Motion Robust MR Fingerprinting Scan to Image Neonates With Prenatal Opioid Exposure

Abstract

Purpose: To explore whether MR fingerprinting (MRF) scans provide motion-robust and quantitative brain tissue measurements for non-sedated infants with prenatal opioid exposure (POE).

Study type: Prospective.

Population: 13 infants with POE (3 male; 12 newborns (age 7-65 days) and 1 infant aged 9-months).

Field strength/sequence: 3T, 3D T1-weighted MPRAGE, 3D T2-weighted TSE and MRF sequences.

Assessment: The image quality of MRF and MRI was assessed in a fully crossed, multiple-reader, multiple-case study. Sixteen image quality features in three types-image artifacts, structure and myelination visualization-were ranked by four neuroradiologists (8, 7, 5, and 8 years of experience respectively), using a 3-point scale. MRF T1 and T2 values in 8 white matter brain regions were compared between babies younger than 1 month and babies between 1 and 2 months.

Statistical tests: Generalized estimating equations model to test the significance of differences of regional T1 and T2 values of babies under 1 month and those older. MRI and MRF image quality was assessed using Gwet's second order auto-correlation coefficient (AC₂) with confidence levels. The Cochran-Mantel-Haenszel test was used to assess the difference in proportions between MRF and MRI for all features and stratified by the type of features. A P value <0.05 was considered statistically significant.

Results: The MRF of two infants were excluded in T1 and T2 value analysis due to severe motion artifact but were included in the image quality assessment. In infants under 1 month of age (N = 6), the T1 and T2 values were significantly higher compared to those between 1 and 2 months of age (N = 4). MRF images showed significantly higher image quality ratings in all three feature types compared to MRI images.

Conclusions: MR Fingerprinting scans have potential to be a motion-robust and efficient method for nonsedated infants.

Level of evidence: 2 TECHNICAL EFFICACY STAGE: 1.

Keywords: MR Fingerprinting; image quality assessment; motion Robustness; neonate imaging; prenatal opioid exposure.

Figure 1:

Example whole brain T1, T2 and R1R2 maps generated from an MRF scan (pt1, 10 days). Each map defines the appropriate myelination pattern expected in a full-term neonate. The posterior limb of internal capsule (red arrow), lateral thalamus (blue arrow), and coronal radiata (yellow arrow) are best highlighted on R1R2 maps.

Figure 2:

Synthetic MR images (A) and sub-voxel tissue fraction maps (B) generated from an example subject (pt2, 15 days). The myelinated posterior limb of internal capsule is best defined on the double inversion recovery (DIR) and myelin water fraction (MWF) images (arrowed) and not visualized on the white matter fraction maps.

Figure 3:

Comparisons of the image quality, scan time and image quality scores of the MRI-T1w and T2w images and the synthetic T1w and T2w images from MRF scans from two neonate subjects. (A) (Pt3, 13 days old), The first MRI-T2w failed due to severe motion. The repeated T2w still showed shading artifacts (arrowed). The following T1w scan was also corrupted by periodic ghosting artifacts due to motion. (B) (Pt4, 6 days old), All the MRI scans (T2w and two repeated T1w) did not provide acceptable image quality. The MR images were corrupted by severe blurring and ghosting artifacts. As a comparison, the synthetic MRF T1w and T2w images from both subjects were free of motion artifacts.

Figure 4:

Summary of the image quality ratings for MRF and MRI by readers and image quality scores: (A) summary of the raw scores collected from the multi-reader image quality assessment by four neuroradiologists (A, B, C, D). Each number in the table is the number of times a reader rated a certain score. (B) bar plots demonstrate the distribution of the scores for each image type.

Figure 5:

Comparison of the number of grade 1 (minor/no artifacts and well visualized structure and myelination) ratings in each of 16 assessment image items between MRI and MRF. MRF consistently rated with lower image artifacts and better anatomical and myelin visualization compared to MRI.

Figure 6:

(A) Gwet’s AC₂ agreement coefficients for the image features amongst the four readers, estimating the inter-reader agreement for each image feature and each image type. (B) Gwet’s AC₂ agreement for each reader among all features in each image type, estimating the inter-feature agreement for each reader.

Figure 7:

Comparison of the quantitative MRF maps from babies aged 10 days and 270 days. (A) images demonstrate differences in the maturation of brain structure as well as the progression of myelination. (B) T1 and T2 values of white matter from eleven babies with opioid exposure. The box plot of each subject shows the T1 or T2 value distribution from eight white matter ROIs. (C) Subjects’ ages correspond to the subject Numbers on x-axis in (B)