Principal Investigator:
E. R. John, Ph.D., Laboratories of Brain Research
New York University School of Medicine

The amount and accuracy of the information obtained from a PET scan by physicians depends upon their training, experience, ability and attention to details. Some of the information present in the PET images may be missed, even by experienced personnel, because the image intensity variations are too complex to be comprehensively evaluated or too slight to be detected by visual inspection. These difficulties in the interpretation of PET images are due to the lack of precise definition for a normal image and the absence of a method to quantify the significance of apparent deviations from the normal range.

Project Description
Dr. John has successfully introduced the statistical evaluation of electroencephalograms (EEG) and evoked potentials (EP) . He and his research team have over the last four years invented and implemented a computer-based method that analyzes PET scans of the brain and produces a differential diagnosis of subtle brain dysfunctions.

A normative database has been assembled from PET scans images of a group of healthy, normal subjects. The method is based on a statistical evaluation of the differences between a patient’s PET scans and this normative data.

First, all slices are translated to the centroid (center of mass), midline axes aligned, and subjected to shape and size normalization. Then, the analysis system compares the mean value of each pixel in the normal group and the value of each corresponding subject pixel to estimate the degree of abnormality, using a pixel by pixel Z-transformation.

The degree of abnormality of the subject at each pixel, calculated from the Z-score, is displayed in a topographic map, color-coded to reflect the significance of deviations from the normative values. Red represents significant positive Z value, or deviation of increased absolute metabolic rate. Green represents Z values not significantly different from zero, or normal metabolic rate. Blue represents significant negative Z values, or deviations of decreased absolute metabolic rate.

Corrections for differences in overall uptake reflecting non-specific factors, such as arousal level, are made by similar statistical evaluations of relative regional metabolism.

Major Applications
The method can be applied in the quantitative detection and classification of brain dysfunctions as well as in the evaluation of the efficacy of treatments. This new interpretation of PET images significantly enhances the clinical use of PET scans.

Patent Status
Patent pending

For further information, please contact:
New York University
Industrial Liaison/Technology Transfer
650 First Avenue, New York, N.Y. 10016
Tel: (212) 263-8178 Fax: (212) 263-8189