A robust and sensitive measurement for sulcal depth from geometrical approach

Abstract

Sulcal depth has been used for brain morphological analysis. Many previous approaches for measuring sulcal depth in cortical surfaces have been provided, but they have some limitations due to complex and convoluted sulcal geometry. We propose a novel algorithm for sulcal depth measurement, named adaptive distance transform (ADT). Reflecting sulcal geometry, ADT may be a sensitive method for brain atrophy and have robustness against sulcal variability. We demonstrate that ADT is sensitive approach by applying clinical data set. The group differences of mean sulcal depth between 25 patients with Alzheimer's disease (AD) and 25 normal controls using ADT and previous approaches (Euclidean and geodesic distance) are compared. The area under the receiver operating characteristic curve of three approaches is calculated for classifying performance of the algorithms. The area of receiver operating characteristic curve indicated that all three algorithms are sensitive measure to classifying subject into controls or AD. To compare robustness of algorithms, we designed simulation data set which is similar to sulcal shape. At the result, ADT has lower error than other algorithms estimated in simulation data. These results are discussed in terms of robust and sensitive measurement of ADT for sulcal depth. (2012 World Congress on Engineering and Computer Science, WCECS 2012San Francisco24 October 2012 through 26 October 2012)