Evaluating the Utility of EPIK in a Finger Tapping fMRI Experiment using BOLD Detection and Effective Connectivity

Seong Dae Yun, Ralph Weidner, Peter H. Weiss, N. Jon ShahScientific 2019 · DOI

TL;DR

This paper investigates the advantages of EPIK, a novel fMRI imaging technique, over the standard EPI method for detecting brain activity during a finger-tapping task. EPIK demonstrates superior spatial resolution and reduced image distortions, leading to stronger BOLD signal detection in key motor and visual areas, and more robust estimation of functional connectivity parameters.

What Problem Does This Paper Address?

Functional Magnetic Resonance Imaging (fMRI) is a powerful tool for understanding brain activity, but traditional Echo Planar Imaging (EPI) sequences suffer from limitations like image distortions and relatively lower spatial resolution. These issues can impact the accuracy of detecting subtle brain signals and estimating the complex network interactions within the brain. The paper addresses the need for improved fMRI techniques that can offer higher resolution and better signal quality without compromising temporal performance, particularly for tasks involving fine motor control and intricate neural communication.

Methods

The study directly compares EPIK (EPI with Keyhole) with the standard single-shot EPI sequence on a 3T MRI scanner. Both sequences were optimized to achieve the highest possible spatial resolution under identical imaging conditions (FOV, TR, TE). EPIK utilizes a hybrid approach where the central k-space is fully sampled, and peripheral k-space data is acquired more sparsely and reconstructed by sharing information from consecutive scans. This "keyhole" technique aims to reduce acquisition time and minimize distortions.

Key methodological aspects include:

  • Optimized Sequences: Both EPIK and EPI were tuned for maximum in-plane resolution (2.08 x 2.08 mm² for EPIK with 32 slices, and 3.13 x 3.13 mm² for EPI with 36 slices).
  • Finger-Tapping Paradigm: A visually-guided finger-tapping task was used to reliably elicit BOLD responses in motor and visual brain regions.
  • Point Spread Function (PSF) Analysis: PSF simulations were conducted to quantitatively assess image blurring and spatial resolution differences between EPIK and EPI, both under stationary and dynamic signal change conditions.
  • Dynamic Causal Modelling (DCM): DCM was employed to analyze effective connectivity between identified brain regions (primary motor cortex, supplementary motor area, and visual cortex). Bayesian model selection was used to determine the best model of connectivity, and the robustness of estimated coupling parameters was compared between EPIK and EPI.

Datasets

The study utilized data from a single healthy male volunteer. The fMRI experiments were conducted over multiple sessions, with data acquired using both EPIK and EPI sequences in an alternating order to control for potential habituation effects. No new benchmark datasets were introduced; rather, the study focused on comparing the performance of the two sequences on a well-established fMRI paradigm.

Key Insights and Ideas

  • EPIK's Superior Spatial Resolution and Distortion Reduction: EPIK consistently produced images with higher spatial resolution and significantly reduced susceptibility artifacts, particularly in challenging regions like the sinuses and the fourth ventricle. This is attributed to EPIK's shorter effective echo spacing and optimized k-space sampling.
  • Enhanced BOLD Signal Detection with EPIK: The study found that EPIK yielded significantly stronger BOLD amplitudes in key brain regions involved in the finger-tapping task, including the primary motor cortex (M1), supplementary motor area (SMA), and visual cortex. This suggests EPIK is more sensitive to neural activity.
  • Robust Functional Connectivity Estimation: While both EPIK and EPI identified the same optimal model for effective connectivity (a bidirectional model with task-related modulations), EPIK offered more robust estimation of intrinsic connection parameters, specifically for the connection from the left motor cortex to the right visual cortex.
  • EPIK as a Promising Alternative: The findings strongly suggest that EPIK is a valuable alternative to standard EPI for fMRI, offering a better balance of spatial resolution, signal quality, and temporal performance.

Key Results

  • Spatial Resolution: EPIK achieved an in-plane resolution of 2.08 x 2.08 mm² compared to 3.13 x 3.13 mm² for EPI.
  • Image Quality: Visual inspection and PSF analysis confirmed EPIK's reduced susceptibility artifacts and sharper point spread function, indicating improved spatial localization. The FWHM of EPIK was approximately 41% smaller than EPI.
  • BOLD Amplitude: EPIK showed significantly stronger BOLD responses in several brain regions, with one voxel in the primary motor cortex exhibiting a 36% increase in BOLD signal with EPIK compared to a 10.3% increase with EPI.
  • Functional Connectivity: Bayesian model selection identified the same winning DCM model for both EPIK and EPI. However, the variance of coupling parameters for the intrinsic connection between the left motor cortex and the right visual cortex was significantly smaller for EPIK (F(13,13) = 4.8796, p > 0.05, Bonferroni corrected).

Limitations and Future Work

  • Single Subject Study: The study was conducted on a single healthy male volunteer. While this allows for direct comparison under controlled conditions, future work should validate these findings across a larger and more diverse cohort.
  • No Acceleration Techniques: The study deliberately excluded acceleration techniques (e.g., parallel imaging, partial Fourier) to isolate the benefits of EPIK itself. Future research could explore the combined advantages of EPIK with these acceleration methods for even higher resolution and faster acquisition.
  • Exploration of Resting-State and Event-Related fMRI: The authors suggest that EPIK's properties make it suitable for resting-state and event-related fMRI, but these applications require further dedicated investigation.
  • Columnar Resolution and Sub-Second Acquisition: The potential for EPIK, especially when combined with acceleration, to achieve columnar resolution and sub-second acquisition warrants further exploration.

Summary updated

Abstract

EPI with Keyhole (EPIK) is a hybrid imaging technique that overcomes many of the performance disadvantages associated with EPI. Previously, EPIK was shown to provide a higher temporal resolution and fewer image distortions than EPI whilst maintaining comparable performance for the detection of BOLD-based signals. This work carefully examines the putative enhanced sensitivity of EPIK in a typical fMRI setting by using a robust fMRI paradigm - visually guided finger tapping - to demonstrate the advantages of EPIK for fMRI at 3 T. The data acquired were directly compared to the community standard fMRI protocol using single-shot EPI to ascertain a clear comparison. Each sequence was optimised to offer its highest possible spatial resolution for a given set of imaging conditions, i.e., EPIK and EPI achieved an in-planar resolution of 2.08 × 2.08 mm<sup>2</sup> with 32 slices and 3.13 × 3.13 mm<sup>2</sup> with 36 slices, respectively. EPIK demonstrated a number of clear improvements, such as superior spatial resolution with favourable robustness against susceptibility artefacts. Both imaging sequences revealed robust activation within primary motor, premotor and visual regions, although significantly higher BOLD amplitudes were detected using EPIK within the primary and supplementary motor areas. Dynamic causal modelling, in combination with Bayesian model selection, identified identical winning models for EPIK and EPI data. Coupling parameters reflecting task-related modulations and the connectivity of fixed connections were comparably robust for both sequences. However, fixed connections from the left motor cortex to the right visual cortex were estimated as being significantly more robust for EPIK data.
BibTeX
@inproceedings{yun2019evaluating,
  author = {Seong Dae Yun and Ralph Weidner and Peter H. Weiss and N. Jon Shah},
  title = {{Evaluating the Utility of EPIK in a Finger Tapping fMRI Experiment using BOLD Detection and Effective Connectivity}},
  booktitle = {Scientific},
  year = {2019},
  doi = {10.1038/s41598-019-47341-y},
}
Figure