An Adaptive Low-Complexity Abnormality Detection Scheme for Wearable Ultrasonography

Abstract

Doppler ultrasonography (DUS) is widely used in medical diagnosis due to its low-cost, non-invasive nature, and real-time operation. Its applications have further expanded with the emergence of point-of-care and wearable devices, the demand for which is rapidly increasing. However, current DUS abnormality detection methods are too computationally intensive for such resource-constrained platforms. This brief presents a low-complexity real-time abnormality detection scheme that enables development of wearable DUS devices. It uses an approximated Fourier transform and a novel greedy algorithm to detect spectrogram envelopes on-the-fly from the stream of samples, thus significantly reducing power and area requirements while achieving a detection accuracy of 96% on a mixture of 25 normal and abnormal test cases. A real-time ASIC implementation of the scheme in 180-nm CMOS consumes 16.8 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\mu }\text{W}$ </tex-math></inline-formula> at a clock frequency of 80 kHz while occupying a layout area of 0.64 mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> .