Inverse problems, deep learning, and computational methods for medical imaging
Our research sits at the intersection of applied mathematics, signal processing, and deep learning, with a focus on developing computational methods that push the boundaries of medical imaging quality and accessibility. We work on inverse problems where the goal is to recover high-fidelity images from limited, noisy, or otherwise degraded measurements.
A core theme across our work is making advanced imaging more practical: reducing the data and hardware requirements, improving reconstruction speed, and enabling new clinical applications in medical imaging with a focus on ultrasound, photoacoustic, and cardiovascular imaging.
Developing deep learning-based reconstruction methods for photoacoustic tomography under limited-data and noisy conditions. Work includes sinogram super-resolution, dimensionality reduction with plug-and-play priors, denoising diffusion models, and image quality assessment.
Deep learning for echocardiographic image analysis including left ventricle segmentation (LVNet), bandwidth enhancement, beamforming, and automated segmentation of aortic aneurysms in 3D time-resolved ultrasound.
Designing efficient deep learning architectures for multi-organ and multi-modal segmentation, including brain tumor segmentation (MTAU), breast ultrasound, and cardiac structure delineation using transformers.
Investigating robustness and security of federated learning in medical settings, including minimal data poisoning attacks on distributed medical image classification systems.
Deep learning methods for ECG classification, cardiac rhythm analysis, respiratory distress quantification, and smartphone-based estimation of physiological parameters for remote monitoring.
Image-guided filtering for histopathology microscopy, whole slide imaging augmented by computational methods, and tumor cellularity assessment in breast cancer histology images.
Indian Institute of Science, Bangalore
Iowa State University
Eindhoven University of Technology / UMC Utrecht
Eindhoven University of Technology