Compressed-sensing photoacoustic computed tomography in vivo with partially known support

Abstract

Compressed sensing (CS) can recover sparse signals from under-sampled measurements. In this work, we have developed an advanced CS framework for photoacoustic computed tomography (PACT). During the reconstruction, a small part of the nonzero signals’ locations in the transformed sparse domain is used as partially known support (PKS). PACT reconstructions have been performed with under-sampled in vivo image data of human hands and a rat. Compared to PACT with basic CS, PACT with CS-PKS can recover signals using fewer ultrasonic transducer elements and can improve convergence speed, which may ultimately enable high-speed, low-cost PACT for various biomedical applications.