The conventional techniques for refractive error measurements (myopia, hypermetropia, and astigmatism) have been considered inadequate for several optometry researches. In this context, they have investigat.ed alternative methodologies for refractive error measurement. A new strategy is the determination of refractive errors from images of the globe of the eye. A process named Hartmann-Shack can obtain these images. The HS images should be analysed in order t o extract relevant inforniation for identification of refractive errors. The present paper investigates a technique based on Radial Basis Functions (RBFs), a n Artificial Neural Network (ANN), and on Support Vector Machines (SVMs), which automatically performs analysis of images from the globe of the eye and identifies refractive errors. The most relevant data of these images are extracted using Gahor wavelets transform, and then these Machine Learning techniques carry out the image analysis.
Este artigo relata uma nova abordagem para a avaliação dos vícios refrativos do olho humano (miopia, hipermetropia e astigmatismo) por meio de uma técnica de Aprendizado de Máquina. Esses vícios são diagnosticados a partir da análise de imagens do olho adquiridas por uma técnica específica, denominada Hartmann-Shack (ou Shack-Hartmann), as quais são pré-processadas utilizando análise de histograma e informações geométricas e espaciais do domínio da aplicação. Em seguida, vetores de características são extraídos por meio de duas técnicas: Análise de Componentes Principais e Tranformada wavelet de Gabor. Por fim, o conjunto de dados com os vetores de características extraídos é analisado por Support Vector Machines. Apesar das limitações do conjunto de imagens, resultados encorajadores foram obtidos, indicando o potencial desta abordagem na área de Optometria/Oftalmologia. The article introduces a new image analysis approach for measuring refractive errors in the human eye (myopia, hypermetropia and astigmatism) using Machine Learning techniques. These refractive errors are identified through the analysis of images of the eye obtained with a specific technique known as Hartmann-Shack (or Shack-Hartmann), which are preprocessed with histogram analysis considering spatial and geometrical information on the application domain. Afterwards, feature vectors are extracted using two techniques: Principal Component Analysis and Gabor Wavelets Transform. Finally, the dataset with the extracted feature vectors is analyzed using Support Vector Machines. In spite of the limitations of the image dataset, encouraging results were obtained, suggesting the potential of the proposed approach in Optometry/Ophthalmology
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.