Implementation of smartphone based blood flow diagnoses from Doppler spectrogram

This paper introduces a new scheme to identify the blood flow condition based on a processing of the Ultrasound Doppler spectrograms on an Android Smartphone. A set of blood flow spectrograms are processed to denoise and some features are extracted. These features are used in supervised classifiers (support vector machine and k-nearest neighbors) to detect the blood flow abnormalities. Due to better performance of support vector machine, an Android application is implemented based on this classifier for diagnosing arterial diseases.