Computer aided peripheral arterial disease diagnosis and blood pressure estimation via pulse oscillation

Peripheral arterial disease (PAD) is a collective term for all diseases caused by the obstruction of large peripheral arteries. Early diagnosis allows for faster treatment and the reduction of further tissue destruction. In this paper we present a novel approach for early PAD diagnosis based on the pulse oscillation measurement of any extremity. In contrast to standard pulse oscillography we record all pulse oscillations during intervals of five seconds at constant cuff pressures. Different statistical parameters describing the single pressure steps are then used as features to automatically classify PAD positive and negative persons. We evaluate nine features that are extracted from a pre-classified data pool containing the pressure step oscillograms (PSOs) of 50 healthy and 50 ill persons. By means of receiver operation characteristic (ROC) analysis the most promising features for classification are determined. Additionally, the systolic and diastolic blood pressure is automatically deduced from the feature showing the best classification performance.