Selection of fluidic flow region of interest in a medical device using proper orthogonal decomposition

The present study aims at identifying an optimal spatial region of interest (ROI) by means of image analysis of a temporal data set obtained by a novel medical device. To achieve this, a lower-dimensional approximation technique, Proper Orthogonal Decomposition(POD) is used since it is well-known for highlighting a region of interest in a flow field effectively. Primary attention was devoted to studying the regions of the highest absorbance and intensity that were successfully computed by mode one of the decomposition. Due to the running cost of POD with higher memory storage requirements, some techniques related to dimensionality reduction, such as averaging data, economy singular value decomposition (SVD) tool, and preconditioning techniques, were applied. Results indicated that the region of interest chosen merely by visual inspection differs from the one chosen through POD in terms of the highest change in absorbance and intensity. This simple methodology can be easily extended to a variety of applications involving image processing of temporal data.