0003

Cross-Vendor Multiparametric Mapping of the Human Brain Using 3D-QALAS: A Multicenter & Multivendor Study

Shohei Fujita^1,2,3,4, Borjan Gagoski^4,5, Ken-Pin Hwang⁶, Akifumi Hagiwara¹, Marcel Warntjes^7,8, Issei Fukunaga¹, Wataru Uchida¹, Yuya Saito¹, Towa Sekine¹, Rina Tachibana¹, Tomoya Muroi¹, Toshiya Akatsu¹, Akihiro Kasahara², Ryo Sato², Tsuyoshi Ueyama², Christina Andica^1,9, Koji Kamagata¹, Shiori Amemiya², Hidemasa Takao², Yasunobu Hoshino¹⁰, Yuji Tomizawa¹⁰, Kazumasa Yokoyama¹⁰, Berkin Bilgic^3,4,11, Nobutaka Hattori¹⁰, Osamu Abe², and Shigeki Aoki¹
¹Department of Radiology, Juntendo University, Tokyo, Japan, ²Department of Radiology, The University of Tokyo, Tokyo, Japan, ³Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA, United States, ⁴Department of Radiology, Harvard Medical School, Boston, MA, United States, ⁵Fetal Neonatal Neuroimaging and Developmental Science Center, Boston Children's Hospital, Boston, MA, United States, ⁶Department of Imaging Physics,, MD Anderson Cancer Center, Houston, TX, United States, ⁷SyntheticMR, Linköping, Sweden, ⁸Center for Medical Imaging Science and Visualization (CMIV), Linköping University, Linköping, Sweden, ⁹Faculty of Health Data Science, Juntendo University, Chiba, Japan, ¹⁰Department of Neurology, Juntendo University, Tokyo, Japan, ¹¹Harvard/MIT Health Sciences and Technology, Cambridge, MA, United States

Synopsis

Keywords: YIA, Neuro

Motivation: To address the unmet need for a cross-vendor, multiparametric technique to facilitate data pooling across sites.

Goal(s): To evaluate a vendor-standardized multiparametric mapping scheme based on 3D-QALAS for whole-brain T1, T2, and proton density (PD) mapping.

Approach: Intra-scanner repeatability and inter-vendor reproducibility were evaluated in vivo on five different 3T systems from four vendors (GE, Philips, Siemens, and Canon). Patients with multiple sclerosis were scanned on systems from different vendors to assess the feasibility of the scheme in real-world clinical settings.

Results: 3D-QALAS provided T1, T2, and PD with coefficient of variations <4.0% using 3T scanners from different manufacturers.

Impact: The four major vendors used in this study constitute a considerable portion of the global installation base, demonstrating the value of cross-vendor quantitative technique 3D-QALAS for imaging in clinical sites with multiple vendors, as well as in multicenter research settings.

Introduction

A multiparametric technique^1-5that is compatible with multiple vendors is necessary to pool data among different sites and vendors. This can increase the size and diversity of the applicable patient population, thus enhancing the statistical power.^6,7 However, few technologies have shown compatibility across vendors or provided reproducible values across MRI scanners from different vendors.
One cross-vendor multiparametric technique is 3D-quantification using an interleaved Look–Locker acquisition sequence with a T2 preparation pulse (3D-QALAS).^4,8 Here, we developed a vendor-standardized whole-brain multiparametric mapping scheme based on 3D-QALAS and assessed the intra-scanner repeatability and inter-vendor reproducibility of T1, T2, and proton density (PD) values in five different 3T systems from four major MRI vendors. Further, we performed an inter-vendor validation on patients with multiple sclerosis to assess the feasibility of 3D-QALAS in real-world clinical settings

Methods

Sequence and acquisition protocol
Our prospective multi-institutional study was approved by the local institutional review board of each institution. Participants provided written informed consent. Five 3T scanners manufactured by four vendors (GE Healthcare, Philips Healthcare, Siemens Healthcare, and Canon Medical Systems) were included from two academic institutions. The 3D-QALAS sequence was implemented in the MRI systems of all vendors; corresponding imaging parameters were kept as close as possible across all scanners (Figure 1).

