Using Principal Component Analysis to Detect the Wavefront of Cortical Spreading Depression

Abstract

Cortical spreading depression (CSD) is an important neurophysiologic phenomenon which correlates with migraines and focal cerebral ischemia. A series of images are acquired by optical intrinsic signal imaging (OISI) during CSD in rats. To delineate the spatiotemporal evolution of CSD effectively, principle component analysis (PCA) has been proposed for analyzing the optical imaging data (X), which are 300 frames, and each frame is 120times160 pixels. The first 80 spatial components (P) with the corresponding coefficient matrix (A) are sufficient to reconstruct all the original imaging data without useful information loss (X = PA). Successive image subtraction (running subtraction) can be done by processing the matrix A, which takes the form deltaX = PB. The running subtraction based on PCA significantly sharpens the boundary of leading and trailing edges (wavefront) of the CSD wave. Thus, PCA proves to be an appropriate approach to detect the wavefront of CSD.