A5094124285- Radio Astronomy Observations and Technology
- Medical Imaging Techniques and Applications
- Astrophysics and Cosmic Phenomena
- Advanced MRI Techniques and Applications
- Optical measurement and interference techniques
- Statistical and numerical algorithms
- Computational Physics and Python Applications
- Radiomics and Machine Learning in Medical Imaging
- Geophysics and Gravity Measurements
- Pulsars and Gravitational Waves Research
Heriot-Watt University
2024
Abstract Radio-interferometric imaging entails solving high-resolution high-dynamic-range inverse problems from large data volumes. Recent image reconstruction techniques grounded in optimization theory have demonstrated remarkable capability for precision, well beyond CLEAN’s capability. These range advanced proximal algorithms propelled by handcrafted regularization operators, such as the SARA family, to hybrid plug-and-play (PnP) learned denoisers, AIRI. Optimization and PnP structures...
Abstract A novel deep-learning paradigm for synthesis imaging by radio interferometry in astronomy was recently proposed, dubbed “Residual-to-Residual DNN series high-Dynamic range imaging” (R2D2). In this work, we start shedding light on R2D2's algorithmic structure, interpreting it as a learned version of CLEAN with minor cycles substituted deep neural network (DNN) whose training is iteration-specific. We then proceed first demonstration real data, monochromatic intensity the galaxy...
Radio-interferometric (RI) imaging entails solving high-resolution high-dynamic range inverse problems from large data volumes. Recent image reconstruction techniques grounded in optimization theory have demonstrated remarkable capability for precision, well beyond CLEAN's capability. These advanced proximal algorithms propelled by handcrafted regularization operators, such as the SARA family, to hybrid plug-and-play (PnP) learned denoisers, AIRI. Optimization and PnP structures are however...
We propose a new approach for non-Cartesian magnetic resonance image reconstruction. While unrolled architectures provide robustness via data-consistency layers, embedding measurement operators in Deep Neural Network (DNN) can become impractical at large scale. Alternative Plug-and-Play (PnP) approaches, where the denoising DNNs are blind to setting, not affected by this limitation and have also proven effective, but their highly iterative nature affects scalability. To address scalability...
The ``Residual-to-Residual DNN series for high-Dynamic range imaging'' (R2D2) approach was recently introduced Radio-Interferometric (RI) imaging in astronomy. R2D2's reconstruction is formed as a of residual images, iteratively estimated outputs Deep Neural Networks (DNNs) taking the previous iteration's image estimate and associated data inputs. In this work, we investigate robustness R2D2 estimation process, by studying uncertainty with its learned models. Adopting an ensemble averaging...