Compressive Sensing

Compressive sensing (CS) aims to reconstruct high-dimensional signals from significantly fewer measurements than traditionally required, leveraging signal sparsity or low-rank structure. Current research focuses on developing efficient algorithms and neural network architectures, such as deep unfolding networks and deep equilibrium models, to improve reconstruction accuracy and speed, often incorporating adaptive sampling strategies and incorporating prior knowledge like physics-based models or non-local image priors. CS has broad applications across various imaging modalities (e.g., photoacoustic, MRI, tomography) and machine learning (e.g., federated learning), offering significant potential for reducing data acquisition time, storage needs, and computational costs while maintaining image quality.