Deep Neural Networks in Medical Image Classification
Subject and Keywords:diagnostyka obrazowa ; rak piersi ; sztuczne sieci neuronowe (SSN) ; Polska ; raport z badań ; rozprawa doktorska
Abstract:The paper presents two approaches to the problem of classifying cytological images. The first approach is based on the extraction of deep features resulting from training the network to correctly classify a set of cytological and histopathological images. The second approach starts with semantic segmentation of objects in the image. A diverse set of features is then created based on the segmented cell nuclei. Extraction, which belongs to dimensionality reduction methods, is one of the basic approaches to building a set of features. ; Thanks to extraction, the original space of the input image is reduced to a set of features describing objects located in the space of this image. Features can be generated in a traditional way, i.e. by determining the morphometric, colorimetric and textural properties of cell nuclei. The set of activities performed in this approach constitutes the comprehensive classi cation system presented in this work. ; To achieve the goal, it was necessary to perform detailed research on the veri cation of the best classi ers, dimensionality reduction methods and the issue of fusion between deep and manual features. The dissertation presents the results obtained in the course of comprehensive research meant to select the most fitting classification model.
Description:seria: Lecture Notes in Control and Computer Science, vol. 28
Publisher:Zielona Góra: Oficyna Wydawnicza Uniwersytetu Zielonogórskiego
Date: Resource Type: Format: DOI:DOI https://doi.org/10.59444/2024MONaSko_LNCCS_v28
Language: Rights: