Distortion-free diffusion-weighted imaging using echo planar time-resolved imaging with prospective motion correction: a feasibility study in motion-sensitive patients

Introduction

Echo planar imaging (EPI) has been widely used for diffusion-weighted imaging (DWI) in clinical neuroimaging to probe tissue microstructure by measuring water diffusion (1-4). However, conventional single-shot echo-planar imaging (ss-EPI) is highly susceptible to geometric distortion, as its long readout duration and low effective bandwidth in the phase-encoding direction make it particularly sensitive to B₀ inhomogeneity (5,6). Multi-shot EPI (ms-EPI) techniques, such as interleaved ms-EPI (7,8) and readout-segmented EPI (rs-EPI) (9,10), have been developed to reduce distortion and blurring by increasing the sampling bandwidth and reducing the readout window, enabling studies at higher field strengths as well as examining structures that are proximal to regions with inhomogeneous fields (11). However, because strong diffusion gradients are required for the diffusion-encoding process, even minor bulk motion or physiological fluctuations (e.g., involuntary motion, cardiac pulsation, respiration, and peristalsis) can induce background phase variations (12-15). The relative longer scan time of ms-EPI DWI and the mismatch of motion-induced phase between shots render ms-EPI acquisitions less robust than ss-EPI, leading to severe artifacts and inaccuracies in diffusion parameter estimation (16-19). The issue becomes particularly pronounced in motion-sensitive patients, such as children, elderly subjects, or patients with neurological disorders, who often struggle to remain still during magnetic resonance imaging (MRI) scans (20,21).

Navigator echoes have been employed in ms-EPI and rs-EPI to estimate shot-to-shot phase variations induced by physiological motion, such as cerebrospinal fluid (CSF) pulsation (8,10,22), but these approaches generally fail to handle through-plane motion caused by rigid-body macroscopic head motion. Retrospective image registration methods, such as the eddy tool in FMRIB Software Library (FSL) (23), can correct motions and distortions induced by eddy current in ss-EPI DWI, but they are insufficient for ms-EPI in the presence of substantial rigid-body motion and through-plane motion. Due to these limitations, robust prospective motion correction (PMC) methods have been developed (24-29), which aim to estimate rigid-body head motion in real time and update the imaging coordinates accordingly. Unlike retrospective approaches that can only compensate after acquisition, PMC dynamically adapts the imaging field of view (FOV) during scanning, thereby reducing spin-history artifacts and improving data consistency. Navigator-based approaches do not require additional hardware, but their temporal resolution is limited, typically providing only low-frequency motion estimates once per repetition time (TR), which cannot correct intra-TR or through-plane motion. Navigator-free methods (30,31) can retrospectively correct motion-induced phase variations across shots without additional navigators. Although some navigator-free methods (32,33) can address inter-shot rigid-body motion, correcting through-plane motion remains particularly challenging for navigator-free methods in diffusion imaging. The external tracking systems, including marker-based optical devices and more recent markerless PMC systems (25,26), enable high-frequency sampling of head position with sub-voxel accuracy, allowing real-time correction of diffusion gradient orientations and other sequence parameters. These advances have demonstrated improved robustness across structural (24), functional (34), and diffusion MRI (25,26), making PMC a promising strategy to overcome the fundamental motion sensitivity of multi-shot diffusion imaging.

Echo-planar time-resolved imaging (EPTI) (35) is a novel EPI-based MRI technique that employs optimized k-t space sampling to provide distortion-free, multi-echo, and multi-contrast images from both single-shot (36) and multi-shot acquisitions. Locally low rank (LLR) subspace reconstruction can be applied to further improve the reconstruction performance of EPTI (37). As a promising alternative to conventional EPI, EPTI has recently been extended to enable high-resolution, distortion-free and multi-contrast DWI (38,39). Moreover, the reconstructed multi-contrast images of EPTI enables T₂* mapping, which could provide information on local field inhomogeneity for applications such as susceptibility-weighted imaging (SWI) (40) and blood-oxygen-level-dependent (BOLD) functional MRI (fMRI) (41). However, high-resolution multi-shot EPTI implementations remain susceptible to motion. Existing implementations predominantly rely on retrospective correction strategies, in which trajectory design and reconstruction algorithms are optimized to suppress motion-induced artifacts. Representative strategies have included the use of four-dimensional (4D) navigators for phase estimation (42), propeller-style rotated sampling for improved robustness (43), and rotational view super-resolution with self-navigation (44). These retrospective or trajectory-based approaches can partially mitigate motion artifacts through sequence design and algorithmic optimization without requiring a PMC system or additional hardware. But PMC system based markerless optical motion tracking can more effectively correct intra-volume inter-shot motion in diffusion acquisition. Thus, integrating PMC with EPTI provides a unified framework that simultaneously resolves susceptibility-induced distortion and motion sensitivity.

In this work, we employed a 3-shot EPTI acquisition to provide distortion-free, multi-contrast diffusion imaging. To address the motion sensitivity inherent to multi-shot acquisitions, a markerless PMC system was integrated with EPTI to correct inter-shot motion artifacts in real time. Specifically, this study assessed the impact of PMC in a controlled volunteer experiment with controlled head motion, and evaluated reader-rated image quality and apparent diffusion coefficients (ADC) agreement between EPTI-DWI with PMC and conventional ss-EPI in motion-sensitive patient cohorts. These evaluations were designed to establish the clinical feasibility of EPTI-DWI with PMC as a robust alternative to conventional EPI-DWI, particularly for motion-sensitive patients.

Methods

Participants

The study included one healthy volunteer instructed to perform periodic head movements during EPTI-DWI to evaluate PMC, and a motion-sensitive cohort comprising 24 adolescent patients with anxiety disorder (mean age 15.5±2.0 years) and 14 patients with Alzheimer’s disease (mean age 77.1±7.3 years). All participants were recruited from Ningbo Kangning Hospital Affiliated to Ningbo University, Ningbo, China. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Ningbo Kangning Hospital (Nos. NBKNYY-2023-LC-50 and NBKNYY-2023-LC-46). Written informed consents were obtained from all participants or their legal guardians.

EPTI with PMC

Figure 1A shows the markerless PMC system. The markerless hardware-based motion tracking system (26,34) employs a three-dimensional (3D) camera to continuously track rigid-body head motion by performing real-time 3D point cloud reconstruction and registration, and then estimates six degrees of freedom (DOF) including three translations and three rotations. The estimated motion parameters are fed back to the MRI scanner for real-time sequence update, enabling accurate correction of head motion during acquisition. Prior to each excitation, the MRI system updated the imaging coordinates based on the tracked pose to maintain spatial alignment. These prospective updates were applied to the excitation pulse, refocusing pulse, and diffusion gradients through the updated coordinate frame, ensuring that the slice position and orientation, the targeted FOV, and the diffusion-encoding directions followed the subject’s motion in real time with minimal latency. The PMC system can track head motion at 30 fps with a latency of ~90 ms and a precision of 0.1 mm translation and 0.1° rotation. Despite prospective correction, shot-to-shot background phase variations caused by the strong diffusion gradients still required additional correction.

Figure 1 Overview of the PMC system and the EPTI-DWI reconstruction workflow. (A) The PMC system. (B) The pulse sequence diagram of EPTI. (C) The sampling trajectory of EPTI. (D) The LLR subspace reconstruction. 3D, three-dimensional; ADC, analog-to-digital converter; DWI, diffusion-weighted imaging; EPTI, echo planar time-resolved imaging; Gp, phase encoding gradient; Gr, readout gradient; Gs, slice selection gradient; LLR, locally low-rank; MRI, magnetic resonance imaging; PMC, prospective motion correction; RF, radio frequency pulse; SVD, singular value decomposition.

EPTI is a recently developed multi-contrast technique that eliminates geometric distortion by extending conventional EPI into the two-dimensional k_y–t space (35). Through LLR subspace reconstruction, EPTI enables the recovery of distortion-free multi-echo images. Figure 1B,1C demonstrate the sequence diagram and k_y–t sampling trajectory of 3-shot EPTI-DWI. For high resolution DWI, the 3-shot trajectory was adopted (38), which consisted of one central shot that repeatedly sampled the low frequency region around the k-space center, and two complementary shots that together provided full coverage of the remaining k-space to control aliasing. More intensive sampling at the center of k-space and at start time allow improving signal-to-noise ratio (SNR) and more accurate LLR constrained reconstruction. Each shot traverses k_y–t space, and passes through the center of k-space for self-navigated shot-to-shot background phase correction.

LLR Subspace reconstruction

The LLR subspace reconstruction (45) was performed with the under-sampling data from EPTI. The reconstruction problem now is to solve the following problem:

argminα12‖y−AΦKα‖22+∑rλ‖Rr(α)‖∗[1]

Here A is the overall sequence encoding operator, including sensitivity map, B₀ inhomogeneity induced phase and shot-to-shot background phase and the k-space under-sampling trajectory. Shot-wise LLR subspace reconstruction was performed for each shot using only the k-space center data to generate low-resolution, distortion-free and multi-echo navigator images. The shot-to-shot phase variations were then estimated from the background phase of the navigator images. Φ_K is the first K temporal subspace basis derived from Bloch simulation by singular value decomposition (SVD). R_r is the LLR regularization term. With the coefficients α, the distortion-free multi-echo images can be obtained reconstructed as shown in Figure 1D.

All the processing and reconstructions were performed offline in MATLAB (R2019a, MathWorks, Inc., Natick, MA, USA). Berkeley Advanced Reconstruction Toolbox (BART) (46) was used for the LLR subspace reconstruction. The first K=3 temporal subspace basis were used, which were sufficient to represent the simulated signal evolutions with over 99% accuracy. The reconstruction was solved using an alternating direction method of multipliers (ADMM) algorithm (47) with a maximum of 100 iterations and an LLR regularization weight of λ=0.01.

MRI acquisition

The MRI scans were performed at Ningbo Kangning Hospital Affiliated to Ningbo University, Ningbo, China. All scans were performed on a 3.0 T United Imaging uMR NX system (Shanghai, China) using a 64-channel head coil. The protocol comprised T₂-weighted fast spin-echo (T₂ FSE), single-shot EPI-DWI, and EPTI-DWI with PMC. EPTI-DWI was performed with the following parameters: FOV =220 mm × 220 mm; in-plane resolution =1 mm ×1 mm; slice thickness =5 mm; number of slices =19; TR =3,000 ms; range of echo time (TE) =38–97 ms; number of shots =3; averages =1; number of echoes =64; echo spacing (ESP) =0.95 ms, bandwidth =2,000 Hz/pixel; b values =0/1,000 s/mm²; diffusion directions =3; total acquisition time =60 s including dummy scan and low-resolution calibration scan for reconstruction; the low-resolution calibration scan with a matrix size =224×32×6 (k_x×k_y×N_echo) was used to compute coil sensitivity maps, and estimate B₀ inhomogeneity.

Single-shot EPI-DWI was performed with the following parameters: FOV =220 mm × 220 mm; the in-plane resolution =1.25 mm × 1.25 mm; slice thickness =5 mm; number of slices =19; TR =2,839 ms; TE =80 ms; averages =2; ESP =0.89 ms, bandwidth =1,800 Hz/pixel; b values =0/1,000 s/mm²; diffusion directions =3; parallel imaging acceleration factor =2; total acquisition time =45 s.

T₂ FSE was acquired to provide high-resolution anatomical reference for structural comparison with the following parameters: FOV =200 mm × 230 mm; in-plane resolution =0.51 mm × 0.51 mm; slice thickness =5 mm; number of slices =19; TR =5,515 ms; TE =135 ms; bandwidth =210 Hz/pixel; parallel imaging acceleration factor =2; partial Fourier in-phase encoding =7/8, echo train length =27; total acquisition time =28 s.

Qualitative scoring

Two radiologists (2 and 6 years of experience in MRI, respectively) independently reviewed paired EPI-DWI and EPTI-DWI images presented in randomized order. Absolute image quality was rated on a 5-point Likert scale (1= nondiagnostic, 2= limited, 3= fair, 4= good, 5= excellent) for 4 metrics including anatomical clarity, geometric distortion, diagnostic confidence, and overall image quality. Relative preference (EPTI-DWI vs. EPI-DWI) was evaluated on a separate 5-point Likert scale (1= worse, 2= slightly worse, 3= same, 4= slightly better, 5= better). Image quality scoring was performed on DWI images (b=0 and b=1,000 s/mm²) acquired with both EPI-DWI and EPTI-DWI at two representative TEs (40 and 80 ms). Two scores of the two readers were averaged to obtain the final score.

Statistical analysis

Qualitative scoring results from two readers were compared using the Wilcoxon signed-rank test. Interreader agreement for absolute scores was quantified using an intraclass correlation coefficient (ICC) analysis with 95% confidence intervals (CIs). ICC values were interpreted as 0.00–0.20 (poor), 0.21–0.40 (fair), 0.41–0.60 (moderate), 0.61–0.80 (good), and 0.81–1.00 (excellent).

For EPI-DWI, ADC maps were derived directly from the scanner’s standard reconstruction pipeline. For EPTI-DWI, ADC maps were computed offline from b=0 and 1,000 s/mm² images at TE =40 ms using a mono-exponential model:

S(b)=S0e−b⋅ADC[2]

Region of interest (ROI) placement was performed by two radiologists in consensus. Four ROIs were placed on the ADC maps derived from both EPI-DWI and EPTI-DWI: bilateral hippocampi (two symmetric ROIs), the genu of the corpus callosum, and frontal white matter. These regions were chosen because of their established relevance to Alzheimer’s disease and mood disorders (48-51). ROI areas were constrained to 30–50 mm² and were drawn with side-by-side reference to the distortion-free T₂ FSE to ensure anatomical consistency. To ensure data integrity and mitigate partial-volume effects, ROI placement strictly excluded areas with severe geometric distortion, as well as pixels containing CSF, large vessels, or tissue-interface boundaries.

For cross-sequence comparison, the mean ADC values within the ROIs from EPTI-DWI and EPI-DWI were analyzed across all patients. Linear association was evaluated using Pearson’s correlation coefficient (r). Agreement was evaluated by Bland-Altman analysis with the difference defined as ADC_EPTI − ADC_EPI, reporting the mean bias and 95% limits of agreement [mean ± 1.96 standard deviation (SD)]. A two-sided P<0.05 was considered statistically significant.

Results

Although subjects were stabilized with foam padding and instructed to remain still, some residual motion inevitably occurred. Table 1 summarizes the motion metrics (mean, SD, maximum, and 95th percentile), with mean values reflecting the Euclidean distance across three directions measured per TR for each group. Average displacements of 1.08 mm (14 patients with Alzheimer’s disease) and 1.65 mm (24 adolescent patients with anxiety disorder) with larger excursions in some cases were observed, confirming that in-scan motion was non-negligible in these motion-sensitive patients. To investigate the impact of such motion, a healthy volunteer was instructed to perform controlled periodic head movements during EPTI-DWI acquisition. Figure 2A,2B show the motion patterns captured by the PMC system during the ETPI-DWI acquisitions and Figure 2C shows the reconstruction results of EPTI-DWI with/without PMC. In the controlled volunteer experiment, the induced head motion was 1.46±0.28 mm in translation and 0.04±0.02 rad in rotation (mean ± SD), which closely matched the motion range observed in the patient cohorts in Table 1, thereby reproducing clinically relevant motion conditions. The employed 3-shot EPTI sampling trajectory repeatedly traverses the k-space center, which enables the reconstruction of low-resolution subspace images from each shot. This design allows direct visualization of shot-to-shot background phase variations. Without PMC, these low-resolution images exhibited evident misalignment caused by head motion, leading to severe motion artifacts in the final high-resolution reconstructions. In contrast, enabling PMC effectively mitigated motion-induced artifacts from head movements, leading to artifact-free EPTI-DWI reconstructions.

Figure 2 EPTI-DWI with PMC for a healthy volunteer performing controlled head movements. (A) Motion patterns during the EPTI acquisitions with PMC on. (B) Motion patterns during the EPTI acquisitions with PMC off. (C) EPTI-DWI reconstruction results without PMC (upper panel) and with PMC (lower panel). Low-resolution subspace reconstruction was first applied to each shot of EPTI-DWI to assess shot-to-shot background phase variations. Given the high temporal resolution of EPTI, inter-shot misalignment was observed in the low-resolution images with PMC off, leading to severe motion artifacts in the final reconstructed images (red arrows). In contrast, with PMC on, motion artifacts were effectively corrected resulting in blurring-free images. DWI, diffusion-weighted imaging; EPTI, echo planar time-resolved imaging; PMC, prospective motion correction.

Figure 3A shows the motion patterns recorded by the PMC system in a 16-year-old adolescent with anxiety disorder. The head motion was characterized by maximum translations of approximately 3 mm. Figure 3B compares T₂ FSE, single-shot EPI-DWI, and EPTI-DWI with PMC. With T₂ FSE serving as the distortion-free reference, single-shot EPI-DWI (1.25 mm × 1.25 mm) demonstrated noticeable geometric distortions, especially in the frontal lobe, as indicated by the red arrow. In contrast, EPTI-DWI with PMC achieved higher resolution (1 mm ×1 mm) reconstructions without distortions, preserving anatomical fidelity in the frontal lobe.

Figure 3 EPTI-DWI with PMC in a 16-year-old patient with anxiety disorder. (A) Motion patterns recorded by the PMC system during the scan, showing substantial head movements. (B) Comparison among T₂ FSE, single-shot EPI-DWI, and EPTI-DWI with PMC at TE =40, 60, and 80 ms. For both b=0 and 1,000 s/mm² images, EPI-DWI exhibits geometric distortions in the frontal lobe (red arrow) compared to the T₂ FSE reference, while EPTI-DWI with PMC achieves distortion-free reconstruction results. DWI, diffusion-weighted imaging; EPI, echo planar imaging; EPTI, echo planar time-resolved imaging; PMC, prospective motion correction; T₂ FSE, T2-weighted fast spin echo; TE, echo time.

Figure 4A demonstrates the motion patterns captured by the PMC system in a 70-year-old patient with Alzheimer’s disease. The head motion was characterized by translations of approximately 1 mm. Figure 4B compares T₂ FSE, single-shot EPI-DWI, and EPTI-DWI with PMC at TE =40, 60, and 80 ms. T₂ FSE provided the distortion-free anatomical reference. Single-shot EPI-DWI (1.25 mm × 1.25 mm) exhibited pronounced susceptibility-induced distortions in the hippocampal region, as indicated by the red arrow, along with signal displacement and blurring that obscured fine structures. By comparison, EPTI-DWI with PMC yielded higher in-plane resolution (1 mm ×1 mm) and distortion-free images, with improved delineation of hippocampal boundaries, reduced displacement, and sharper depiction of deep structures, closely matching the T₂ FSE reference despite the presence of patient motion.

Figure 4 EPTI-DWI with PMC in a 70-year-old patient with Alzheimer’s disease. (A) Motion patterns recorded by the PMC system during the scan. (B) Comparison among T₂ FSE, single-shot EPI-DWI, and EPTI-DWI with PMC at TE =40, 60, and 80 ms. Single-shot EPI-DWI showed pronounced geometric distortions in the hippocampal region, highlighted by the red arrow in the zoomed view. In contrast, EPTI-DWI with PMC achieved distortion-free reconstructions with preserved hippocampal anatomy. DWI, diffusion-weighted imaging; EPI, echo planar imaging; EPTI, echo planar time-resolved imaging; PMC, prospective motion correction; T₂ FSE, T2-weighted fast spin echo; TE, echo time.

Figure 5 illustrates multi-contrast EPTI-DWI with PMC in an 81-year-old patient with Alzheimer’s disease. Two representative slices at TE =40, 60, and 80 ms demonstrate that high-resolution diffusion-weighted images (b=0 and b=1,000 s/mm²) were consistently reconstructed without distortion. ADC maps at TE =40 ms showed stable and reliable values, whereas ADC maps at long TE (e.g., 80 ms) exhibited localized overestimation and signal dropout due to SNR loss from T₂* decay near air-tissue interfaces. EPTI-DWI provided multi-contrast images at different TE. The cystic lesion at the inferior region became more conspicuous in images at longer TE (80 ms). In addition, T₂* maps were fitted separately from both b=0 and b=1,000 s/mm² datasets. T₂* map from b=1,000 s/mm² datasets showed lower and more homogeneous values, which may be explained by the marked attenuation of extracellular water signal after diffusion weighting, given that extracellular water typically has a longer T₂* than axonal and myelin water.

Figure 5 Multi-contrast images from EPTI-DWI with PMC in an 81-year-old patient with Alzheimer’s disease. Results from two representative slices at TE =40, 60, and 80 ms are shown, including diffusion-weighted images at b=0 and b=1,000 s/mm², ADC maps calculated at each TE. Additionally, T₂* maps are derived separately from the multi-echo b=0 and b=1,000 s/mm² data. Red arrow highlights additional tissue contrast provided by the multi-echo acquisitions, while yellow arrows indicate ADC overestimation and signal dropout near air-tissue interfaces at longer TE (80 ms). ADC, apparent diffusion coefficients; DWI, diffusion-weighted imaging; EPTI, echo planar time-resolved imaging; PMC, prospective motion correction; TE, echo time.

Figure 6 shows the distribution of qualitative reader scores for EPI-DWI and EPTI-DWI, as well as the relative preference ratings. The detailed results across all enrolled patients are summarized in Table 2. Compared with EPI-DWI, EPTI-DWI significantly reduced geometric distortion (4.74±0.43 vs. 3.46±0.54; P<0.001), improved diagnostic confidence (4.66±0.56 vs. 4.26±0.38; P<0.001), and enhanced overall image quality (4.76±0.41 vs. 4.39±0.44; P<0.01). There was no significant difference between the two methods in anatomical clarity (4.66±0.57 vs. 4.74±0.45; P=0.49). Relative preference ratings further demonstrated a consistent preference for EPTI-DWI over EPI-DWI across all metrics. The ICCs of the two readers for anatomical clarity, geometric distortion, diagnostic confidence, and overall image quality were 0.88 (95% CI: 0.66–0.97), 0.79 (95% CI: 0.50–0.93), 0.83 (95% CI: 0.63–0.93), and 0.77 (95% CI: 0.49–0.92), respectively, indicating moderate to excellent interreader agreement.

Figure 6 Distribution of qualitative reader scores comparing EPI-DWI and EPTI-DWI. (A) Distribution of reader scores for EPI-DWI. (B) Distribution of reader scores for EPTI-DWI. (C) Distribution of reader scores comparing EPTI-DWI with EPI-DWI. Absolute scores in (A) and (B) were rated independently by two readers on a 5-point Likert scale (1= nondiagnostic, 2= limited, 3= fair, 4= good, 5= excellent) for each metric: anatomical clarity, geometric distortion, diagnostic confidence, and overall image quality. Relative preference scores in (C) were rated on a separate 5-point Likert scale (1= worse, 2= slightly worse, 3= same, 4= slightly better, 5= better). DWI, diffusion-weighted imaging; EPI, echo planar imaging; EPTI, echo planar time-resolved imaging.

Figure 7 shows the correlation and Bland-Altman analyses of ADC derived from EPTI-DWI and EPI-DWI across all the patients. A strong correlation was observed between ADC values derived from EPTI-DWI and those obtained with conventional EPI-DWI (r=0.89; P<0.001). Bland-Altman analysis further revealed that EPTI-DWI yielded systematically higher ADC values than EPI-DWI, with a mean bias of 63.72×10⁻⁶ mm²/s and 95% limits of agreement ranging from 0.20×10⁻⁶ to 127.25×10⁻⁶ mm²/s. Most data points fell within the ±1.96 SD interval, suggesting good agreement between the two approaches despite a small systematic offset.

Figure 7 Quantitative comparison of ADC values between EPI-DWI and EPTI-DWI across all patients. (A) Scatter plot demonstrating a significant linear correlation between ADC values obtained from EPTI-DWI and EPI-DWI (r=0.89, P<0.001). The solid line indicates the linear regression fit; the dashed line represents the line of identity. (B) Bland-Altman plot illustrating the agreement between the two methods. The mean bias was 63.72×10⁻⁶ mm²/s, with 95% limits of agreement of 0.20×10⁻⁶ and 127.25×10⁻⁶ mm²/s. ADC, apparent diffusion coefficients; DWI, diffusion-weighted imaging; EPI, echo planar imaging; EPTI, echo planar time-resolved imaging; SD, standard deviation.

Discussion

In this work, we implemented a 3-shot EPTI acquisition to achieve distortion-free, higher in-plane resolution (1 mm × 1 mm) DWI, compared with EPI-DWI (1.25 mm × 1.25 mm). While higher resolution can also be achieved with single-shot EPI, the longer TE and extended echo train inevitably exacerbate geometric distortion. Ms-EPI, such as 4-shot acquisition (26), can reduce distortion and enable higher resolution, but remains highly sensitive to inter-shot motion with only single contrast. By contrast, EPTI not only eliminates distortions but also provides multi-echo images with a temporal spacing of 0.95 ms, enabling characterization of signal evolution across TEs. Nonetheless, the multi-shot acquisition of EPTI still makes it inherently sensitive to subject motion, which may limit robustness in clinical applications.

Several motion-robust extensions of EPTI have recently been proposed, such as PEPTIDE (43), and Romer-EPTI (44). Although these strategies enhance robustness, they are essentially retrospective or trajectory-based corrections performed during reconstruction, and remain limited in handling severe or through-plane motion. In contrast, our study integrates markerless PMC into EPTI, enabling real-time head motion correction at the acquisition level and providing a more direct and clinically feasible solution. The impact of PMC was first demonstrated in a controlled healthy volunteer experiment, where periodic head movements of approximately 1 mm in translation and 0.1 rad in rotation were intentionally introduced. Without PMC, patient motion introduced substantial inter-shot misalignment, leading to pronounced blurring and motion artifacts in the final reconstruction. By contrast, PMC stabilized the acquisition through real-time pose updates, thereby confirming its effectiveness in mitigating motion-induced inconsistencies.

In the patient cohort, motion metrics recorded during scanning confirmed the presence of involuntary head motion despite foam padding and instructions to remain still. Average displacements of 1–2 mm with larger excursions in some cases were observed (Table 1), magnitudes that would typically compromise multi-shot diffusion imaging. Under these conditions, EPTI-DWI with PMC consistently produced multi-contrast, distortion-free images without motion artifacts. Reader-based assessments demonstrated significant improvements in geometric distortion, diagnostic confidence, and overall image quality with EPTI-DWI, while anatomical clarity remained comparable between EPTI-DWI and EPI-DWI. These benefits were particularly evident in clinically relevant regions such as the frontal lobes and hippocampi, as illustrated in Figures 3,4, which are critical targets in the evaluation of psychiatric and neurodegenerative disorders including Alzheimer’s disease and anxiety disorder. Importantly, the quantitative reliability of EPTI-DWI was preserved. The ADC values exhibited strong correlation with those from ss-EPI, thereby confirming that enhanced image quality did not come at the expense of quantitative accuracy. EPTI-DWI may show systematically higher ADC values than EPI-DWI because ADC fitting was performed using nominal b-values. In EPTI, the effective b-value is echo-dependent, and using nominal b-value can bias the ADC estimates. More accurate ADC fitting may require calculating specific effective b-values for each echo. By improving both qualitative and quantitative aspects of diffusion imaging, EPTI-DWI with PMC has the potential to increase diagnostic reliability in motion-sensitive patients, reduce the need for repeat scans, and support longitudinal studies and treatment planning. These advantages highlight its clinical utility and underscore its potential as a robust alternative to conventional EPI-DWI in routine practice, especially for the motion-sensitive patients.

Several limitations should be acknowledged. First, the study was conducted in a single center with a relatively small sample size. Although feasibility was demonstrated in adolescents with anxiety disorder and elderly patients with Alzheimer’s disease, the statistical power was limited, and the findings may not be generalizable to broader clinical populations. Studies with larger sample sizes and more heterogeneous cohorts are needed to confirm reproducibility. Second, direct patient-level comparisons of EPTI-DWI with and without PMC were not performed. Although motion trajectories were recorded by the PMC system, acquiring paired PMC-on and PMC-off datasets in patients was not feasible due to practical and ethical constraints. Thus, the observed benefits of PMC in patients were inferred from comparisons with ss-EPI and supported by the volunteer experiment. Besides, the LLR subspace reconstruction framework requires an additional calibration scan for B₀ inhomogeneity estimation, which prolongs scan time compared with the parallel imaging calibration of conventional EPI. Moreover, EPTI can generate dozens of echo images, but many of these are not used for clinical interpretation and impose additional computational demands, and the later echoes with long TE inherently suffer from reduced SNR, which diminishes the conspicuity of fine structures and may limit the diagnostic value.

Conclusions

This study demonstrates that EPTI with markerless PMC provides distortion-free, multi-echo, high-resolution diffusion imaging with consistent ADC quantification compared with ss-EPI. Integrating with PMC, the 3-shot EPTI-DWI can effectively correct motion-induced artifacts in motion-sensitive patients. These results highlight the clinical feasibility of EPTI-DWI with PMC as a robust alternative to EPI-based diffusion imaging, with particular relevance for pediatric and elderly patients who are prone to motion during scanning.

Acknowledgments

None.

