V Congreso Latinoamericano de Ingeniería Biomédica (CLAIB2011)

Segmentación de Lesiones Cérvico Uterinas Aplicando Contornos Activos sobre Patrones Espacio Temporales
Acosta Mesa Hector-Gabriel, Hernández Jiménez Rodolfo, Gutiérrez Fragoso Karina, Magallanes Lugo Ariadna Jesús

##manager.scheduler.building##: Palacio de Convenciones de La Habana
##manager.scheduler.room##: Sala 12
Fecha: 2011-05-20 09:00  – 10:30
Última modificación: 2011-04-29 10:31


Cervical cancer is the second cause of death in Mexican women. After Pap smear test, colposcopy is the most used technique to diagnose this disease due to its higher sensitivity and specificity. Colposcopy is used to guide the biopsy selection. One of the most important colposcopic findings is the level of white color intensities that the cervical tissue reaches after acetic acid application (aceto-white effect). In order to improve the interpretation of colposcopic images, some digital image processing techniques have been proposed, however a recurrent problem with these methods is the presence of  specular reflection produced by the moist nature of the cervical tissue. To overcome this problem, active contours have been applied  to colposopic images using multi scale Gaussian Piramids (GP) representation. However the boundary gradients are computed using the gray scale differences in just one image taken at a specific time point after acetic acid application. In the present work we investigate how the dynamic properties of the aceto- white effect can be integrated in the analysis in order to improve the segmentation.  Results shows that under this approach the analysis not only does not requires GP but also doest not depends on the time at which the image was taken.