Neural network based on local binary patterns for medical image analysis

• Grant number: 20/12549-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: May 01, 2021
• End date: February 28, 2022
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Joao Batista Florindo
• Grantee: Juliano Amadeu Lopes Moura
• Host Institution: Instituto de Matemática, Estatística e Computação Científica (IMECC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil

Abstract

This project proposes the study and development of mathematical/computational tools for medical image analysis, introducing image descriptors based on local binary patterns into the pipeline of a multilayer perceptron neural network. In fact, neural networks, especially convolutional neural networks, have become popular in image analysis in general and for medical images in particular. However, other approaches to neural networks have recently been investigated and one of these is based on the combinations of classical neural networks with other image descriptors. A popular example of these descriptors is local binary patterns, obtained by comparing the value of each pixel with its neighboring pixels. This type of network has shown promising results in object recognition for general purposes and we intend to investigate its application to texture images, with an especial focus on diagnostic aid in the medical field. More specifically, the methodology will be applied to the identification of types of lung cancer. The obtained results are expected to have important implications for society, helping to better understand carcinogenic processes and thus fostering the possibility of earlier diagnosis and more effective treatment, improving patient life quality and life expectancy. (AU)