Software

SeqSeg

Deep learning framework for automated vascular segmentation and model construction from CT and MR images.

Try it Code Paper Guide

MeshGrow

Unified pipeline for building simulation-ready cardiac and vascular meshes from CT and MR images.

Code Paper Explainer

MIROS

Medical image to patient-specific reduced-order hemodynamic simulation in minutes, integrating SeqSeg with SimVascular's 1D and 0D solvers.

Code

SimVascular

Open-source platform for patient-specific blood flow simulation. I contribute software, documentation and user support.

Website Code

Vascular Model Repository

A public resource for sharing high-quality vascular data and models, with standardized metadata and quality control.

Website Design notes

Extensions and utilities

• SlicerSeqSeg 3D Slicer extension for seed-based vascular segmentation using SeqSeg and nnU-Net
• SlicerLinFloNet 3D Slicer extension for simulation-ready whole-heart meshes and segmentations using LinFlo-Net
• LinFlo-Net Deep learning package for diffeomorphic whole-heart mesh generation
• vascular-segment-sampler Blood vessel image data processing for vascular segmentation and analysis

Research projects

Automatic Vessel Tracking

using machine learning to algorithmically track blood vessels without manual input

SeqSeg - Patient-Specific Vascular Geometric Modeling

a machine learning approach to segment CT and MR images

Object Detection for Blood Vessels

Utilizing deep object detection techniques for blood vessel detection in medical image data

reduced-order blood flow modeling

improving ROM parameters and automatizing medical image -> model pipeline

Transformers for Tracking

Utilizing transformer neural network architecture to exploit the sequential nature of blood vessel tracking

MeshGrow - Combining Vascular and Cardiac Models into One

combining the models of the heart and the blood vessels into a single simulation-ready model for patient-specific simulations

Deep Learning to Locate Heart in Medical Image Volume

Utilizing deep object detection techniques for heart location prediction in medical image volume

Collaborations

SimVascular

an open-source software for patient-specific blood flow simulations

VMR

an open-access dataset of vascular models with simulation results

hemodynamic simulations of aneurysms modeled using machine learning

validating the use of machine learning for segmenting abdominal aortic aneurysms from CT and MR images for hemodynamic simulations

Bioprinting; Modeling Cell Survival

Modeling chemical concentration via diffusion equation to inform bioprinting template design

Continuous blood pressure analysis

understanding blood pressure waveforms from wearable devices

Coursework archive

Effect of Modeling Variability on Aneurysm Analysis

Final Project for MEC213 - Biofluid Mechanics

Learning Blood Vessel Centerline from Medical Images

Final Project for CS280 - Computer Vision

Leveraging Transformers for Blood Vessel Tracing

Final Project for CS282A - Designing and Understanding Deep Neural Networks

Computational Parallelism to Boost Blood Vessel Tracking

Final Project for CS267 - Applications of Parallel Computers

Modeling Effect of Bad Working Posture on Vertebral Health

Final Project for ME210 - Orthopaedic Biomechanics