The Medical Image Analysis Lab (MIAL) was established by Dr. Ghassan Hamarneh (Computing Science) and Dr. Faisal Beg (Engineering Science) in September 2003.
Abnormalities of brain structures have been shown to be a key biologic characteristic of neuropsyciatric diseases such as Schizophrenia, Alzheimer's or Multiple Sclerosis. Abnormalities in cardiac structure are also implicated in altered mechanical and electrical properties and changes in proper functioning of the heart. The common theme in biology is that structure and function are interdependent on each other, and in-vivo imaging capabilities such as magnetic resonance imaging (MRI), computed tomography (CT), ultrasound, diffusion-tensor MRI (DTMRI), and functional MRI (fMRI) are now acquiring data that enables this dependence to be observed and quantified. Small animal medical imaging modalities (ultrasound bio-microscopy, micro-CT, optical projection tomography, and MRI microscopy) are now complementing existing physiological and behavioural screenings of transgenic/gene-knockout mice. Paralleling the tremendous increase in ability to image internal anatomical structure and function at increasingly higher resolution and signal-to-noise is the need for sophisticated computational tools for analysis of such images. Such computational tools will extract "information" from anatomical images to quantify normal and abnormal structure, while taking into account the tremendous biological variability inherent in normal and diseased states across the population. This is the emergent field of mathematical and computational representation/analysis of anatomy, also referred to as Computational Anatomy (CA), that forms the main thrust of our medical image analysis laboratory (MIAL) at SFU.
This computational study of anatomy consists of three core areas (1) segmentation or automated labelling of anatomical features such as points, curves, surfaces, volumes (2) registration, or geometric transformation of these anatomical features into common coordinates for comparison, and (3) statistical analysis via probability measures that allow for inferences on normal and diseased states.
Research at MIAL spans the development of techniques for image analysis in each of the above three core areas. Within these three core areas are numerous interrelated projects for developing techniques that are jointly aimed towards the overall theme of relating structural and functional aberrations to clinical disease within a precise and quantitative anatomical framework. Dr. Hamarneh trained at Chalmers University medical image analysis lab (Sweden) and University of Toronto, Visual Modeling lab of Dr. Terzopoulos, founder of the classical deformable models ('Snakes') which revolutionized medical image analysis. Dr. Hamarneh has contributed to the first core area of image segmentation by developing techniques for incorporating prior knowledge through building AI and A-Life models of geometry, physics, statistics, behaviours, and cognition into deformable shape models, leading to artificial deformable organisms. Dr. Beg trained with the Johns Hopkins Group led by Michael Miller who pioneered high dimensional brain mapping (HDBM). Dr. Beg developed the Large Deformation Metric Mapping tool that allows the representation of anatomical structures in a common coordinate system and computes a metric distance that quantifies the notion of close and far based on geodesic length in the space of high dimensional diffeomorphic transformations linking anatomical structures. Dr. Beg has pioneered the application of HDBM to study of functional MRI data for inferences on brain activity and for the study of cardiac geometry. Drs. Hamarneh and Beg's contributions also extend to the study of statistical analysis of shape variation. These three core areas form the basis of ongoing research efforts at MIAL.
- Medical image analysis
- Computational Anatomy
- Image segmentation
- Image registration
- Shape analysis
- Deformable models
- Geometry-, physics-, and statistics-based shape modeling
- Small animal image analysis
- Neuro-degenerative disorders
- Biosignal processing
- Optimization theory
- High-performance computation methods
Some of our current collaborators are: