Contenido principal del artículo
Sixty brain volumes were analyzed from separate magnetic resonance images in two populations: healthy control subjects and subjects with mild cognitive impairment. For each element, the Box Counting algorithm was applied to obtain the characterization of the 3D Fractal Dimension that it presented. In addition, other morphological indices of volume, discrete compactness and surrounding area were added. Finally, a classification strategy is implemented using the Gaussian Process with a radial-based Kernel to compare the correct discrimination in the populations studied. The classifier model after the validation process gave a 60% success rate for the fractal dimension and for the morphological indices. The highest hit rate was when both metrics were combined with 80%. With these results, it is believed that the fractal index could become a biomarker for the prediagnosis of Alzheimer's disease, although for this, more studies are needed in the future.
Detalles del artículo
DERECHOS DE AUTOR Y DERECHOS CONEXOS, las MEMORIAS CONGRESO NACIONAL DE INGENÍERIA BIOMÉDICA es una publicación editada por la Sociedad Mexicana de Ingeniería Biomédica A.C., Plaza Buenavista, núm. 2, Col. Buenavista, Delegación Cuauhtémoc, C.P. 06350, México, D.F., Tel. +52 (555) 574-4505, www.somib.org.mx, correo-e: firstname.lastname@example.org. Editor responsable: Elliot Vernet Saavedra. Reserva de Derechos al Uso Exclusivo No. 04-2015-011313082200-01, ISSN: 2395-8928, ambos otorgados por el Instituto Nacional de Derechos de Autor.