Data Acquisition
Ten healthy volunteers underwent test-retest of vendor-standardized 3D-QALAS sequence with each scanner. Subjects were repositioned between the acquisitions to assess variability due to subject positioning. Each participant underwent imaging on all the scanners within two months. To further assess the feasibility of 3D-QALAS in real-world clinical settings, nine female patients with multiple sclerosis underwent imaging on two scanners from different vendors.

Postprocessing
Whole-brain T1, T2, and PD maps were derived via voxel-wise signal modeling-based parameter fitting on the 3D-QALAS source images using prototype SyMRI software. FreeSurfer⁹ was used to perform brain region-wise analysis with the Desikan-Killiany atlas.¹⁰ The mean T1, T2, and PD values were calculated for each brain structure in cortical gray matter (GM), subcortical GM, and white matter (WM) structures and WM lesions in patients.

Assessment
Intra-scanner repeatability for each scanner and structure was assessed using the within-subject coefficient of variation (CV), defined as the standard deviation [SD] of the three quantitative maps derived from test and retest over the mean of both time points.¹¹ Inter-vendor reproducibility was also assessed using within-subject CV, defined as the SD divided by mean value calculated across all five scanners. Linearity, slope, intercept, and limits of agreement were used for comparisons among scanners. The intraclass correlation coefficient (ICC) using two-way random-effects model was also calculated to assess agreement between scanners.

Results

Quantitative values of five scanners from four vendors
Representative quantitative maps of a healthy subject are shown in Figure 2. The mean intra-scanner CVs for each scanner and structure ranged from 0.4 to 2.6% (Figure 3A). The overall structure-wise test-retest repeatability (mean intra-scanner CVs of the five scanners) were 1.6%, 1.1%, and 0.7% for T1, T2, and PD, respectively (Figure 3B).
Overall, a high inter-vendor reproducibility was observed for all parameter maps (Figure 4A and 4B, ICC = 0.99, 0.97, and 0.98 for T1. T2, and PD, respectively). The structure-wise inter-scanner CV values are shown in Figure 4C.

Inter-vendor patient study
Representative quantitative maps of a patient with multiple sclerosis are shown in Figure 5A. High agreement was observed among the scanners for all structure measurements, including WM lesions (Figure 5B, ICC = 0.98, 0.92, and 0.94 for T1. T2, and PD, respectively). The white matter in healthy individuals and the normal-appearing white matter in those with multiple sclerosis were distinguishable, even when using scanners from different vendors (Figure 5C).

Discussion

In this study, we examined the intra-scanner and inter-vendor reproducibility of quantitative measurements obtained with 3D-QALAS, a cross-vendor whole-brain multiparametric mapping technique. Although the number of participants was small, this study included scanners from four major vendors, covering more than 90% of the global install base¹². Our results show that the T1, T2, and PD values measured with 3D-QALAS were comparable across scanners and vendors.
Importantly, the results demonstrate that the reproducibility coefficient (defined as 2.77 × within-subject CV) was smaller than the difference between the quantitative values of the lesion and the normal tissue: indicating that lesions can be distinguished from normal tissue even when data from scanners from multiple vendors are pooled.

Conclusion

In a cohort consisting of healthy volunteers and patients, T1, T2, and PD values measured with 3D-QALAS were in high agreement across 3T scanners from four major MRI vendors. This validation study may serve as a base for larger multi-institutional studies.

Acknowledgements

No acknowledgement found.

References

1. Jara H, Sakai O, Farrher E, et al. Primary Multiparametric Quantitative Brain MRI: State-of-the-Art Relaxometric and Proton Density Mapping Techniques. Radiology. 2022;305(1):5-18. doi:10.1148/radiol.211519

2. Ma D, Gulani V, Seiberlich N, et al. Magnetic resonance fingerprinting. Nature. 2013;495(7440):187-192. doi:10.1038/nature11971

3. Wang F, Dong Z, Reese TG, et al. Echo planar time-resolved imaging (EPTI). Magn Reson Med. 2019;81(6):3599-3615. doi:10.1002/mrm.27673

4. Fujita S, Hagiwara A, Hori M, et al. Three-dimensional high-resolution simultaneous quantitative mapping of the whole brain with 3D-QALAS: An accuracy and repeatability study. Magn Reson Imaging. 2019;63:235-243.