Footnote

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-1979/dss

Funding: This work was supported by the National Natural Science Foundation of China (Nos. 62301309 and 62471295), and Medical-Engineering Interdisciplinary Research Fund of Shanghai Jiao Tong University (No. YG2025LC01), Municipal Key R&D Program of Ningbo (Nos. 2023Z170 and 2023Z175).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-1979/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Ningbo Kangning Hospital (Nos. NBKNYY-2023-LC-50 and NBKNYY-2023-LC-46). Written informed consents were obtained from all participants or their legal guardians.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E, Laval-Jeantet M. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 1986;161:401-7. [Crossref] [PubMed]
2. Kim YJ, Chang KH, Song IC, Kim HD, Seong SO, Kim YH, Han MH. Brain abscess and necrotic or cystic brain tumor: discrimination with signal intensity on diffusion-weighted MR imaging. AJR Am J Roentgenol 1998;171:1487-90. [Crossref] [PubMed]
3. Jones DK, Knösche TR, Turner R. White matter integrity, fiber count, and other fallacies: the do's and don'ts of diffusion MRI. Neuroimage 2013;73:239-54. [Crossref] [PubMed]
4. Wu W, Miller KL. Image formation in diffusion MRI: A review of recent technical developments. J Magn Reson Imaging 2017;46:646-62. [Crossref] [PubMed]
5. Jezzard P, Barnett AS, Pierpaoli C. Characterization of and correction for eddy current artifacts in echo planar diffusion imaging. Magn Reson Med 1998;39:801-12. [Crossref] [PubMed]
6. Andersson JL, Skare S. A model-based method for retrospective correction of geometric distortions in diffusion-weighted EPI. Neuroimage 2002;16:177-99. [Crossref] [PubMed]
7. Chen NK, Guidon A, Chang HC, Song AW. A robust multi-shot scan strategy for high-resolution diffusion weighted MRI enabled by multiplexed sensitivity-encoding (MUSE). Neuroimage 2013;72:41-7. [Crossref] [PubMed]
8. Butts K, de Crespigny A, Pauly JM, Moseley M. Diffusion-weighted interleaved echo-planar imaging with a pair of orthogonal navigator echoes. Magn Reson Med 1996;35:763-70. [Crossref] [PubMed]
9. Holdsworth SJ, Skare S, Newbould RD, Guzmann R, Blevins NH, Bammer R. Readout-segmented EPI for rapid high resolution diffusion imaging at 3 T. Eur J Radiol 2008;65:36-46. [Crossref] [PubMed]
10. Porter DA, Heidemann RM. High resolution diffusion-weighted imaging using readout-segmented echo-planar imaging, parallel imaging and a two-dimensional navigator-based reacquisition. Magn Reson Med 2009;62:468-75. [Crossref] [PubMed]
11. Holdsworth SJ, O'Halloran R, Setsompop K. The quest for high spatial resolution diffusion-weighted imaging of the human brain in vivo. NMR Biomed 2019;32:e4056. [Crossref] [PubMed]
12. Ordidge RJ, Helpern JA, Qing ZX, Knight RA, Nagesh V. Correction of motional artifacts in diffusion-weighted MR images using navigator echoes. Magn Reson Imaging 1994;12:455-60. [Crossref] [PubMed]
13. Anderson AW, Gore JC. Analysis and correction of motion artifacts in diffusion weighted imaging. Magn Reson Med 1994;32:379-87. [Crossref] [PubMed]
14. Eichner C, Cauley SF, Cohen-Adad J, Möller HE, Turner R, Setsompop K, Wald LL. Real diffusion-weighted MRI enabling true signal averaging and increased diffusion contrast. Neuroimage 2015;122:373-84. [Crossref] [PubMed]
15. Miller KL, Pauly JM. Nonlinear phase correction for navigated diffusion imaging. Magn Reson Med 2003;50:343-53. [Crossref] [PubMed]
16. Robson MD, Anderson AW, Gore JC. Diffusion-weighted multiple shot echo planar imaging of humans without navigation. Magn Reson Med 1997;38:82-8. [Crossref] [PubMed]
17. Atkinson D, Porter DA, Hill DL, Calamante F, Connelly A. Sampling and reconstruction effects due to motion in diffusion-weighted interleaved echo planar imaging. Magn Reson Med 2000;44:101-9. [Crossref] [PubMed]
18. Pipe JG, Farthing VG, Forbes KP. Multishot diffusion-weighted FSE using PROPELLER MRI. Magn Reson Med 2002;47:42-52. Erratum in: Magn Reson Med 2002;47:621. [Crossref] [PubMed]
19. Liu C, Bammer R, Kim DH, Moseley ME. Self-navigated interleaved spiral (SNAILS): application to high-resolution diffusion tensor imaging. Magn Reson Med 2004;52:1388-96. [Crossref] [PubMed]
20. Alhamud A, Taylor PA, Laughton B, van der Kouwe AJ, Meintjes EM. Motion artifact reduction in pediatric diffusion tensor imaging using fast prospective correction. J Magn Reson Imaging 2015;41:1353-64. [Crossref] [PubMed]
21. Yendiki A, Koldewyn K, Kakunoori S, Kanwisher N, Fischl B. Spurious group differences due to head motion in a diffusion MRI study. Neuroimage 2014;88:79-90. [Crossref] [PubMed]
22. Jeong HK, Gore JC, Anderson AW. High-resolution human diffusion tensor imaging using 2-D navigated multishot SENSE EPI at 7 T. Magn Reson Med 2013;69:793-802. [Crossref] [PubMed]
23. Andersson JLR, Sotiropoulos SN. An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage 2016;125:1063-78. [Crossref] [PubMed]
24. Frost R, Wighton P, Karahanoğlu FI, Robertson RL, Grant PE, Fischl B, Tisdall MD, van der Kouwe A. Markerless high-frequency prospective motion correction for neuroanatomical MRI. Magn Reson Med 2019;82:126-44. [Crossref] [PubMed]
25. Berglund J, van Niekerk A, Rydén H, Sprenger T, Avventi E, Norbeck O, Glimberg SL, Olesen OV, Skare S. Prospective motion correction for diffusion weighted EPI of the brain using an optical markerless tracker. Magn Reson Med 2021;85:1427-40. [Crossref] [PubMed]
26. Chen H, Dai K, Zhong S, Zheng J, Zhang X, Yang S, Cao T, Wang C, Karasan E, Frydman L, Zhang Z. High-resolution multi-shot diffusion-weighted MRI combining markerless prospective motion correction and locally low-rank constrained reconstruction. Magn Reson Med 2023;89:605-19. [Crossref] [PubMed]
