Anirudha Shastri MSK · New York

I build computational methods for seeing, understanding, and intervening in disease.

Medical imaging researcher and engineer developing methods across segmentation, registration, uncertainty, ablation assessment, and clinically grounded AI.

View researchGoogle Scholar ↗

Anirudha ShastriMedical AI researcher

Medical AIMedical ImagingImage-Guided InterventionTranslational Research

01 / Selected research

Methods built around the clinical problem.

Investigations across interactive segmentation, deformable registration, and diagnostic imaging.

Diffusion models3D segmentationUncertainty

Interactive Lung Tumor Segmentation using Denoising Diffusion Models

Human-assisted diffusion framework using a single user-defined point, compact 3D ROI processing, and uncertainty-aware predictions.

0.773D mean Dice

20DDIM steps

49.7×sampling speedup

View on SPIE

From interaction to uncertainty

→→→→

Step 1: User click — a single point marked near the tumor center on the CT slice.

A clinician places a single point inside the suspected tumor on the CT slice — the only manual input the method needs.

Deformable registrationFoundation modelsLung ablation

Comparing B-spline and uniGradICON-AI Foundational Registration

Comparison for pre/post lung ablation CT, examining the tradeoff between accuracy and computational speed.

0.43 mmB-spline TRE

1.11 mmuniGradICON TRE

7–10×runtime difference

View on SPIE

Drag to compare · Unregistered / Registered

UnregisteredRegistered

Object detectionRadiographsClinical AI

Automated Detection and Characterization of Bone Lesions and Fractures

Deep-learning framework for pelvic radiographs with emphasis on normal-image sampling, difficult cases, and error analysis.

Drag to compare · Ground Truth / Predictions

Ground TruthPredictions

02 / Publications

Selected scholarly work.

2026

First author

Interactive Lung Tumor Segmentation using Denoising Diffusion Models

Anirudha Shastri, Yanjie Wu, Robert Salkin, Arvind Dev, Julia Wolleb, Florentin Bieder, Etay Ziv, Krishna Nand Keshava Murthy

SPIE Medical Imaging · 2026

DOI

First author

Comparing B-spline and uniGradICON-AI Foundational Deformable Image Registration Methods

Anirudha Shastri, Basar Demir, Robert Salkin, Arvind Dev, Marc Niethammer, Etay Ziv, Krishna Nand Keshava Murthy

SPIE Medical Imaging · 2026

DOI

First author

AI-Based Automated Detection and Characterization of Bone Lesions and Fractures on Pelvic Radiographs

Anirudha Shastri et al.

RSNA · 2025

Co-authored conference contributions

AI-Guided Ablation Assessment Framework for Clinical Workflows Implemented as a 3D Slicer ExtensionSociety of Interventional Oncology · 2026 · Co-author

Spatial Analysis of Ablation Margins and Tumor Recurrence in Microwave Lung AblationSociety of Interventional Oncology · 2026 · Co-author

03 / Experience

Research in translation.

2025—Present

Memorial Sloan Kettering Cancer Center

Senior Research Technician · Interventional Radiology · New York

Translational AI research across deformable registration, diffusion segmentation, ablation assessment, lesion and fracture detection, medical imaging pipelines, and optical spectroscopy.

Earlier

Healthcare AI & engineering

Applied machine learning and research engineering across clinical data, computer vision, and healthcare workflows.

04 / Patent

Automated Cervical Vertebral Staging to determine skeletal maturation using Lateral Cephalograms

Indian Patent Application No. 202541002362

Computer-vision based cervical vertebral staging using deep-learning methods and landmark localization.

True vs. predicted cervical vertebra landmarks on a lateral cephalogram

Landmark Detection for Bone Age Prediction U-Net and InceptionResNetV2 models implementing the staging method above. GitHub ↗

05 / About

“I like problems where computation has to survive contact with reality.”

My work sits between machine learning and clinical workflows, where model performance is only part of the problem. Registration error, image geometry, annotation quality, uncertainty, inference time, and clinical usability all matter.

Outside the lab, I’m usually exploring tools, cameras, cycling routes, 3D printing, electronic repair or the infrastructure that makes technical work possible.

Anirudha Shastri with his road bicycle — Beyond the labCycling · building · exploring

Side projects

Dental Assistance Tool web app interface

Dental Assistance ToolA web app giving dentists AI-assisted analysis of dental images, built with Node.js and a Flask inference server · GitHub ↗

Coming later · /labExperiments / Lab

Interactive demonstrations for registration, uncertainty, ablation margins, and spectroscopy.

→