HIGHER EDUCATION:
WORK EXPERIENCE.
JOURNAL PUBLICATIONS.
SUBMITTED TO JOURNAL PUBLICATIONS.
PUBLICATIONS IN CONFERENCE PROCEEDINGS.
LANGUAGES. Spanish: Mother tongue. English: Speak and write fluently.
French: Basics.
PROGRAMMING LANGUAGES.
The project is divided into two stages. The first is the detection and segmentation of the blood vessels to convert the original image into a binary image. For this purpose we use an automatic segmentation algorithm based upon the scale-space analysis of the first and second derivative of the intensity image which gives information about its topology and overcomes the problem of variations in contrast inherent in these images. We use the local maxima over scales of the magnitude of the gradient and the maximum principal curvature as the two features used in a region growing procedure. Initially, the growth is constrained to regions of low gradient magnitude. Later on this constraint is relaxed to allow borders between regions to be defined. The algorithm is tested in both red-free and fluorescein clinical retinal images.
The second stage of this work is to perform the analysis of the tree geometry from the binary images by making measurements of the morphometrical parameters such as lengths, widths, and branching angles of the blood vessels, from these measurements other important geometrical parameters are extracted. The measurements are made with a semiautomatic process in which the skeleton of the segmented trees are tracked and marked with a chain code. Skeletons should be properly corrected and marked to distinguish, for example, between a bifurcation and an intersection of vessels, or to erase false bifurcations, etc. The data for each individual vascular tree are taken and lengths, areas and branching angles are measured. The data base of measurements is organised in such a way that it is possible to track the vessel tree in any direction, from the root to the leaves or from the leaves towards the root.
We validated the measurements using a set of red-free and fluorescein images taken in the same subjects in order to have measurements from the same blood vessel taking fluorescein as our standard measure. Finally, we will apply our segmentation and measurement methods to clinical retinal images from hypertensive patients and age matched controls to study the relationship between vascular patterns and disease.
The accuracy of any image segmentation method depends not only on the correct estimation of the parameters of the different regions present in the image, but also on the correct labelling of the pixels. By using robust estimators and relaxation labelling techniques, an unsupervised segmentation algorithm was developed. The mean gray value of each region is estimated from the histogram by using robust clustering analysis. The gray level distribution of each individual region is approximated through the mean gray value cooccurrence data. The standard deviation of the gray levels of each region is estimated from this distribution by using the Least Median of Squares (LMedS) robust estimator. The labelling of the pixels is done through an iterative relaxation region growing process, taking into account both spectral and spatial information. The method is tested in various images and validated with synthetic data, where it is shown that the known true parameters are recovered accurately.
Wed Aug 9 10:18:53 CDT 2006