Perelman School of Medicine at the University of Pennsylvania

Section for Biomedical Image Analysis (SBIA)

Advanced Image Computing and Analytics Core (AICAC)


AICAC was formed in July 2007 by the Section of Biomedical Image Analysis, Department of Radiology, University of Pennsylvania, in an effort to facilitate translational research that needs advanced image processing and analysis. As the size and needs of imaging studies have grown exponentially in the past decade, the need for highly automated and quantitative tools for image analysis has also grown. Although many sophisticated software tools have been developed by image computing laboratories, they don't regularly reach the clinical researchers, in part because of the absence of translational research bridges and in part because many such tools are not easy for the typical clinical researcher to use. The goal of AICAC has been to facilitate this flow of high-tech image analysis from computational labs to clinical researchers.

Current Projects

A variety of projects are currently served via AICAC, which include imaging in their aim to study:

-- Normal Aging, Alzheimer's and Parkinson's diseases

-- Brain development

-- Schizophrenia and bipolar disorder

-- Effects of diabetes on the brain

-- Cerebrovascular disease

-- Effects on alcohol on the brain

-- Hormonal therapy and its effects on the brain

-- Brain changes during mid-age

-- Epidemiologic studies including brain imaging and investigating relationships between race and socioeconomic factors, and brain health

-- Autism

-- Brain tumors

-- Breast tumors

-- Depression

-- Chronic kidney disease

Close


Image Processing Pipelines

Currently, AICAC provides various standardized image processing pipelines, as well as consultation and development of customized pipelines that meet more specialized needs:

Standardized pipelines:

 — Image preprocessing

Tasks such as image reorientation, automated and semi-manual segmentation (e.g. skull stripping), field inhomogeneity correction, motion correction, and other image preparation steps.

— Brain MRI parcelation and ROI measurement

Atlas warping and automated definition of ROis in brain images. Multi-atlas labeling using hierarchical parcelation of brain MRI into 262 ROIs hierarchically organized from whole brain ROIs to individual cortical gyri.

— Regional volumetric measurements of brain ROIs, and measurement of co-registered structural (e.g. sMRI, DTI), physiological (bold fMRI, ASL) and molecular images (e.g. amyloid images).

— A description of the anatomical ROIs and the hierarchy: Example
— A 3D rendering of a parcelation : Example

— Voxel-based analysis of brain images, utilizing high-dimensional image warping techniques.

— Fiber tractography and basic connectivity analyses.

— Measurement of brain and breast tumor size and progression.

— High-dimensional pattern analysis for classification of MRI scans in Alzheimer's Disease and other dementias; SPARE-AD scores from sMRI of the brain.

Customized/Specialized pipelines:

Needs must be discussed with the directors of the core to determine whether they can be addressed by AICAC's expertise and time constraints, and if they can, a pipeline is developed and applied.

Close


Personnel

Overseeing Director: Christos Davatzikos

Core Manager and Technical Director: Guray Erus

Staff:

  • Harsha Battapady
  • Xiao Da
  • Drew Parker
  • Jimit Doshi
  • Michael Hsieh
  • Evi Parmpi (financial administration)
  • Mark Bergman (IT director)

Close


Fees and Project Proposals

A short description of the project and its image analysis needs should be sent to Evi Parmpi. AICAC will respond with an estimate of the cost, if the project can be assisted by AICAC's facilities.

Close