A hierarchical multi-phase image segmentation using the original and a modified Chan-Vese 2-phase
method is considered. A method of capturing features inside a pre-selected region of interest (ROI) is
proposed that effectively restricts the segmentation operation to the ROI. At the first step, a modified
image is created by setting the portion of the image outside the ROI to a uniform intensity equal to the
mean image intensity inside the ROI. Effectively, this procedure partitions the initial image into two
phases, in such a way that the ROI effectively becomes a 'segmented' feature. At the second step, the
segmentation procedure is applied to the modified image, partitioning the image in two phases - <i>object</i> and
<i>background</i> - inside the ROI. By confining segmentation to the ROI, it is shown, using an artificial image,
that objects can be discriminated that could not have been found if segmentation had been performed on
the entire image. If necessary, this second step can be repeated to further segment features of interest
within the ROI, thereby providing a multi-phase segmentation procedure. ROI placement around features
of interest requires prior knowledge, and may be derived from an atlas or manually prescribed by the
operator. In this way, segmentation is possible on low-contrast features of interest, while ignoring features
irrelevant for a particular application. Examples are provided for segmentation of several 2D/3D images
performed both on entire images and inside a ROI.
A method is proposed for cross-modal image registration based on mutual information (MI) matching criteria. Both conventional and "normalized" MI are considered. MI may be expressed as a functional of a general image displacement field u. The variational principle for MI provides a field equation for u. The method employs a set of "registration points" consisting of a prescribed number of strongest edge points of the reference image, and minimizes an objective function D defined as the sum of the square residuals of the field equation for u at these points, where u is expressed as a sum over a set of basis functions (the affine model is presented here). D has a global minimum when the images are aligned, with a “basin of attraction” typically of width ~0.3 pixels. By pre-filtering with a low-pass filter, and using a multiresolution image pyramid, the basin may be significantly widened. The Levenberg-Marquardt algorithm is used to minimize D. Tests using randomly distributed misalignments of image pairs show that registration accuracy of 0.02 - 0.07 pixels is achieved, when using cubic B-splines for image representation, interpolation, and Parzen window estimation.
EvIdent (EVent IDENTification) is an exploratory data analysis system for the detection and investigation of novelty, identified for a region of interest and its characteristics, within a set of images. For functional magnetic resonance imaging, for instance, a characteristic of the region of interest is a time course, which represents the intensity value of voxels over several discrete instances in time. An essential preprocessing step is the rapid registration of these images prior to analysis. Two dimensional image registration coefficients are obtained within EvIdent by solving a regression problem based on integration of a linearized matching equation over a set of patches in the image space. The registration method is robust to noise, offers a flexible hierarchical procedure, is easily generalizable to 3D registration, and is well suited to parallel processing. EvIdent, written in Java and C++, offers a sophisticated data model, an extensible algorithm framework, and a suite of graphical user interface constructs. We describe the registration algorithm and its implementation within the EvIdent software.