作者: Hermilo Sánchez-Cruz , Humberto Sossa , J. Luis Quintanar , Alfonso Vizcaíno
DOI: 10.1155/2021/6663977
关键词: Moment (mathematics) 、 Random forest 、 Pattern recognition 、 Pixel 、 Radial basis function 、 Categorization 、 Metric (mathematics) 、 Computer science 、 Artificial intelligence 、 Multilayer perceptron 、 Support vector machine
摘要: This paper presents a method for pixel-wise classification applied the first time on hippocampus histological images. The goal is achieved by representing pixels in 14-D vector, composed of grey-level information and moment invariants. Then, several popular machine learning models are used to categorize them, multiple metrics computed evaluate performance different models. multilayer perceptron, random forest, support vector machine, radial basis function networks were compared, achieving perceptron model highest result accuracy metric, AUC, score with highly satisfactory results substituting manual task, due an expert opinion