City Research Online

Abstract

This dissertation investigates the volumetric analysis of a variety of cervical cancer cells called HeLa cells. HeLa cells were derived from cervical cancer cells taken from Henrietta Lacks at the Johns Hopkins Hospital and hence the name HeLa remains. The shape of cells is important as the regular or irregular shape of the cell and its structures can be related to some conditions of health or disease.

In this dissertation, a traditional image processing algorithm to segment the nuclear envelope of HeLa cells imaged with Serial Block Face Scanning Electron Microscopy is proposed. The algorithm is fast, robust and accurate and it was compared against different deep learning architectures. Three deep learning architectures were deployed through transfer learning and U-Net was trained from scratch for semantic segmentation of HeLa cells. The algorithm outperformed all four deep learning architectures and active contours (snakes) in both accuracy and time as suggested by the similarity metrics. The segmented nuclear envelope was further investigated through a visualisation technique to obtain a graphical model. This model provides volume and surface metrics which can be used to compare different cells. Wild-type of HeLa cells were compared against Chlamydia trachomatis-infected HeLa cells and geometric differences were revealed.

The open-source image processing algorithm, developed in programming environment of Matlab® (The MathworksTM, Natick, USA), provides cell segmentation in a fraction of manual segmentation time therefore it is an alternative to expensive commercial software and manual segmentation, which is still widely used despite the significant disadvantages of time and inter- and intra-user variability.

Publication Type:	Thesis (Doctoral)
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering
Departments:	Doctoral Theses School of Science & Technology > School of Science & Technology Doctoral Theses School of Science & Technology > Engineering

Export

Downloads