Automated Diagnosis of Autism Using Fourier Series Expansion of Corpus Callosum Boundary

We explored the possibility of developing an automatic diagnostic tool for detecting autism based on MRI measurements. Since the two previous structural imaging studies [1] [2] strongly suggested there were significant abnormality in the corpus callosum (CC) region, the methodology is concentrated in this area. For this purpose, we have developed a new framework for representing and classifying the CC boundary as a Fourier descriptor [3] [6]. The Fourier coefficients can be viewed as a multivariate measurement that characterizes the CC boundary, and later feed into a classification algorithm.