Spatiotemporal Isosurfaces of Physioinformatic Data
A guide to spatialization of information to enhance perceptualization

Jeff Sale
UCSD - San Diego Supercomputer Center

EEG Evoked Potential, 1 second
ECG Isosurface, 2 seconds of data

These are images of human bioelectric activity, such as electroencephalographic (EEG) and electrocardiographic (ECG) spatiotemporal isosurfaces rendered using the powerful AVS data rendering software on a Stardent Titan 3000 workstation (MIPS 3000 cpu). We did this work a few years ago and are belatedly getting it up on the web for you to see. Your comments and suggestions are welcomed.

What are these images? Click for an explanation.

Alpha Bursts: The image to the left was created by calculating power in the alpha band (8-13 Hz) using a fast fourier transform (fft) on the eeg in Matlab, and outputing to AVS.

EEG Evoked Potential: Created from EEG data acquired using the P300 "oddball paradigm". This image comprises roughly one second of data. The blue region is N100, the red region is P300.

EEG Evoked Potential, 1 second

EEG Evoked Potential, 1 second

EEG Evoked Potential - Detail: Detail of same data as above.

Alpha Bursts: Created by calculating power in the alpha band (8-13 Hz) using a fast fourier transform (fft) on the eeg in Matlab, and outputing to AVS.

Alpha waves

Raw Alpha Waves:
These are alpha waves acquired from a human subject, relaxed, eyes closed, comprising roughly 3-4 seconds of activity. This particular subject had especially clean and strong alpha activity with eyes closed. This is apparent from the 'striped' red/blue pattern we see in the data. This is typical of very coherent (in phase) sinusoidal data.

Raw Alpha Waves - Detail:
This is a detail of the same waves, focusing on a particular area comprising about one second of data. You can see that part of the eeg is slightly out of phase with the rest of the data.

Alpha waves, detail, 1 second

EEG from Electroconvulsive Therapy:
This data was acquired intraoperatively from a patient undergoing ECT. Note the "helical" pattern as time increases from left to right.

ECG Isosurface, 2 seconds of data

Electrocardiogram (ECG):
This is two heartbeats worth of electrocardiographic data (ECG) acquired by placing a 5x5 grid of electrodes on the subject's chest. Note the repeating pattern of the beats.

EEG Correlation Dimension:
This is my personal favorite. The others were easy to create compared to this one. It took me a week to write the code and the Stardent 2 weeks to calculate the values. I wrote the algorithm to calculate the correlation dimension, similar to the fractal dimension, which is a measure of the complexity of a system, or more generally, a measure of how a system fills a space, real or abstract. The rest was simple. The data consists of one minute of EEG, the first 30 seconds the subjects eyes were open, then they shut their eyes for the remaining 30 seconds. Time goes from left to right. As usual, the visual cortex corresponds to the lower half of the volume, which is where one would expect to see the most significant change when the eyes are closed, and this is in fact quite distinct.

EEG Correlation Dimension, 1 minute of data

It is difficult to explain what these images are, but click here to get a glimpse...

last updated 3/12/00