hrf

Problem

This was one of the projects of my first co-op at Massachusetts General Hospital's Martinos Center for Biomedical Imaging. Most often, MRI scans can take upwards of an hour. In that time, it is easy for a person to become drowsy. There had not yet been data and analysis on if there is a difference in hemodynamic response functions between those "drowsy" subjects and more "alert" subjects.

Objective

To characterize arousal modulations shown in the hemodyanmic response function (HRF) in "drowsy" and "alert" subjects.

Skills Employed

MATLAB
Linux environment

Figure 1. Table of the total number of missed cues for each subject, and graph of subject 02 during the MRI scan. Graph highlights the switches of the MRI, the task paradigm sections, and the key presses by the subject.

Approach

Given behavioral data from previous projects, I developed code in MATLAB to classify arousal levels. After classifying the arousal levels, the derived HRF was compared in the subject, run, and trial level for both alert and drowsy subjects. Essentially I:

Wrote MATLAB programs to infer subject’s arousal levels based on task behavioral data
Characterized and compared hemodynamic changes between high- vs. low-arousal conditions

Image of Invalid Key Presses

Table 1. This table highlights the percent of invalid keypresses throughout a single run for all subjects.

Table 2. This table highlights the percent of missed keypresses throughout a single run for all subjects.

Results

The developed MATLAB code successfully characterized compared hemodynamic changes between high- vs. low-arousal conditions.

Image of Run Level

Figure 2. Highlights the run level comparisons of the hemodynamic response funciton (HRF) of the drowsy and alert subjects during two types of tasks.

Figure 3. Highlights the trial level comparisons of the hemodynamic response funciton (HRF) of the drowsy and alert subjects during two types of tasks.

modulation of arousal levels on stimulus-driven hemodynamic responses