels across synthetic samples that had 5 different levels of Rician noise added to them. Rician noise level: (a) 1% (b) 2% (c) 3% (d) 4%, and (e) 5% standard deviation. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Figure 3. Comparison of mean square error (MSE) between LLF-fitted T2*, SD-DL-fitted T2*, and T2* labels across synthetic samples that had 5 different levels of Rician noise added to them. Rician noise level: (a) 1% (b) 2% (c) 3% (d) 4%, and (e) 5% standard deviation. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Brainsci 14 00828 g003
Figure 4. Comparison between LLF and SD-DL in T2* maps, tSNR maps, activation maps, and BOLD responses based on in-house fMRI data. (a) T2* maps estimated by LLF and SD-DL and the corresponding ground truth. (b) The tSNR maps of T2* time series estimated by LLF and SD-DL. (c) Activation maps (T-value maps) and BOLD responses. p < 0.0001. BOLD response curves were fitted with predicted responses from the maximum T-value voxel.
Figure 4. Comparison between LLF and SD-DL in T2* maps, tSNR maps, activation maps, and BOLD responses based on in-house fMRI data. (a) T2* maps estimated by LLF and SD-DL and the corresponding ground truth. (b) The tSNR maps of T2* time series estimated by LLF and SD-DL. (c) Activation maps (T-value maps) and BOLD responses. p < 0.0001. BOLD response curves were fitted with predicted responses from the maximum T-value voxel.
Brainsci 14 00828 g004
Figure 5. Comparison of whole-brain mean tSNR distribution of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA in rt-ME-fMRI dataset. **** indicates p < 0.0001. d indicates Cohen’s d. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Figure 5. Comparison of whole-brain mean tSNR distribution of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA in rt-ME-fMRI dataset. **** indicates p < 0.0001. d indicates Cohen’s d. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Brainsci 14 00828 g005
Figure 6. Comparison of percentage signal change within the UFWE of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. The plots (from top to bottom) correspond to the following: (a) fingerTapping, (b) fingerTappingImagined, (c) emotionProcessing, and (d) emotionProcessingImagined. n.s., *, **, ***, and **** indicate no significance, p < 0.05, 0.01, 0.001, and 0.0001, respectively. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Figure 6. Comparison of percentage signal change within the UFWE of SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. The plots (from top to bottom) correspond to the following: (a) fingerTapping, (b) fingerTappingImagined, (c) emotionProcessing, and (d) emotionProcessingImagined. n.s., *, **, ***, and **** indicate no significance, p < 0.05, 0.01, 0.001, and 0.0001, respectively. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Brainsci 14 00828 g006
Figure 7. Comparison of the T-values within the UFWE for SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. The plots (from top to bottom) correspond to the following: (a) fingerTapping, (b) fingerTappingImagined, (c) emotionProcessing, and (d) emotionProcessingImagined. n.s., *, ***, and **** indicate no significance, p < 0.05, 0.001, and 0.0001, respectively. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Figure 7. Comparison of the T-values within the UFWE for SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. The plots (from top to bottom) correspond to the following: (a) fingerTapping, (b) fingerTappingImagined, (c) emotionProcessing, and (d) emotionProcessingImagined. n.s., *, ***, and **** indicate no significance, p < 0.05, 0.001, and 0.0001, respectively. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Brainsci 14 00828 g007
Figure 8. Comparison of functional contrast distribution within the UFWE for SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. The plots (from top to bottom) correspond to the following: (a) fingerTapping, (b) fingerTappingImagined, (c) emotionProcessing, and (d) emotionProcessingImagined. n.s., *, **, and **** indicates no significance, p < 0.05, 0.01, and 0.0001, respectively. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Figure 8. Comparison of functional contrast distribution within the UFWE for SLLF, SSD-DL, SEC, SSD-DL-EC, SMEICA, and SSD-DL-MEICA. The plots (from top to bottom) correspond to the following: (a) fingerTapping, (b) fingerTappingImagined, (c) emotionProcessing, and (d) emotionProcessingImagined. n.s., *, **, and **** indicates no significance, p < 0.05, 0.01, and 0.0001, respectively. Purple color indicates results using the LLF and orange-red color indicates results using the SD-DL.
Brainsci 14 00828 g008
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Zhao, Y.; Yang, Q.; Qian, S.; Dong, J.; Cai, S.; Chen, Z.; Cai, C. Improving the Sensitivity of Task-Based Multi-Echo Functional Magnetic Resonance Imaging via T2* Mapping Using Synthetic Data-Driven Deep Learning. Brain Sci. 2024, 14, 828. https://doi.org/10.3390/brainsci14080828

AMA Style

Zhao Y, Yang Q, Qian S, Dong J, Cai S, Chen Z, Cai C. Improving the Sensitivity of Task-Based Multi-Echo Functional Magnetic Resonance Imaging via T2* Mapping Using Synthetic Data-Driven Deep Learning. Brain Sciences. 2024; 14(8):828. https://doi.org/10.3390/brainsci14080828

Chicago/Turabian Style

Zhao, Yinghe, Qinqin Yang, Shiting Qian, Jiyang Dong, Shuhui Cai, Zhong Chen, and Congbo Cai. 2024. "Improving the Sensitivity of Task-Based Multi-Echo Functional Magnetic Resonance Imaging via T2* Mapping Using Synthetic Data-Driven Deep Learning" Brain Sciences 14, no. 8: 828. https://doi.org/10.3390/brainsci14080828

APA Style

Zhao, Y., Yang, Q., Qian, S., Dong, J., Cai, S., Chen, Z., & Cai, C. (2024). Improving the Sensitivity of Task-Based Multi-Echo Functional Magnetic Resonance Imaging via T2* Mapping Using Synthetic Data-Driven Deep Learning. Brain Sciences, 14(8), 828. https://doi.org/10.3390/brainsci14080828

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop