Abstracts

The folding or gyrification of the cerebral cortex is a developmental process, allowing a large cortical surface area and inter-neuron connectivity without a disproportionate increase in cranial size. Cortical developmental malformations (CDM) may give rise to disruptions in the folding process. The gyrification index (GI), a measure of the degree of cortical folding, is defined as the ratio of the total cortical surface area to superficially exposed surface area. With the advent of MRI, CDMs are increasingly recognised as a cause of epilepsy. An abnormal GI may be a surrogate for subtle CDMs.
Direct visual inspection of two-dimensional images of patients with partial epilepsy show that approximately 25% are normal, without the use of additional quantitative measurement techniques. Inclusion of [italic]in-vivo[/italic] MRI based GI measurements may enhance diagnostic imaging techniques.
The stereological technique of exhaustive vertical sections allows mathematically unbiased estimates of a 3D structure[apos]s surface area based on 2D image slices. The method involves systematic sampling of the surface area around the axis of a fixed vertical plane.
We describe the method and application of the stereological technique of exhaustive vertical sections to MR brain images to produce unbiased estimates of cortical surface area. This technique may ultimately be used for [italic]in-vivo[/italic] measurement of the gyrification index, which to date has not been reported for an epilepsy population.
High resolution 3D SPGR MRI image data of a water filled phantom was used to calibrate and test the method of exhaustive vertical sections. The same MRI sequence was used to acquire brain images from five normal control subjects. Rules were established for identification of the pial layer on these images, on which cortical surface areas were based. Using the computerised stereological method of vertical sections, experiments were carried out to establish appropriate sampling rules to achieve accurate, precise and repeatable measurements.
From repeated measurements, the surface area estimated from the phantom image data measured 340.78cm² [plusmn] 10.3% (mean [plusmn] CE), which compared well ([plusmn] 4.97%) to actual surface area of 358.6cm². Surface area measurements from brain data (1886cm² - 2179cm²) were found to be accurate within experimental error, which ranged consistently between 10.3% and 10.8%.
Robust rules have been established for stereological surface area estimation based on the pial layer. Our experiments show that the technique of exhaustive vertical sections is reliable and accurate. All brain measurements were carried out under identical stereological conditions of sampling slice separation and orientation. Further experiments are required to optimise this method in cortical surface area estimation. The ultimate application of the technique to the measurement of GI is epilepsy may be of clinical utility.
[Supported by: The Irish Brain Research Foundation.]