Harmonization of an SMS-EPI fMRI protocol using Pulseq: Sequence implementation and ABCD QC metrics

2384 

Abstract Submission 

Scott Peltier¹, Maximillian Egan¹, Rex Fung¹, Qingping Chen², Maxim Zaitsev², Jon-Fredrik Nielsen¹

¹University of Michigan, Ann Arbor, MI, ²University Medical Center Freiburg, Freiburg, Germany

Scott Peltier  
University of Michigan
Ann Arbor, MI

Maximillian Egan  
University of Michigan
Ann Arbor, MI

Rex Fung  
University of Michigan
Ann Arbor, MI

Qingping Chen  
University Medical Center Freiburg
Freiburg, Germany

Maxim Zaitsev  
University Medical Center Freiburg
Freiburg, Germany

Jon-Fredrik Nielsen  
University of Michigan
Ann Arbor, MI

Harmonizing fMRI acquisition protocols for multi-site studies can be a difficult and laborious task that may involve custom pulse sequence programming on each vendor platform and continued maintenance across scanner software updates. Moreover, protocols are typically only matched with respect to the high-level protocol parameters accessible to the end user such as field-of-view and resolution; subtle but potentially important differences in, e.g., sequence timing or RF and gradient pulse shapes, are typically not known to the researchers and are beyond their control. Furthermore, each vendor implements their own image reconstruction and post-processing (e.g., image filtering) algorithms. The impact of these vendor-specific protocol implementation details on the reproducibility of functional, structural, and quantitative MRI measures is generally unknown.

Pulseq is an open, vendor-independent MR pulse programming platform that allows an MRI sequence to be created and analyzed in interactive programming environments such as MATLAB or Python, and executed on hardware using sequence-agnostic interpreters [Layton 2017]. Mature interpreters for the Pulseq sequence specification now exist for Siemens and GE scanners [Layton 2017, Nielsen 2018]. In addition, two independently-developed interpreters for Philips were recently reported [Shaik 2023, Roos 2023]. The Pulseq platform enables, for the first time, exact harmonization of all low-level sequence details across sites and vendor platforms. Pulseq also greatly simplifies protocol maintenance, since only the interpreter needs to be recompiled every software upgrade. Here we report on a Pulseq implementation of a simultaneous multi-slice echo-planar imaging (SMS-EPI) protocol that is similar to the fMRI protocol used in the ABCD study, along with quality control (QC) metrics on two scanners from one vendor (GE) following the ABCD QC protocol.

Sequence design and scanner implementation: Using the Pulseq MATLAB toolbox we designed the SMS-EPI sequence shown in Figure 1, which consists of a fat saturation pulse, SMS excitation (multiband factor 6), and EPI readout. We compiled the GE Pulseq interpreter source code for three different software versions: For MR30.2 for the purpose of simulating the sequence with GE's 'Pulse Studio' sequence simulator; for MR29.1 for execution on a GE 3T MR750 scanner; and for RX28.0 for execution on a GE 3T UHP scanner.

Data acquisition and QC metrics: We scanned the FBIRN phantom using the Pulseq SMS-EPI sequence on both scanners, using separate 32 channel Nova Medical head arrays. A total of 388 volumes were acquired, with the first 11 discarded (resulting in a time series duration of 5:01). We also acquired corresponding data using the ABCD QC protocol, and calculated several QC metrics, based on the fBIRN protocol, and as implemented in the ABCD study [https://github.com/ABCD-STUDY/FIONA-QC-PHANTOM]. Metrics were compared against average values obtained using the corresponding ABCD vendor-native sequences.

Figure 1a shows the Pulseq SMS-EPI sequence in the Pulse Studio simulator, i.e., exactly as it appears on scanner hardware. Sequence waveforms and timing are identical to the Pulseq file description (not shown). Figure 1b shows representative images obtained with this sequence.

Table 1 summarizes the calculated QC metrics. We observe that Pulseq performs favorably against the ABCD protocol. The higher SNR in the ABCD UHP acquisition is partly ascribed to the acquisition of full ky, as opposed to the fractional ky on the MR750 and in the Pulseq acquisition. As the Pulseq acquisition is the same across scanners, SNR values are more similar across the two scanners as compared to the ABCD protocol.

We implemented a Pulseq version of the ABCD SMS-EPI protocol and obtained similar or improved QC metrics compared to the vendor-native protocols. Future work will investigate its performance across vendors and different sequence types.

BOLD fMRI ²

Imaging Methods Other ¹

Acquisition

FUNCTIONAL MRI

Open-Source Software

Other - Harmonization