5. Ma S, Nguyen CT, Han F, et al. Three-dimensional simultaneous brain T1 , T2 , and ADC mapping with MR Multitasking. Magn Reson Med. 2020;84(1):72-88. doi:10.1002/mrm.28092

6. Seiberlich N, Gulani V, Campbell-Washburn A et al. Quantitative Magnetic Resonance Imaging, Volume 1 – 1st Edition. 2020.

7. QIBA Concepts. https://qibawiki.rsna.org/index.php/QIBA_Concepts

8. Kvernby S, Warntjes MJ, Haraldsson H, Carlhäll CJ, Engvall J, Ebbers T. J Cardiovasc Magn Reson. 2014 Dec 20;16:102. Simultaneous three-dimensional myocardial T1 and T2 mapping in one breath hold with 3D-QALAS.

9. Fischl B. FreeSurfer. Neuroimage 2012;62:774–781

10. Desikan RS, Segonne F, Fischl B, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage 2006;31:968–980.

11. Weingärtner S, Desmond KL, Obuchowski NA, et al. Development, validation, qualification, and dissemination of quantitative MR methods: Overview and recommendations by the ISMRM quantitative MR study group. Magn Reson Med. 2022;87(3):1184-1206.

12. https://seekingalpha.com/article/3405705-fonar-corp-undervalued-niche-player-in-the-world-of-mri-scanners

Figures

Figure 1. (A) Overview of the 3D-QALAS sequence. The 3D-QALAS sequence consists of five equally spaced 3D FLASH readouts with T2 preparation and inversion recovery preparation modules. This cycle is repeated to fill the five sets of k-spaces, one for each readout. The MRI images with green borders are the source images corresponding to the respective acquisition blocks. (B) Standardized sequence parameters of 3D-QALAS. A minimal acceleration technique was applied to reduce confounding effects from vendor-specific acceleration methods.

Figure 2. Representative T1, T2, and proton density (PD) maps from a healthy traveling subject. Each traveling subject underwent a test-retest session on each of the five 3T MRI scanners from four vendors (i.e., GE Healthcare, Philips Healthcare, Siemens Healthcare, Canon Medical Systems). Subjects were repositioned between the acquisitions. For analysis of results, the names of the scanners were anonymized to avoid bias and to ensure that the study does not show superiority/inferiority of a specific scanner or manufacturer over others.

Figure 3. Intra-scanner repeatability. (A) Boxplot showing structure-wise coefficient of variation (CV) for each scanner. Each dot represents a brain region on the Desikan-Killiany atlas. (B) Region-wise intra-scanner repeatability. For each brain structure, the intra-scanner CV value of the five scanners was calculated for each subject, and the mean value across subjects was displayed on schematic brain illustrations.

Figure 4. (A) Scatter and Bland-Altman plots showing inter-scanner and inter-vendor agreement. Each dot represents a brain region. The solid line is the identity line. The dashed line is the linear fit for each scanner. (B) Bland–Altman plots showing inter-scanner biases. (C) Region-wise inter-scanner CVs for each structure calculated using the 1^st scan from each scanner. CV, coefficient of variation; GM, gray matter; PD, proton density; WM, white matter.

Figure 5. Inter-vendor evaluation on multiple sclerosis patients. (A and B) Representative maps and inter-vendor agreement of two scanners from different manufacturers. Each dot represents a brain region. The dashed line is the linear fit. (C) Relaxation times of normal-appearing WM in healthy controls and patients with MS. Each point represents a mean value within corpus callosum. There is a distinction between the normal-appearing WM of healthy controls and patients with MS, regardless of the scanner type. HC, healthy control; MS, multiple sclerosis; WM, white matter.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)

0003

DOI: https://doi.org/10.58530/2024/0003