Workshop Organizer: Jasjit S. Suri, PhD., Fellow AIMBE, Fellow ABI, SrMIEEE, MAAPM

Speakers:
Alex Sarti, PhD., University of Balogna, Balogna, Italy.
Karol Mikula, PhD., Slovak University of Technology, Bratislava.
Claudio Lamberti , PhD., University of Balogna, Balogna, Italy. 
Aly A. Farag,PhD., University of Louisville, KY, USA.

Abstract

Medical Image Analysis has recently being dominated by topology driven segmentation and registration techniques. Since the human anatomy differs from case-to-case, the image data acquired using medical imaging modalities has different topology, gray scale and intensity variations and the physics of different modalities. Segmentation of anatomic structures thus needs topology driven strategies and recently level sets and PDE's have started to show very promising results. This workshop covers the basics of level sets and PDE-based strategies for Medical Image Analysis and advanced applications of level sets. The workshop will be both for beginners and advance image processing research scientists, clinicians, radiologists, interventional radiologists, biomedical engineers, computer science graduates and electrical engineers.

The fundamental level sets for the beginners will cover Eikonal Equations, Solutions to Eikonal Equations such as Fast Marching Strategies, narrow banding and stable solutions. Advanced level sets will cover fast region-based level set approach for extraction of white matter, gray matter, and cerebrospinal fluid boundaries from two dimensional magnetic resonance slices of the human brain, where the speed control functions will be based on region, edge and curvature. This will include the integration of regional statistical techniques with boundary based geometric level sets. We will also cover examples where the flow is derived as the steepest descent of an energy functional taking into account the density probability distribution of the gray levels of the image as well as smoothness constraints. A finite difference approximation of the flow will be derived and numerical experiments will be provided. Results will be presented on ultrasound medical images as fetal echography and echo-cardiography, as well as images with Poisson and Gauss distributions. semi-implicit co-volume level-set method in 3D image segmentation.

For very advanced applications, the workshop will present three-dimensional semi-implicit complementary volume level-set scheme for solving the Riemannian mean curvature flow of graphs in the context of image segmentation. It will show the scheme to segmentation of objects (with interrupted edges) in 3-D images. We will demonstrate its unconditionally stableness, its convergence accuracy and low computational times. We present numerical experiments devoted to analysis of 3-D images in medicine and bioengineering.

Part-I

This workshop is organized by leading researchers in the field of level sets, digital image processing and computer vision. The workshop will be in two parts: Part-I will cover the basics of level set understanding, how they got originated, fundamental Eikonal Equations, Solutions to Eikonal Equations such as Fast Marching Strategies. Particular emphasis will be paid on application of level sets in medical image analysis such as boundary estimation of anatomic structures in two dimensional images. Stopping forces are very important to get the accurate segmentation of objects. Emphasis will be paid on integrating different kinds of classifiers with level sets to make the system robust. Extension of level sets will be presented in 3-D medical imaging volumetric data sets. Lot of applications will be covered such as in Neurology, Cardiology and Mammography.

Part-II

Part-II will present advance applications of level sets such as subjective level sets or missing boundary estimation. A Partial Differential Equation (PDE) based flow is designed in order to achieve a maximum likelihood segmentation of the target in the scene. The flow is derived as the steepest descent of an energy functional taking into account the density probability distribution of the gray levels of the image as well as smoothness constraints. To model gray level behavior of ultrasound images the classic Rayleigh probability distribution is considered. The steady state of the flow presents a maximum likelihood segmentation of the target. A finite difference approximation of the flow is derived and numerical experiments are provided. Results are presented on ultrasound medical images as fetal echography and echocardiography, as well as images with Poisson and Gauss distributions.

In Part-II, we will also cover semi-implicit co-volume level-set method in 3D image segmentation. Here the workshop will present three-dimensional semi-implicit complementary volume level-set scheme for solving the Riemannian mean curvature flow of graphs in the context of image segmentation. It will show the scheme to segmentation of objects (with interrupted edges) in 3-D images. We will demonstrate its unconditionally stability, and the convergence accuracy and low computational times. We present numerical experiments devoted to analysis of 3-D images in medicine and bioengineering.