top of page
signapulse2006Autumn-83370_0052_1.jpg

PREPROCESSING FMRI SCANS:
Improving the Signal to Noise Ratio

BACKGROUND

During my freshman year, I started working in the CONCEPT Lab, which is part of the Western Psychiatric Institute and Clinic of UPMC. The CONCEPT Lab works on Connecting Outcomes, Networks and Cognitions for Early Psychosis Therapeutics.

This is an example of a brain that has been motion corrected

Motion Correction.jpg

This is an example of a brain that has not been motion corrected.

SNRgraphLabeled.png

Signal-to-noise ratio after each preprocessing step

PROJECT OVERVIEW

Before data from fMRI scans can be used in research, it must go through several preprocessing steps. These preprocessing steps help remove unwanted changes to the signal, also known as noise.  In this project, I explored the effect each part of the preprocessing method had on the overall signal to noise ratio (or SNR). 

METHODS

There are five preprocessing steps whose effects I looked at:

  1. Motion correction

  2. Slice timing correction

  3. Gradient distortion correction

  4. Brain extraction

  5. Regularization 

 

The fMRI data was run through a script that performed each of these steps and then saved a copy of the data after each step was completed. 

RESULTS

​To find the signal to noise ratio, an average of the data across subjects was taken and then divided by the standard deviation.   

 

The respective SNRs for each step were then plotted so that the effect of each preprocessing step could be seen in comparison to the others.

ACKNOWLEDGMENTS

Special thanks to:

  • My mentor, Dr. Konasale Prasad

  • Nick Theis, MS (Research Specialist)

  • And the other members of the CONCEPT Lab

bottom of page