Posts by Collection

Fully Automated detection of Globes for Volume Quantification in CT Orbits Images Using Deep Learning

Published in Am J Neuroradiol, 2009

Deep learning based segmentation for orbits in CT images.

Recommended citation: Umapathy L, Winegar B, MacKinnon L, Hill M, Altbach MI, Miller JM, Bilgin A. Fully Automated Segmentation of Globes for Volume Quantification in CT Images of Orbits using Deep Learning. AJNR Am J Neuroradiol. 2020 Jun;41(6):1061-1069. doi: 10.3174/ajnr.A6538. Epub 2020 May 21. PMID: 32439637; PMCID: PMC7342761.
Download Paper

A Stacked Generalization of 3D Orthogonal Deep Learning Convolutional Neural Networks for Improved Detection of White Matter Hyperintensities in 3D FLAIR Images

Published in Am J Neuroradiol, 2021

Stacked Generalization for WMH segmentation in T2-FLAIR images.

Recommended citation: Umapathy L, Perez-Carrillo GG, Keerthivasan MB, Rosado-Toro JA, Altbach MI, Winegar B, Weinkauf C, Bilgin A; Alzheimer’s Disease Neuroimaging Initiative. A Stacked Generalization of 3D Orthogonal Deep Learning Convolutional Neural Networks for Improved Detection of White Matter Hyperintensities in 3D FLAIR Images. AJNR Am J Neuroradiol. 2021 Apr;42(4):639-647. doi: 10.3174/ajnr.A6970. Epub 2021 Feb 11. PMID: 33574101; PMCID: PMC8040994.
Download Paper

Convolutional Neural Network Based Frameworks for Fast Automatic Segmentation of Thalamic Nuclei from Native and Synthesized Contrast Structural MRI

Published in Neuroinformatics, 2021

Representation learning approach to learn synthesis of WMn-MPRAGE images for improved thalamic nuclei segmentation

Recommended citation: Umapathy L, Keerthivasan MB, Zahr NM, Bilgin A, Saranathan M. Convolutional Neural Network Based Frameworks for Fast Automatic Segmentation of Thalamic Nuclei from Native and Synthesized Contrast Structural MRI. Neuroinformatics. 2022 Jul;20(3):651-664. doi: 10.1007/s12021-021-09544-5. Epub 2021 Oct 9. PMID: 34626333; PMCID: PMC8993941.
Download Paper

Reducing annotation burden in MR: A novel MR-contrast guided contrastive learning approach for image segmentation

Published in Med Physics, 2023

Multi-contrast contrastive learning for MR segmentation

Recommended citation: Umapathy L, Brown T, Mushtaq R, Greenhill M, Lu J, Martin D, Altbach M, Bilgin A. Reducing annotation burden in MR: A novel MR-contrast guided contrastive learning approach for image segmentation. Med Phys. 2024 Apr;51(4):2707-2720. doi: 10.1002/mp.16820. Epub 2023 Nov 13. PMID: 37956263; PMCID: PMC10994772.
Download Paper

Prostate Cancer Risk Stratification and Scan Tailoring Using Deep Learning on Abbreviated Prostate MRI

Published in JMRI, 2025

Intelligent scanning for prostate MRI

Recommended citation: Johnson, P.M., Dutt, T., Ginocchio, L.A., Saimbhi, A.S., Umapathy, L., Block, K.T., Sodickson, D.K., Chopra, S., Tong, A. and Chandarana, H. (2025), Prostate Cancer Risk Stratification and Scan Tailoring Using Deep Learning on Abbreviated Prostate MRI. J Magn Reson Imaging. https://doi.org/10.1002/jmri.29798
Download Paper

Leveraging Representation Learning for Biparametric Prostate MRI to Disambiguate PIRADS3 and Improve Biopsy Decision Strategies

Published in Investigative Radiology, 2025

Representation learning for prostate MRI

Recommended citation: Umapathy, Lavanya; Johnson, Patricia M; Dutt, Tarun; Tong, Angela; Chopra, Sumit; Sodickson, Daniel K; Chandarana, Hersh. Leveraging Representation Learning for Biparametric Prostate MRI to Disambiguate PIRADS3 and Improve Biopsy Decision Strategies. Investigative Radiology. 2025.
Download Paper

Automated image interpretation and analysis

• Stacked generalization ensemble of 3D orthogonal deep learning models for white matter hyperintensity segmentation in 3D T2-FLAIR images. Paper
• 2D and 3D deep learning models for multi-organ segmentation in body MRI arXiv
• Detection of subtle globe injuries in CT Orbits images with automated segmentation of globes and volume assessment Paper
• Liver fibrosis staging with texture analysis in delayed-phase Gadolinium-enhanced T1-weighted MRI of the liver. ISMRM Proc

Transformers for learning temporal context

• Learning spatio-temporal characteristics in multi-contrast MR images with self-supervision: An application to accelerating quantitative T2 mapping. ISMRM Proc

Representation Learning for imaging

• Leverage representation learning approaches to model a radiologists assessment of prostate MR images from data in which radiologists are confident in their assessment of risk for clinically significant prostate cancer, and use it to disambiguate the equivocal PI-RADS 3 assessments, and avoid unnecessary biopsies in general. Paper
• Leverage underlying tissue-specific information in multi-contrast MR images to learn representations of local regions in an image such that regions belonging to similar tissue types generate similar representations. This constained contrastive learning approach is used to pretrain deep learning models to reduce the amount of labeled data required for downstream segmentation tasks in MRI. Paper
• Leverage shared information in multi-contrast MR images to synthesize Gd-enhanced contrast MR images from non-contrast enhanced MR images that are routinely acquired in the clinic (T1-weighted, T2-weighted, FLAIR). ISMRM Proc
• Synthesize white matter nulled MPRAGE images from conventional T1-weighted MPRAGE images to improve thalamic nuclei segmentation. Paper

AI-driven longitudinal health monitoring

• Develop AI-based longitudinal and multi-modal risk refinement approaches, modeled on human physicians, that can accomodate variable and limited prior context into risk assessment frameworks. arXiv
• Develop foundation models to learn generalized representations from multi-modal (imaging + clinical variables) medical data
• Self-supervised representation learning approach to identify increases in risk over time for longitudinal imaging ISMRM Proc