Desarrollo de un algoritmo para la determinación de histología virtual de imágenes intravasculares ultrasónicas

Abstract

El Síndrome Coronario Agudo (ACS), así como diversas patologías de índole vascular son actualmente causas potenciales de mortalidad y morbilidad a nivel mundial. Ante esta problemática se tiene como herramienta a la Histología Virtual, (HV-IVUS) la cual permite determinar la composición histológica de la placa aterosclerosa coronaria. El algoritmo aquí presentado tiene como principal fuente imágenes de Ultrasonido Intravascular para someterlas a un post procesamiento. A partir de ellas, se etiquetan los componentes histológicos suficientes para llevar a cabo la clasificación de la placa. De igual forma, se obtiene el volumen de cada uno de los componentes a partir de una reconstrucción en 3D. La clasificación de las placas se realiza con base en un algoritmo de lógica difusa. El control por lógica difusa arrojará un resultado en porcentaje de la probabilidad de que determinada variable pertenezca a cierto conjunto. En términos de la problemática aquí tratada pretende arrojar la probabilidad de que la imagen ilustre un cierto tipo de placa. Así, esta herramienta presenta una ayuda para el médico cardiólogo sin la necesidad de adquirir un equipo alterno al IVUS.

ABSTRACT. The Acute Coronary Sindromes (ACS) as many other vascular pathologies, are nowadays potential causes of mortality and morbidity all around the world. Hence, there has been developed such a tool (The Virtual Histology of Intravascular Ultrasound). This tool allows the determination of the histological composition of the culprit plaque lesion. It is capable to detect four different types of tissue: Necrotic, Fibrous, Fibro-lipidic and Calcified-core. The present algorithm uses Intravascular Ultrasound gray scale images, these images pass through a post-processing treatment. The target is to label each one of the required histological components for classifying the plaque. The Virtual Histology helps us to know the stability of the plaque what directly expresses the possibility of an Acute Coronary Sindrome, it also allows to have a tomographic view of the artery when a treatment procedure is developed, for example the stent angioplasty. The post-processing treatment is based on a texture segmentation, the texture filtering needs the entropy calculated in each zone limited by the window chosen, an average filtering has been managed for getting the joined zones of each texture in order to label and assigned each joined object to a different matrix, each gray scale of the image can be observed in a different figure. The labeling is used also for calculate the area of the joined “high” pixels, in order to indirectly know the area of each tissue. Each binarized matrix could be assigned to a different color, the sum of all matrices result on the display of the Virtual Histology. A 3D reconstruction has been designed in order to get the volume values of each component. The classification of the plaques is executed by fuzzy logic. The outputs of the fuzzy logic represent the probability percentage of each one of the possible types of plaque: Adaptive Intimal Thickening, Pathological Intimal Thickening, Fibro-atheroma, Thin-Cap Fibro-atheroma and Fibrocalcific. The algorithm classifies successfully each gray scale in all the images and videos that were processed, this gives the possibility of analyze images in a remote location. The 3D reconstruction is capable to display all the tissues at the same time of the complete lumen. For further works it is recommended to compare the results of the Virtual Histology designed with the one of Volcano’s including the diagnoses but this images are not easy available.