3941
An Optimized Processing Procedure for Fetal Resting State Imaging1Department of Neuroradiology, San Raffaele Scientific Institute, Milan, Italy, 2Department of Gynecology, San Raffaele Scientific Institute, Milan, Italy, 3Department of Nuclear Medicine, San Raffaele Scientific Institute, Milan, Italy
Synopsis
Human fetal brain development is marked by the rapid growth of connections between brain structures during the last trimester of pregnancy. Crucially, data processing of fetal rs-functional scans (rs-fMRI) results largely affected by high degrees of movement, which impose severe constraints on interpretation of functional activation patterns leading to potentially biased results. In this work we present a standardized yet flexible preprocessing and normalization procedure allowing for data loss minimization (i.e. allowing to preserve the highest amount of volumes while consistently excluding outliers) and thus fostering reliable group-level inferences.
Introduction and Aims of the Work
Human fetal brain development is marked by the rapid growth of connections between brain structures during the last trimester of pregnancy 1.While structural brain connectivity has been consistently described in structural terms 1, little is known about functional coupling in fetal brain structures2.
Crucially, data processing of fetal rs-functional scans (rs-fMRI) results largely affected by high degrees of movement, which impose severe constraints on interpretation of functional activation patterns leading to potentially biased results. To this end a standardized yet flexible preprocessing and normalization procedure is needed in order to minimize data loss (i.e. preserving the highest amount of volumes while consistently excluding outliers) and thus allowing for reliable group-level infereces.
METHODOLOGICAL AIM: The first aim of this work is to develop a consistent and standardized processing procedure for rs-fMRI scans, with two main goals:
1) To develop a reliable outlier scans detection procedure, maximizing signal quality and minimizing scans loss.
2) To develop a flexible preprocessing procedure allowing normalization to a standardized template space, in order to allow for 1st (i.e. subject level) and 2nd (i.e. group level) inferences.
THEORETHICAL AIM: The second aim of this work is to test the outcomes of this newly developed processing procedure by investigating the development of Thalamo-Cortical fetal connectivity at rest.
Materials and Methods
We performed fetal structural and functional resting-state imaging (rs-fMRI) in a sample of 9 women at their third trimester of pregnancy (Gestational Week mean 30, ±3.1; range 26 : 35.6) (Age mean 30.9, ±6.8; range 9 : 40). For each subject 5/6 Resting State fMRI sessions (60 vols, 25 slices, 3mm, ~2 min each) and a 2D axial structural scan (25 slices, 3mm, ~1 min) were acquired on a Philips 1.5T scanner (16-channel body coil).
First reoriented, skullstripped and realigned functional images were evaluated in terms of frame-to-frame and frame-to-mean movement and intensity signals changes 3 , in order to identify outlier scans resulting from both i) sudden and large movements and ii) frequent and small movements. As a result a regressor coding for outliers volumes was created for each subject and included into all subsequent 1st level analyses models. Each structural scan was then normalized to a fetal template and the resulting normalization parameters were applied to functional scans, in order to allow for 2nd level, group statistics (Figure 1).
Finally, normalized rs-fMRI volumes were smoothed and input into statistical models aimed at testing for patterns of positive connectivity between the L and R thalamus and the whole-brain.
Results
- The outliers detection procedure allowed us to consistently exclude outlier volumes (i.e. volumes suffering from both sudden, frequent, large and subtle movements), while preserving ~60% volumes (i.e. between 180 and 220) per subject.
- The normalization procedure showed substantial consistency across subjects, allowing us to investigate and make inference on the evolution of functional connectivity in a group of brains differing in terms of gestational week and, thus, shape.
- As a result, 2nd level analyses showed diffused patterns of thalamo-cortical connectivity (p<.001 unc) (Figure 2), consistent with previously reported evidence4,5.
Conclusions
- Fetal functional datasets are characterized by limited data availability (given the specificity of the required scanning procedures) and high data exclusion rates (due to the intrinsic susceptibility to movement of the fetus). To this end we present a standardized yet flexible processing procedure, allowing for consistent volumes preservation and reliable group-level statistics even when working with a small data sample.
- Although suggestive, the results presented in this work will need further testing in a larger-scale population. This will be necessary in order to fully understand the evolution and nuances of functional connectivity development throughout gestation.
Acknowledgements
No acknowledgement found.References
1 Kostović, I., & Jovanov-Milošević, N. The development of cerebral connections during the first 20–45 weeks’ gestation. In Seminars in Fetal and Neonatal Medicine; 2006, December; Vol. 11, No. 6, pp. 415-422.
2 Thomason, M. E., Scheinost, D., Manning, J. H., Grove, L. E., Hect, J., Marshall, N., ... & Hassan, S. S. Weak functional connectivity in the human fetal brain prior to preterm birth. Scientific reports 2017; 7, 39286.
3 Power, J. D., Barnes, K. A., Snyder, A. Z., Schlaggar, B. L.,
& Petersen, S. E. Spurious but systematic correlations in functional
connectivity MRI networks arise from subject motion. Neuroimage 2012; 59(3),
2142-2154.
4 Lagercrantz, H., & Changeux, J. P. The
emergence of human consciousness: from fetal to neonatal life. Pediatric
research 2009, 65(3), 255;
5 Schöpf, V., Kasprian, G., Brugger, P. C., &
Prayer, D. Watching the fetal brain at ‘rest’. International Journal of
Developmental Neuroscience 2012; 30(1), 11-17.
Figures
