Using fMRI to Guide TMS Target Selection: Reliability and Sensitivity of 3T and 1.5T fMRI Metrics

[DOI: 10.1007/s12021-024-09667-5], published in Neuroinformatics

Background Introduction

The early application of functional magnetic resonance imaging (fMRI) mainly focused on inferring cognitive processes. However, modern medicine is gradually extending its use to more clinical applications, such as preoperative planning and disease differentiation. In the clinical application of repetitive transcranial magnetic stimulation (rTMS) treatments, fMRI has shown potential in optimizing TMS target selection and improving treatment outcomes. Specifically for major depressive disorder (MDD) patients, the U.S. Food and Drug Administration (FDA) has approved an fMRI-guided individualized treatment protocol. However, most current studies focus on 3T scanners, while 1.5T MRI is more common in many primary hospitals. Therefore, a systematic evaluation of fMRI metrics from both 1.5T and 3T scanners may provide a reference for the clinical application of personalized and precise TMS stimulation guided by fMRI.

Research Source

This paper was written by Qiu Ge, Matthew Lock, Xue Yang, Yuejiao Ding, Juan Yue, Na Zhao, Yun-Song Hu, Yong Zhang, and Minliang Yao, belonging to the Affiliated Hospital of Hangzhou Normal University, Zhejiang Provincial Key Laboratory of Cognitive Impairment Research, Institute of Psychology of Hangzhou Normal University, Affiliated Deqing Hospital of Hangzhou Normal University, Shanghai International Studies University, among other institutions. The paper was accepted and published by Neuroinformatics on April 30, 2024.

Research Process and Methods

Experimental Design

The study recruited 20 healthy college students (10 females, aged 19-27), through internet advertisements. The experimental protocol was approved by the Ethics Committee of the Affiliated Hospital of Hangzhou Normal University, and participants signed informed consent forms. The experiment used 3T and 1.5T MRI scanners for a comparative study, with each participant undergoing both 3T and 1.5T scans on the same day, with an interval of approximately 3.87 hours.

Data Collection and Preprocessing

During each scan, three resting-state fMRI scans were first performed, followed by 3D-T1 imaging, and finally a finger-tapping task fMRI scan. Specific scanning parameters are as follows: - 3T Scanner: Multiband GRE-EPI sequence, TR/TE = 700/30ms, FOV = 216mm, matrix = 72 × 72, thickness/gap = 3/0mm. - 1.5T Scanner: Single-shot GRE-EPI, TR/TE = 3000/40ms, FOV = 224mm, matrix = 64 × 64, thickness/gap = 3.5/0mm.

Experimental Methods

Finger-Tapping Task fMRI

Participants performed the finger-tapping task under self-initiated (SI) and visually-guided (VG) conditions. Each task included 42 seconds of SI finger-tapping and VG finger-tapping alternately, with a total task duration of approximately 10 minutes.

Data Preprocessing and Analysis

• RS-fMRI data were preprocessed using the DPABI (V6.1) toolbox, including slice time correction, motion correction, regression of head motion effects, and spatial normalization.
• Finger-tapping task fMRI data were analyzed in SPM12, including slice time correction, motion correction, spatial normalization, and spatial smoothing.
• Individual brain contrast maps were generated using General Linear Model (GLM), and the SI and VG conditions data were analyzed.

Reliability and Sensitivity Assessment

The study evaluated the reliability of fMRI metrics by calculating the consistency of peak activation locations (using intraclass individual distance calculations) and local fMRI metrics (including Amplitude of Low-Frequency Fluctuations (ALFF), Percentage of Amplitude of Fluctuation (PerAF), wavelet-based low-frequency fluctuation amplitude (Wavelet-ALFF), Regional Homogeneity (ReHo), and Degree Centrality (DC)) by test-retest reliability (using Intraclass Correlation Coefficient (ICC)). Sensitivity assessments included comparing differences in fMRI metrics under eyes-open and eyes-closed conditions, as well as evaluating the consistency of peak activation intensity and functionally connected (FC) defined TMS target locations.

Data Processing and Analysis

Consistency of Peak Activation Locations

Peak activations were located in motor-related brain regions (e.g., precentral gyrus). The peak activation locations for 3T and 1.5T scan data were very consistent in these regions, with intraclass individual distances ranging from 15.8 to 19mm.

Consistency of FC-Defined Target Locations

The study selected several deep brain regions as seeds (ROI), including the anterior cingulate cortex (ACC), dorsolateral prefrontal cortex (DLPFC), etc. Functional connectivity was calculated by Pearson correlation and normalized using Fisher’s r-to-z transformation. The results showed similar consistency in peak FC positions in both 1.5T and 3T data, with intraclass individual distances ranging from 19.7 to 31.2mm.

Reliability of Local fMRI Metrics

• Metrics such as ALFF and PerAF exhibited moderate to excellent test-retest reliability in both 1.5T and 3T data.
• PerAF and Wavelet-ALFF demonstrated higher sensitivity, particularly having greater effect sizes and more significant voxels in both 1.5T and 3T scanners.

Reliability of BOLD Signals

The intensity of BOLD signals in both 3T and 1.5T showed excellent test-retest reliability.

Results

• The consistency of finger-tapping task fMRI activation locations, consistency of FC-defined target locations, and reliability of local fMRI metrics were comparable in both 1.5T and 3T scanners.
• The 3T scanner outperformed the 1.5T scanner in sensitivity to detecting changes in resting-state activities, but the overall trends were consistent, with Wavelet-ALFF demonstrating relatively higher sensitivity among all five metrics.

Significance of the Study

This study indicates that although 1.5T is less sensitive than 3T, it performs similarly in several critical metrics (such as peak activation locations, FC target locations, and test-retest reliability of local fMRI metrics). This suggests that 1.5T scanners can be used to guide individualized fMRI TMS target selection, offering potential feasibility for the broader clinical application of fMRI.

Summary

Despite performance differences in some metrics, overall, 1.5T and 3T performances are close, indicating potential application value of 1.5T in fMRI-guided individualized TMS target selection. Future research should continue to evaluate the feasibility of this method in different tasks and larger samples, further promoting the clinical application of fMRI-guided TMS.