Automated Quantification of the Cortical Surface

Period of Performance: 05/15/1998 - 02/28/1999

Unknown

Phase 1 SBIR

Recipient Firm

Cortechs Labs, Inc.
La Jolla, CA 92037
Principal Investigator

Abstract

Most of the neurons in the human brain lie in the cerebral cortex, a two dimensional continuous sheet less than 5 mm thick but with an area of about 2000 cm2. The cortical sheet is highly and irregularly folded, rendering its quantitative measurement and visualization highly problematic. We have developed an automatic method for constructing a detailed, accurate and mathematically correct description of individual cortical surfaces from the structural MRI. The proposed research will further validate the accuracy of this method using segmented closely- spaced histological sections. Simulations will quantify the effects of noise, partial volume and image nonuniformities. The method s power will be enhanced using a training algorithm. This method has broad applications in basic neurology and surgical planning, including functional mapping by painting activations onto the reconstructed surface, automatic skull-stripping, and visualization of cortical blood vessels. Cortical surface reconstruction also provides powerful constraints for localizing electromagnetic activity, such as epileptiform spikes. Quantification of cortical thickness, area, curvature and volume may also assist the diagnosis of neurodegenerative diseases. PROPOSER COMMERCIAL APPLICATION The proposed software would be unique in its capacity to automatically derive quantitative measures of the human cortical surface from MRI. Such measures could become an essential part of neurological diagnosis. In addition the reconstructed cortical surface could become a standard format for the display and integration of functional and structural brain measurement in both the research and clinical communities.