Título: Classificação do Espectro Raman Compactado de Tecidos Arteriais Utilizando Redes Neurais
Autores: Paula Jr, Alderico R.; Sathaiah, Sokki
Resumo: Raman spectroscopy is a powerful non-destructive technique that has a high potential application for in vivo diagnosis of atherosclerotic plaques in human arteries. In some real time applications, a rapid collection and analysis of the Raman spectrum is needed. In such applications the noise generated by the detector may have the same level as the tissue Raman signal what makes the analysis difficult. This article presents the results of a study to process the Raman spectrum acquired from artery tissues when irradiated by a low power infrared laser. After being preprocessed, the Raman spectrum is compressed with the utilization of discrete wavelet transforms or principal component analyzes. Then, the tissue is classified into a non-pathological tissue, atherosclerotic lesion or calcified tissue utilizing neural networks. It was verified that the Raman spectrum, after being compressed from 592 up to 10 variables, presented correct classification rate greater than 95% when neural networks were used for detector exposition time as low as 20 ms.
Código DOI: 10.21528/CBRN2003-009
Artigo em PDF: 6CBRN_009.PDF
Arquivo BibTex: 6CBRN_009.bib