Saltar al contenido
República del Perú · Observatorio de la producción científica universitaria
OApol.org Observatorio de Inteligencia Científica Universitaria del Perú
OApol.org
Inicio Universidades Publicaciones Investigadores
Repositorios
Tesis y trabajos académicosRevistas
Indicadores
Ranking trimestralRanking por carreraObjetivos de Desarrollo SostenibleReporte OCDERed neuronal 3D
Acerca
MetodologíaPrivacidad

Algorithm for Optimization in Medical Image Processing applied in Heterogeneous Architecture

Auccahuasi W.; Meza S.; Porras E.; Reyes M.; Linares O.; Rojas K.; Inciso-Roja M.; Pando-Ezcurr T.; Aiquipa G.; Nicolas-Roja Y.; Auccahuasi A. · CEUR Workshop Proceedings (2022)

Resumen

In these times of pandemic, hospitals are being the focus of many innovations, not only for the adaptation to telemedicine, but also from the perspective of the use and processing of the multiple modalities of medical images, where we find images made up of a single Image such as x-rays, images that are made up of a sequence of images such as tomography and Magnetic Resonance, or in video format as is the case with ultrasound and angiography. One way of working with images is through popular image servers that connect to medical equipment for transfer and storage. In the process of visualization and processing, special workstations with good computational capacity are required for these purposes, in most cases these workstations are connected in the network of medical offices, therefore they are presented in a normal working image display requests at the same time. The methodology presented uses a heterogeneous architecture based on CPU and GPU, in such a way that by means of an algorithm it analyzes the type and dimension of the image to be able to choose where the processing will be carried out, thereby optimizing the use of computational resources. and we can achieve a parallel job that the CPU and GPU are working simultaneously with different imaging modalities. As a result, we present the execution mode of the algorithm where it automatically chooses what type of image is processed by the CPU and what type is processed in the GPU, as well as the execution time in each of them. Finally we can indicate that the algorithm can be scalable towards workstations to optimize its use in clinical practice.

Instituciones: Universidad ESAN; Universidad Científica del Sur; Universidad Tecnológica del Perú; Universidad Continental; Universidad Tecnológica de los Andes; Universidad Privada del Norte; Universidad de Ingeniería y Tecnología