Multilayer Perceptron Models for Classification of Cervical Precancerous Cells Based on FE-SEM/EDX Features
Yessi Jusman, Masayu Alya Nur ’Aini, Karisma Trinanda Putra, Siew‐Cheok Ng, Khairunnisa Binti Hasikin, Kean‐Hooi Teoh · 2023
Field emission scanning electron microscopy (FE-SEM) has been widely applied in research and engineering. Field emission scanning electron microscopy and energy-dispersive X-ray (FE-SEM/EDX) can capture high-resolution cervical cell shape and elemental composition images. This paper aims to utilize the high peaks of the elemental composition based on FE-SEM images and EDX spectra. The features of the cervical cells were extracted the ratio of high peaks. The classification employed neural network Levenberg Marquardt backpropagation and scaled conjugate gradient backpropagation models. The accuracy results reached 84.6%. Levenberg-Marquardt backpropagation classifier required 29 epochs to achieve the best performance.