Automatic colposcopy video tissue classification using higher order entropy-based image registration
Computers in Biology and Medicine
Hi-index | 0.00 |
In this article, a technique for the automated registration of Cervigram驴 images willbe introduced. The motivation for the development of such a technique is warranted by the fact that registration is often a first step to other, more sophisticated, algorithms useful in medical applications. Such algorithms are typically processes developed for the tracking andmonitoring of patient health. The registration described in this article is segmentation-basedand utilizes a combination of clustering- and active contour-based methodologies. Theclustering algorithm is used to obtain an initial contour that will subsequently serve asinitialization for an active contour. The active contour, in conjunction with various internal andexternal forces, should converge to a more precise segmentation of the region of interest which, in this application, is the cervix. Once the segmentations are completed, more traditional registration techniques, such as those of Fourier- or correlation-based techniques, may be used to register the segmented images with more accuracy as the adverse effects stemming from the highly variable background features are no longer a source of error. For illustration purposes, the results from two patients are demonstrated at the end of this article.