Image Data emulation and Analysis (idea) Laboratory

 

 

 


 [Research Summary]  [People] [Research Projects] [Publications] [Sponsors] [Software] [Positions] [Links]


 

Research Summary

 

We conduct both theoretical and applied research in the areas of biomedical image analysis, computer vision and computer graphics. A central theme of the research is to analyze and extract information from images and videos for interpretation, quantification, modeling and synthesis.  The current focus of the lab is on developing robust and novel methods for deformable object modeling, segmentation, registration, tracking and recognition.  The goal of our research is to better solve computational problems in biomedical imaging, computer vision, computer graphics, and multi-modal data archiving/searching/ visualization/integration.  See the IDEA Lab’s homepage for details.

 

Research Projects

 

 

 

 

Metamorphs: Deformable Shape and Appearance Models (PDF)

 

*     Deformable models with both boundary and interior

*     Online and adaptive learning of model-interior intensity/texture statistics

*     Consistent interior appearance during deformation

*     Common deformation scheme using Free Form Deformations (FFD) for both boundary and interior

*     Model-based segmentation and tracking

 

       

 

Shape Registration in Implicit Spaces (PDF)

 

*     Implicit shape representation

*     Global registration by maximizing mutual information in implicit spaces

*     Non-rigid registration by minimizing a sum-of-squared differences criterion w.r.t. Free form Deformation (FFD)  parameters

*     2D/3D shape registration and tracking

*     Establishing accurate correspondences for statistical shape modeling

*     Fitting mesh models to dense point clouds

    

          

Image and Video Registration (PDF-1, PDF-2, PDF-3)

 

*     Salient region feature based image registration

*     Pruning false matches by geometric constraints

*     Image prior-based dynamic texture registration to recover camera motion

*     Simultaneous image transformation and sparse representation recovery for recognition and video registration

 

 

   

Heart Modeling and Wall  Motion Analysis based on Information from Tagged MR, MR, and CT Images (PDF-1, PDF-2, PDF-3)

 

*     Segmentation of heart contours (e.g. epicardium, left/right ventricle endocardium, papillary muscle)

*     Creating heart mesh model based on image information

*     Model-based tracking of heart wall motion

*     Cardiac muscle mechanistic analysis, disease classification, computer-aided diagnosis

 

      

                   

 

Whole-body Image Interpretation in PET/CT (PDF-1, PDF-2, HTML)

 

*     One-click hot spot segmentation in PET

*     Organ (e.g. kidney) detection using boosted 3D Haar wavelet features

*     Robust click-point linking in longitudinal studies

*     Hot spot detection, segmentation and classification

*     Multi-modal image information integration applied to robust segmentation, registration and change quantification

 

 

    

 

Texture (PDF-1, PDF-2)

 

*     Textured object segmentation

*     Texon natural scale computation and appearance representation

*     Texture edges

*      Dynamic textures

 

         

 

 

Computer Graphics Research (PDF-1, PDF-2)

 

*     Physically-based and data-driven simulation of non-rigid deformable bodies

*     Biophysical modeling and simulation of biological processes

*     Visualization

*     Shadow, Transparency

*     Illumination

 

 

 

 

Structuring, Reasoning, Querying, and Retrieval in Large Medical Image Archives (URL, PDF-1, PDF-2, PDF-3)

 

*     Cervigram and Colposcopy image database for cervical cancer research

*     Segmentation of biomarker regions in color cervigram images (URL)

*     Evaluation of multiple-observer segmentations

*     Interactive segmentation and image indexing

 

 

Novel Methods and Applications for Chemical Image Analysis

 

*     Quantitative fluid-and-foam-mixture image analysis to extract parameters for chemical characterization

*     Bubble texture segmentation

 

 

Computational Analysis of Microscopic Cell Images

 

*     High-resolution 3D tomographic cell images

*     Live-cell time lapse movies

*     De-convolution for super-resolution

*     Multi-modal optical imaging

 

 

Machine Learning and Statistical Object Models (PDF-1, PDF-2)

 

*     Active shape and appearance models

*     Supervised learning, unsupervised learning

*     Nonlinear models and manifolds

*     Learning from labeled and unlabeled data

*     Biomedical applications

 

 

Oncology Applications, Molecular Image Analysis, Computer-aided Diagnosis

 

*     Intensity moduldated radiation therapy (PDF)

*     3D Tumor Shape and Location Reconstruction from 2D Bioluminescence Images (PDF)

*     Cancer detection, segmentation, change quantification

*     Multi-modal image registration and information integration

 

 

 

People

 

Professor

 

Xiaolei Huang

 

Ph.D. Students

 

Wei Wang

Yaoyao Zhu, co-advising w/ Prof. Dan Lopresti

Tian Shen

Hongsheng Li

Edward Kim

 

M.S. and Undergraduate Students

 

            Mayura Warge 

 

Former Students

 

Abraham Jun (Sophomore, summer 08)

Christopher Hamilton (Senior, fall 07 & spring 08)

Samuel Wechsler (Senior, fall 07 & spring 08)

Yu Tian (M.S., spring 08)

Yusuf Artan (M.S., summer & fall 07)

Nick Moukhine (M.S., fall 07)

Thomas Boinot (M.S., summer & fall 07)

Hisham Abu-Nabaa (Ph.D., summer 07)

Doug Paul (Senior, spring 07)

 

 

Sponsors

 

Research projects in the IDEA lab are sponsored by:

 

Christian R. & Mary F. Lindback Foundation

 

 



 

                            

 

 

Research Opportunities

           

If you are a Lehigh or a prospective student who would like to gain first-hand experience with research on biomedical imaging, computer vision, or computer graphics, please send an email to xih206@lehigh.edu and inquire regarding current research opportunities.  Interested Lehigh students are welcome to work in our lab during academic year or during summer time for either independent studies or a stipend.

                                

Publications

 

Software

           

Links and Resources