Related Experiment Videos
Discrete pre-processing step effects in registration-based pipelines, a preliminary volumetric study on T1-weighted
Nathan M Muncy1, Ariana M Hedges-Muncy1, C Brock Kirwan1,2
1Department of Psychology, Brigham Young University, Provo, Utah, United States of America.
Plos One
|October 13, 2017
Summary
MRI scan pre-processing significantly impacts region of interest (ROI) volume estimations, particularly affecting the hippocampus, putamen, and middle temporal gyrus. This study quantifies these effects within a single pipeline, offering noise correction for improved accuracy.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Radiology
Background:
- Pre-processing Magnetic Resonance Imaging (MRI) scans is standard before volumetric analysis.
- Existing studies compare different pipelines, but the impact of individual pre-processing steps remains unclear.
- These steps alter voxel intensities, spatial orientation, and data size, potentially affecting volumetric outcomes.
Purpose of the Study:
- To quantify the specific effects of individual pre-processing steps on volumetric measures within a single MRI pipeline.
- To test the hypothesis that pre-processing steps significantly impact region of interest (ROI) volume estimations.
- To provide a method for correcting noise introduced by pre-processing.
Main Methods:
- Utilized T1-weighted MRI scans from 115 participants in the OASIS dataset.
- Applied a step-wise pre-processing pipeline to scans, assessing volume estimations after each step.
- Performed repeated-measures analyses to evaluate the effects of pipeline steps, scan-rescan variability, and repeated pipeline runs.
Main Results:
- A significant main effect of pipeline step on volumetric measures was detected.
- An interaction between pre-processing steps and specific ROIs (hippocampus, putamen, middle temporal gyrus) was observed.
- No significant effects were found for scan-rescan consistency or repeated pipeline runs.
Conclusions:
- Individual pre-processing steps demonstrably influence volumetric estimations in MRI studies.
- The observed interaction highlights that the impact of pre-processing varies depending on the specific brain region.
- A novel correction for pre-processing-induced noise is proposed to enhance data reliability.