MICCAI2005

MICCAI 2005
8th International Conference on Medical Image
Computing and Computer Assisted Intervention

Tutorials and Workshops

Tutorial Submission is closed

Tutorials will be held on October 26 and will complement and enhance the scientific program of MICCAI 2005. The purpose is to provide educational material for training new professionals (students, clinicians, new researchers in the field). They will also provide discussions of technical and application issues. Tutorial chair: Larry Staib (staib at miccai2005.org).

Tutorial Program (click on title of tutorial for a description of the content):

Morning	Afternoon
Computer Integrated Surgery Organizer: Gregory Hager	2-D and 3-D Image Registration and Fusion Organizer: Ardy Goshtasby
Statistics of Anatomic Geometry Organizer: Stephen Pizer	Cardiovascular MR: acquisition and post-processing Organizer: Boudewijn Lelieveldt

Workshops

Title		Date	Paper Due Date
ISC / NA-MIC / MICCAI Workshop on Open-Source Software	Workshop Program	October 30, 2005	August 5,2005

Description of Tutorials

Computer Integrated Surgery

Gregory Hager, Gabor Fichtinger, Peter Kazanzides, Russell Taylor (Johns Hopkins)

This tutorial will broadly survey the spectrum of Computer-Integrated Surgical (CIS) systems currently available and under development. Two CIS system paradigms, surgical CAD-CAM and surgical assistance, will be introduced. The general structure of these systems will be discussed, and several specific examples will be presented. In the area of surgical CAD-CAM, a detailed description of several percutaneous image-guidance systems, with particular emphasis on prostate applications will be presented. This will be followed by a discussion of current developments in the area of assistance methods for minimally invasive surgery. Finally, a discussion of system development tools and processes will close the tutorial.

Topics:
   - Overview of CIS systems (Taylor)
        A brief history of the field of CIS systems
        System paradigms for CIS
        Applications potential of CIS systems
        Current technological barriers to CIS

    - Percutaneous therapy systems (Fichtinger)
        CT guided insertions
        MRI compatible robots
        Interventional US

   - Minimally invasive assistance systems (Hager)
        Paradigms for surgical assistance
        Virtual fixtures for guidance and safety boundaries
        Information assistance and sensor substitution
        Sensing and real-time registration technologies

   - Building clinical systems (Kazanzides).
        Review of regulatory requirements
        Risk analysis methods (FMECA)
        Safety system design
        User acceptance (ergonomics and economics)
        Data collection and system monitoring
        Architectures and toolkits

Statistics of Anatomic Geometry

Stephen Pizer (UNC), Ian Dryden (U. of Nottingham), P. Thomas Fletcher (U. of Utah), Sarang Joshi (UNC), Xavier Pennec (INRIA Sophia-Antipolis), Carole Twining (U. of Manchester)

Humans form a population with an anatomy common enough that it can be usefully described probabilistically. An individual human varies in a way across time that can usefully be described probabilistically. This tutorial will cover the methods for producing and using probability distributions on anatomic objects or regions as well as on images on these geometric entities. We will survey the representation of anatomic objects and collections thereof, anatomic spaces, and images on these spaces, as well as diffusion tensors and diffusion tensor images, and spatiotemporal descriptions of the foregoing, together with transformations on these spaces. The geometric models covered will include landmarks and point distribution models, boundary and medial models, both sampled and parametrized, pseudo-distance functions, and atlas + diffeomorphism models, as well as tensors, similarity and affine transformations, and various nonlinear transformations. Intensity models will include appearance and profile models and histograms, as well as diffusion tensors and texture features. Methods of statistical analysis for the formation of probabilities on single-object and multi-object geometric representations and of probabilities of image intensities, textures, or diffusion tensor values conditional on geometry will be presented. These include Fréchet means, principal and independent component analyses, and extensions of principal component analysis to the Riemannian manifolds needed for rotational transformations and tensors, as well as canonical correlation and Mallow distributions (Earth Mover’s distances). Applications in segmentation and registration by posterior optimization or likelihood optimization, in hypothesis testing and discrimination in regard to geometric properties such as volume and shape, and in formation of didactic atlases with variation will be sketched. The mathematical background assumed for this tutorial will be limited to multivariable calculus, and the background in probability and statistics will be limited to a university-level course in probability densities of multiple random variables.

2-D and 3-D Image Registration and Fusion

Ardy Goshtasby (Wright State), Colin Studholme (UCSF), Bill Crum, Graeme Penney (University College London)

This half-day tutorial covers the fundamentals of image registration and fusion for visualization and image analysis. Methods for the registration of mono-modality and multi-modality images are reviewed, and methods for the fusion of registered mono- and multi-modality images are discussed. Steps in image registration includeing feature selection, feature correspondence, transformation functions, and image resampling are covered in detail, and methods for evaluation of image registration methods are described. Programs to demonstrate both rigid and nonrigid image registration and programs showing fusion of mono- and multi-modality images are demonstrated.

Topics:

Introduction:

What is image registration?
What is image fusion?
Applications of image registration and fusion

Feature Selection:

Point landmarks
Contours
Regions
Template

Landmark Correspondence :

Using coherence property
Using clustering
Using image invariance

Template Matching:

Similarity measures: cross-correlation, mutual information
Search techniques

Transformation Functions:

Rigid
Similarity

Nonrigid: thin-plate splines, multiquadrics, rational Gaussians, weighted mean

Image Fusion:

Image fusion through image blending
Fusion of mono-modality images
Fusion of multi-modality images

Validation Methods

For feature selection
For feature correspondence
For image transformation
Accuracy, reliability, and robustness of image registration software

Summary and Conclusions

Future work in image registration
Future work in image fusion

Cardiovascular MR: acquisition and post-processing

Boudewijn Lelieveldt (Leiden University Medical Center), Derek Hill (King's College), Alejandro Frangi (Pompeu Fabra University), Daniel Rueckert (Imperial College London)

Topics:

MR acquisition (Hill) MR modalites for the heart:

Anatomy, function, flow, perfusion, late enhancement, tagging. The challenge of fast imaging with MRI
Steady State Free Precession contrast
Gating vs real time
The respiratory motion problem Undersampling techniques
Partially parallel imaging eg: SENSE
KT undersampling eg: sliding window, KT-BLAST, RTR Future challenges
Dynamic volume imaging
Free breathing acquisitions
Automated planning
Competing with CT

Vascular image analysis (Frangi)

Vascular image segmentation
Vascular image registration
Vascular characterisation
Centerline extraction and navigation
Image-based computational fluid dynamics simulation

Registration based approaches (Rueckert)

Atlas-based segmentation
Registration of deformable and biomechanical models
Motion and deformation analysis
Motion correction
Perfusion
Multimodality fusion

Misc. 3D cardiac modeling approaches (Lelieveldt)

3D Active Shape Models
2D + time, 3D, 3D + time
Active Appearance Models
Biomechanical modeling
Analytical template modeling
Medial cardial modeling

Web-administration: Martin Styner (styner at miccai2005.org)