My Work Experience

My Projects

【Medical Image Processing and Analysis】

Aorta Segmentation, Visualization and Calculation
-preprocess Phase Contrast MRI data, including bias field correction and flow field denoising
-segment aorta in subpixel precision and retrieve flow field from Phase Contrast MRI
-retrieve aortic centerline and perpendicular plane
-generate 3D mesh from point cloud of aortic surface
-make a moving picture of blood flow
-calculate aortic Wall Shear Stress(WSS) from PC-MRI, and compare it with the result from computational fluid simulation

Coronary Artery Segmentation, 3D Reconstruction and Computation ` -preprocess CTA data and apply vesselness filters
-segment coronary arteries
-retrieve centerlines, calculate its arclength to inflow point and get radius of perpendicular section plane
-coronary artery centerline refinement -generate 3D mesh from point cloud of coronary artery surface
-display coronary arteries using MPR -compare Fractional Flow Reserve by CFD with the one measured with PressureWire
-detect suspicious stenosis

3D Medical Imaging Integrated Software Development
-add/modify segmentation algorithms (Graph Cut, Fast Marching, Live Wire and etc)and integrate them into the software
-integrate live cancer reporting and diagnostic system (LiRADS)
-integrate trained liver cancer LiRADS features classifier

Evaluation on Effectiveness of Liver Cancer Drug
-segment liver tumor in a time series
-compute drug permeation parameters (Ktrans, ve and etc)
-evaluate effectiveness through statistics

Aorta CT and MRI Registration
-segment aortas of same patient from different sources (CT, MRI)
-retrieve surface point cloud
-use revised Iterative Closest Point method and RANSAC affine transform for registeration

Brain Tumor Detection using Asymmetry
-find 3 major axis through PCA, fine tune using a homogeneous transform and last get symmtery plane
-apply SLICO superpixel algorithm on image slices
-compute the texture difference of corresponding superpixels on each side of symmetry plane
-estimate tumor position using texture difference measures

Liver Vessels Segmentation and Couinaud Hepatic Classification (unfinished)
-retrieve hepatic veins
-find centerline, arclength, curvature, radius of perpendicular plane and other geometries
-locate bifurcation points -conduct Couinaud Hepatic Classification according to previous geometries and bifurcations

【Shape Deformation】

Geodesic Shooting Method in Image Registration
-set up geodesic shooting equations for Large Diffeomorphic Deformation Metric Mapping that is used for registering source image to target image
-using various kernels/filters, to compare the results
-calculate transform field and get transformed image
-compute geodesic distance between transformed image and target image

【General Image Processing】

Beef Marbling Grading
-segment beef ribeye
-utilize superpixel algorithm to obtain blocks, then distinguish fat and lean blocks
-calculate far portion, fat blocks portion, fractal dimension and other texture features
-set up a marbling ranking classifier using these features

【Teaching Experience】

Teaching Assistant
-I have been instructing students in problem sessions at University of Texas at Dallas
-classes include Calculus(1,2,3), Linear Algebra, Ordinary Differential Equation, Calculus(Cohort), Multivariable Calculus and ect
-my average rating from students is 4.4 out of 5