Experience & Education
-
Carnegie Mellon University
View Akshay’s full experience
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Licenses & Certifications
Publications
-
LLMs Accelerate Annotation for Medical Information Extraction
Proceedings of Machine Learning Research
See publication1️⃣ Our Large Language Model (LLM) pipeline achieves F1 scores comparable to those of average medical annotation specialists, with higher recall but lower precision.
2️⃣ Crucially, incorporating a human-in-the-loop approach allows for expert-level labeling, equivalent to top-performing annotation specialists, while saving 58% of human time. -
Deployed Deep Learning Kidney Segmentation for Polycystic Kidney Disease MRI
Radiology: Artificial Intelligence
See publication➤➤➤ Python Code and Model Weights: https://github.com/aksg87/adpkd-segmentation-pytorch
1️⃣ Model performance on validation MRIs (53 prospective, 20 external institution) demonstrated a
Dice similarity coefficient > 0.97 (quartile Q1 > 0.94) and Bland-Altman mean % difference in
Total Kidney Volume (model versus manual reference) < 3.6% (95% confidence interval: 2.0%,
5.2%).
2️⃣ Model deployment into the clinical pipeline was accomplished by pushing new MRI…➤➤➤ Python Code and Model Weights: https://github.com/aksg87/adpkd-segmentation-pytorch
1️⃣ Model performance on validation MRIs (53 prospective, 20 external institution) demonstrated a
Dice similarity coefficient > 0.97 (quartile Q1 > 0.94) and Bland-Altman mean % difference in
Total Kidney Volume (model versus manual reference) < 3.6% (95% confidence interval: 2.0%,
5.2%).
2️⃣ Model deployment into the clinical pipeline was accomplished by pushing new MRI scans to a
clinical server within the PACS firewall which runs inference on axial T2-weighted sequences
and outputs results formatted for radiologist label refinement and final reporting.
3️⃣ Prospective assessment on 53 patients showed a 51% reduction in radiologist time for model-assisted segmentation compared with segmenting without the model (P < .001). -
Breast MRI screening for average‐risk women: A monte carlo simulation cost–benefit analysis
Journal of Magnetic Resonance Imaging
-
Ethnic Difference in Proximal Aortic Stiffness: An Observation From the Dallas Heart Study
JACC: Cardiovascular Imaging
See publicationEstablished significant ethnic differences in proximal aortic stiffness independent of blood pressure and relevant risk factors. Study was conducted in a multi-ethnic population-based-sample (Dallas Heart Study).
-
Fully automated tool to identify the aorta and compute flow using phase-contrast MRI: Validation and application in a large population based study
Journal of Magnetic Resonance Imaging
See publicationDeveloped a robust fully automated tool to localize the aorta and provide flow volume measurements on phase contrast MRI was validated on a large population‐based study (Dallas Heart Study). The resulting data from this tool served as the basis for additional high impact publications.
Other similar profiles
-
Francisco Ocegueda
Connect -
Lane McIntosh
Connect -
Micael Carvalho
Connect -
Yi Liu
Connect -
François Modave, PhD.
Professor of Artificial Intelligence, Digital Health, Dept. of Pediatrics, Center for Remote Health Monitoring
Connect -
Manish Gupta
Connect -
Andrea Poretti
Research Associate Division of Pediatric Radiology, The Johns Hopkins Hospital
Connect -
Rene Angeles
Connect -
Rory Lewis JD PhD
Connect -
Simas Glinskis
Artificial Intelligence Scientist. Quantum Physics PhD.
Connect
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content