Bridging the Gap between Quantitative and Clinical MRI
Poster No:
2283
Submission Type:
Abstract Submission
Authors:
Shachar Moskovich1, Oshrat Shtangel1, Aviv Mezer1
Institutions:
1The Hebrew University in Jerusalem, Jerusalem, Israel
First Author:
Co-Author(s):
Introduction:
Quantitative MRI (qMRI) enables in-depth study of brain microstructure and can provide in vivo histological information (Weiskopf et al., 2021). However, its use in clinical settings is limited, with reliance on weighted images that lack microstructural detail. Our study introduces three new weighted image ratios, including T1w/PDw and ln(T2w/PDw), to approximate qMRI parameters R1 and R2. Validated with lipid phantoms and human datasets, these ratios, alongside a modified T1w/T2w (T1w/ln(T2w)), effectively represent qMRI parameters. Applied to Parkinson's disease data, they reveal microstructural differences in key brain regions, enhancing clinical MRI analysis.
Methods:
We employed three datasets:
Lipid phantom data (Shtangel and Mezer, 2020) with various lipid types.
HUJI subjects: from Filo et al., 2019, qMRI and weighted data of 22 healthy individuals (age 41±19.6).
PPMI subjects: 4 ROIs (caudate, putamen, globus pallidus, and midbrain) of 99 Parkinson's patients, age 65±6, and 46 controls, age 65±6 from Parkinson's Progression Marker Initiative were analyzed.
Using the typical T1 and T2 signal equations and the previously measured R1 and R2, we produced synthetic images for the phantom data.
Lipid phantom data (Shtangel and Mezer, 2020) with various lipid types.
HUJI subjects: from Filo et al., 2019, qMRI and weighted data of 22 healthy individuals (age 41±19.6).
PPMI subjects: 4 ROIs (caudate, putamen, globus pallidus, and midbrain) of 99 Parkinson's patients, age 65±6, and 46 controls, age 65±6 from Parkinson's Progression Marker Initiative were analyzed.
Using the typical T1 and T2 signal equations and the previously measured R1 and R2, we produced synthetic images for the phantom data.
Results:
We found perfect linear relationships between synthetic T1w/PDw and R1 map, and between synthetic ln(T2w/PDw) and R2 map (R2=1) in phantoms. Thus, weighted images can perfectly estimate quantitative maps in a noise-free environment.
Next, we evaluated the correlation of R1 with T1w, T1w/T2w, and T1w/PDw. We found T1w/PDw showed the strongest correlation in phantoms (Fig. 1A-C). Similarly, we calculated the correlations of R2 with R2w, T1w/T2w, and ln(T2w/PDw). We found that the strongest correlation was with ln(T2w/PDw) in phantoms (Fig. 1D-F).
We then carried the same analysis we did for phantoms and tested it on HUJI subjects. We replicated the phantoms results for R1 (Fig. 1G-I) and R2 approximations (Fig. 1J-L). Using B0 images from DTI scans as T2w, we observed the same results.
We also ask whether T1w/T2w could more accurately approximate the quantitative maps R1 and R2. We calculated two modifications for this ratio: ln(T1w/T2w) and T1w/ln(T2w), in humans. We found T1w/ln(T2w) to be a better approximation of R1 than is T1w/T2w. We also found the highest correlation of R2 to be with ln(T1w/T2w).
Finally, to test whether the new ratios can be utilized in a clinical setting, we analyzed weighted MRI data from PPMI subjects. We calculated four structural measurements: T1w/PDw, ln(T2w/PDw), T1w/ln(T2w), and T1w/T2w. For each ratio, we assessed group differences between patients and controls in ROIs that are thought to be affected in Parkinson's disease (Dickson, 2012). We found differences between groups only in our new measurements (Fig. 2). These results suggest there is additional information in our new ratios.
Next, we evaluated the correlation of R1 with T1w, T1w/T2w, and T1w/PDw. We found T1w/PDw showed the strongest correlation in phantoms (Fig. 1A-C). Similarly, we calculated the correlations of R2 with R2w, T1w/T2w, and ln(T2w/PDw). We found that the strongest correlation was with ln(T2w/PDw) in phantoms (Fig. 1D-F).
We then carried the same analysis we did for phantoms and tested it on HUJI subjects. We replicated the phantoms results for R1 (Fig. 1G-I) and R2 approximations (Fig. 1J-L). Using B0 images from DTI scans as T2w, we observed the same results.
We also ask whether T1w/T2w could more accurately approximate the quantitative maps R1 and R2. We calculated two modifications for this ratio: ln(T1w/T2w) and T1w/ln(T2w), in humans. We found T1w/ln(T2w) to be a better approximation of R1 than is T1w/T2w. We also found the highest correlation of R2 to be with ln(T1w/T2w).
Finally, to test whether the new ratios can be utilized in a clinical setting, we analyzed weighted MRI data from PPMI subjects. We calculated four structural measurements: T1w/PDw, ln(T2w/PDw), T1w/ln(T2w), and T1w/T2w. For each ratio, we assessed group differences between patients and controls in ROIs that are thought to be affected in Parkinson's disease (Dickson, 2012). We found differences between groups only in our new measurements (Fig. 2). These results suggest there is additional information in our new ratios.
Conclusions:
Despite the high potential of qMRI, its clinical use is limited, with a reliance on weighted images that lack quantitative detail. Addressing this, our research introduced three novel image ratios to approximate qMRI parameters. T1w/PDw closely approximates R1, while ln(T2w/PDw) effectively represents R2. We refined the T1w/T2w ratio for better accuracy and extended our analysis to DTI databases. Applied to Parkinson's disease data, our methods successfully differentiated patients from healthy controls. This work provides a new approach to utilize widely available weighted images for more detailed neuroimaging in both research and clinical contexts.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s) 2
Modeling and Analysis Methods:
Other Methods
Novel Imaging Acquisition Methods:
Anatomical MRI 1
Diffusion MRI
Keywords:
MRI
STRUCTURAL MRI
WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC
Other - quantitative MRI
1|2Indicates the priority used for review
·Fig. 1 – R1 and R2 can be approximated using T1w/PDw and ln(T2w/PDw) in lipid phantom and human data
·Fig. 2 – Our new measurements reveal group differences between Parkinson’s patients and controls
Provide references using author date format
Filo, S. et al. (2019) ‘Disentangling molecular alterations from water-content changes in the aging human brain using quantitative MRI’, Nature Communications [Preprint]. Available at: https://doi.org/10.1038/s41467-019-11319-1.
Glasser, M.F. and Essen, D.C. Van (2011) ‘Mapping human cortical areas in vivo based on myelin content as revealed by T1- and T2-weighted MRI’, The Journal of Neuroscience [Preprint]. Available at: https://doi.org/10.1523/jneurosci.2180-11.2011.
Dickson, D.W. (2012) ‘Parkinson’s Disease and Parkinsonism: Neuropathology’, Cold Spring Harbor Perspectives in Medicine, 2(8), pp. a009258–a009258. Available at: https://doi.org/10.1101/cshperspect.a009258.
Shtangel, O. and Mezer, A. (2020) ‘A phantom system for assessing the effects of membrane lipids on water proton relaxation.’, NMR in Biomedicine [Preprint]. Available at: https://doi.org/10.1002/nbm.4209.
Weiskopf, N. et al. (2021) ‘Quantitative magnetic resonance imaging of brain anatomy and in vivo histology’, Nature Reviews Physics, 3(8), pp. 570–588. Available at: https://doi.org/10.1038/s42254-021-00326-1.