27. Herbst M, Zahneisen B, Knowles B, Zaitsev M, Ernst T. Prospective motion correction of segmented diffusion weighted EPI. Magn Reson Med 2015;74:1675-81. [Crossref] [PubMed]
28. Maclaren J, Herbst M, Speck O, Zaitsev M. Prospective motion correction in brain imaging: a review. Magn Reson Med 2013;69:621-36. [Crossref] [PubMed]
29. Slipsager JM, Ellegaard AH, Glimberg SL, Paulsen RR, Tisdall MD, Wighton P, van der Kouwe A, Marner L, Henriksen OM, Law I, Olesen OV. Markerless motion tracking and correction for PET, MRI, and simultaneous PET/MRI. PLoS One 2019;14:e0215524. [Crossref] [PubMed]
30. Guo H, Ma X, Zhang Z, Zhang B, Yuan C, Huang F. POCS-enhanced inherent correction of motion-induced phase errors (POCS-ICE) for high-resolution multishot diffusion MRI. Magn Reson Med 2016;75:169-80. [Crossref] [PubMed]
31. Hu Z, Ma X, Truong TK, Song AW, Guo H. Phase-updated regularized SENSE for navigator-free multishot diffusion imaging. Magn Reson Med 2017;78:172-81. [Crossref] [PubMed]
32. Guhaniyogi S, Chu ML, Chang HC, Song AW, Chen NK. Motion immune diffusion imaging using augmented MUSE for high-resolution multi-shot EPI. Magn Reson Med 2016;75:639-52. [Crossref] [PubMed]
33. Dong Z, Wang F, Ma X, Dai E, Zhang Z, Guo H. Motion-corrected k-space reconstruction for interleaved EPI diffusion imaging. Magn Reson Med 2018;79:1992-2002. [Crossref] [PubMed]
34. Liu Y, Wei Y, Yang Y, Zhang X, Zhao J, Lee PK, Tal A, Chen H, Zhang Z. Combining frequency navigator and optical prospective motion correction for functional MRS during motor activation at 3 T: A feasibility study. Med Phys 2025;52:e17861. [Crossref] [PubMed]
35. Wang F, Dong Z, Reese TG, Bilgic B, Katherine Manhard M, Chen J, Polimeni JR, Wald LL, Setsompop K. Echo planar time-resolved imaging (EPTI). Magn Reson Med 2019;81:3599-615. [Crossref] [PubMed]
36. Dong Z, Wald LL, Polimeni JR, Wang F. Single-shot echo planar time-resolved imaging for multi-echo functional MRI and distortion-free diffusion imaging. Magn Reson Med 2025;93:993-1013. [Crossref] [PubMed]
37. Dong Z, Wang F, Reese TG, Bilgic B, Setsompop K. Echo planar time-resolved imaging with subspace reconstruction and optimized spatiotemporal encoding. Magn Reson Med 2020;84:2442-55. [Crossref] [PubMed]
38. Dong Z, Wang F, Wald L, Setsompop K. SNR-efficient distortion-free diffusion relaxometry imaging using accelerated echo-train shifted echo-planar time-resolving imaging (ACE-EPTI). Magn Reson Med 2022;88:164-79. [Crossref] [PubMed]
39. Dai E, Lee PK, Dong Z, Fu F, Setsompop K, McNab JA. Distortion-Free Diffusion Imaging Using Self-Navigated Cartesian Echo-Planar Time Resolved Acquisition and Joint Magnitude and Phase Constrained Reconstruction. IEEE Trans Med Imaging 2022;41:63-74. [Crossref] [PubMed]
40. Rossi M, Ruottinen H, Elovaara I, Ryymin P, Soimakallio S, Eskola H, Dastidar P. Brain iron deposition and sequence characteristics in Parkinsonism: comparison of SWI, T₂* maps, T₂-weighted-, and FLAIR-SPACE. Invest Radiol 2010;45:795-802. [Crossref] [PubMed]
41. Heunis S, Breeuwer M, Caballero-Gaudes C, Hellrung L, Huijbers W, Jansen JF, Lamerichs R, Zinger S, Aldenkamp AP. The effects of multi-echo fMRI combination and rapid T(2)*-mapping on offline and real-time BOLD sensitivity. Neuroimage 2021;238:118244. [Crossref] [PubMed]
42. Dong Z, Wang F, Setsompop K. Motion-corrected 3D-EPTI with efficient 4D navigator acquisition for fast and robust whole-brain quantitative imaging. Magn Reson Med 2022;88:1112-5. [Crossref] [PubMed]
43. Fair MJ, Wang F, Dong Z, Reese TG, Setsompop K. Propeller echo-planar time-resolved imaging with dynamic encoding (PEPTIDE). Magn Reson Med 2020;83:2124-37. [Crossref] [PubMed]
44. Dong Z, Reese TG, Lee HH, Huang SY, Polimeni JR, Wald LL, Wang F. Romer-EPTI: Rotating-view motion-robust super-resolution EPTI for SNR-efficient distortion-free in-vivo mesoscale diffusion MRI and microstructure imaging. Magn Reson Med 2025;93:1535-55. [Crossref] [PubMed]
45. Trzasko J, Manduca A, Borisch E, editors. Local versus global low-rank promotion in dynamic MRI series reconstruction. The International Society for Magnetic Resonance in Medicine; 2011; Montreal, Quebec.
46. Uecker M, Virtue P, Ong F, Murphy MJ, Alley MT, Vasanawala SS, et al., editors. Software toolbox and programming library for compressed sensing and parallel imaging. ISMRM Workshop on Data Sampling and Image Reconstruction; 2013; Sedona, Arizona, USA.
47. Boyd S, Parikh N, Chu E, Peleato B, Eckstein J. Distributed optimization and statistical learning via the alternating direction method of multipliers. Found Trends Mach Learn 2011;3:1-122.
48. Jack CR Jr, Bennett DA, Blennow K, Carrillo MC, Dunn B, Haeberlein SB, et al. NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimers Dement 2018;14:535-62. [Crossref] [PubMed]
49. Dickerson BC, Stoub TR, Shah RC, Sperling RA, Killiany RJ, Albert MS, Hyman BT, Blacker D, Detoledo-Morrell L. Alzheimer-signature MRI biomarker predicts AD dementia in cognitively normal adults. Neurology 2011;76:1395-402. [Crossref] [PubMed]
50. Dupont RM, Jernigan TL, Heindel W, Butters N, Shafer K, Wilson T, Hesselink J, Gillin JC. Magnetic resonance imaging and mood disorders. Localization of white matter and other subcortical abnormalities. Arch Gen Psychiatry 1995;52:747-55. [Crossref] [PubMed]
51. Sacher J, Neumann J, Fünfstück T, Soliman A, Villringer A, Schroeter ML. Mapping the depressed brain: a meta-analysis of structural and functional alterations in major depressive disorder. J Affect Disord 2012;140:142-8. [Crossref] [PubMed]