Resumen

HEROLD GARCIA, Silena  y  ESCOBEDO NICOT, Miriela. MEDICAL IMAGE SEGMENTATION APPLYING DEFORMABLE MODELS TO ENDOSCOPY AND ULTRASOUND. uct [online]. 2008, vol.12, n.47, pp.87-92. ISSN 1316-4821.

Endoscope and ultrasound images have become in one of the most important not invasive diagnostic methods in medicine, because they allow the study of areas of smaller size and difficult access. Segmentation in these images is a partially resolved problem, given the necessity to create new methods and to optimize the existent ones to achieve a better automation, due to the diagnostic repercussion that is derived of this process. A prototype that allows applying the snake’s segmentation techniques to endoscope and ultrasound images of cerebral tumours is presented, to achieve the detection of the tumour’s area. The process is carried out with parametric snakes in two ways, using the gradient of the image and the Gradient Vector Flow (GVF), this last one because it allows a less demanding initialization of the snake and bigger precision in the detection of the wanted contours. This allows solving a considerable percentage of practical problems related with the analysis of images. During the validation of the obtained prototype it was proven the effectiveness of the implemented method applying it to several images of the referred modalities and it was determined that it is more efficient and has better results in endoscope images.

Palabras clave : Medical Images Segmentation; Deformable Models; Endoscopy; Ultrasound; Parametric Snake..