Home > Academics > Academic Units > Faculty of Engineering and Architecture (FoE&A) > Electrical Engineering > Sanjay Ghosh
Assistant Professor Grade-I
Electrical Engineering
Myresearch interests fall under the joint theme of Image Processing and Brain Sciences. I aim to build a research group with focus on the following themes:
R1. Image Enhancement and Analysis: With my expertise in fast algorithms for classical methods, my current focus is on developing artificial intellegence (AI) methods for image quality enhancement and low-level computer vision. My research focus will be also on translating domain knowledge such as sparsity, structural prior, and rank properties etc towards building interpretable deep neural networks. On the applications side, I will be focussing on the emerging imaging techniques such as burst-images photography, and high dynamic range imaging, and light field imaging.
R2. Machine Learning Driven Brain Disorder Analysis: Over the past few decades, various imaging/signaling modalilies such as MRI, EEG, MEG etc.have become effective tools to efficiently capture both anatomical and functional information about the human brain. My research plan to develop state-of-the-art methods to analyze neuroimaging data for early diagnosis of neurological disorders such as dementia, tinnitus, autism, and depression. In particular, I plan to my machine learning expertise to build interpretable machine learning algorithms to help discover disease-specific biomarkers from imaging data. These research outcomes could directly contribute to precision care of mental health.
R3. Brain Signal Processing: Here I aim to investigate the following three aspects of brain signal processing: (a) reconstruction of functional brain activity measured by neuroimaging (NI) modalities such as magneto-enchaphalography (MEG) and electroencephalography (EEG) and (b) inference and understand ing of functional signals from a perspective of graph signal processing. I intend to intensively explore the capacity of deep networks towards understanding and analyzing brain signals (EEG/MEG/fMRI). In summary, I would work on data-driven computational methods & algorithms for analysis of human brain, behavior, and disorders.
Joint learning of full-structure noise in hierarchical Bayesian regression models by Hashemi A., Cai C. , Gao Y. , Ghosh S. , Müller K. , Nagarajan S. S., Haufe S. IEEE Transactions on Medical Imaging 43 610-624 (2024)
A Joint Subspace Mapping Between Structural and Functional Brain Connectomes by Ghosh S., Raj A. , Nagarajan S. S. NeuroImage 272 119975-119975 (2023)
Bayesian adaptive beamformer for robust electromagnetic brain imaging of correlated sources in high spatial resolution by Cai C., Long Y. , Ghosh S. , Hashemi A. , Gao Y. , Diwakar M. , Haufe S. , Sekihara K. , Wu W. , Nagarajan S. S. IEEE Transactions on Medical Imaging 42 2502-2512 (2023)
Image downscaling via co-occurrence learning by Ghosh S., Garai A. Journal of Visual Communication and Image Representation 91 103766- (2023)
Multi-tasking deep network for tinnitus classification and severity prediction from multimodal structural MR images by Lin C., Ghosh S. , Hinkley L. B., Dale C. L., Souza A. C., Sabes J. H., Hess C. P., Adams M. E., Cheung S. W., Nagarajan S. S. Journal of Neural Engineering 20 016017-016017 (2023)
Fast scale-adaptive bilateral texture smoothing by Ghosh S., Gavaskar R. G., Panda D. , Chaudhury K. N. IEEE Transactionson Circuits and Systems for Video Technology 30 2015-2026 (2020)
Optimized Fourier bilateral filtering by Ghosh S., Nair P. R., Chaudhury K. N. IEEE Signal Processing Letters 25 1555-1559 (2018)
Artifact reduction for separable nonlocal means by Ghosh S., Chaudhury K. N. Journal of Electronic Imaging 26 063012- (2017)
Pruned non-local means by Ghosh S., Mandal A. K., Chaudhury K. N. IET Image Processing 11 317-323 (2017)
On fast bilateral filtering using Fourier kernels by Ghosh S., Chaudhury K. N. IEEE Signal Processing Letters 23 570-573 (2016)
Suvrojit Mitra
Area of Research: Machine learning methods for efficient image understanding
