Abstract
dc:description.abstractWe propose a novel deep learning based solution to overlapping cervical cell separation from cervical cell cytology images, aiming to improve the diagnostic process for cervical cancer by providing clear and detailed images of individual cells. The proposed solution involves a deep learning-based image separation pipeline for overlapping cervical cells. This pipeline first computes the overlapped portion of cells using a segmentation technique, which is a cascaded deep learning network comprising the Meta-Polyp model and a modified U-Net model employing transfer learning from a pretrained backbone network. The proposed segmentation pipeline's application is extended to obtain the union mask of overlapping cells. The resulting non-overlap segmentation mask is used in an instance segmentation network to yield individual cell masks. The cell separation pipeline, using a conditional generative adversarial network, generates separated cell images. The proposed pipeline generates accurate predictions of individual cervical cells, outperforming existing techniques.
Degree
thesis:*- Name thesis:degree_name
- Master of Applied Science (M.App.Sc.)
- Level thesis:degree_level
- Master's
- Discipline thesis:degree_discipline
- Engineering, Electrical and Computer
- Grantor dc:publisher
- Carleton University
- Year dc:date.issued
- 2024
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Fazal, Shaad
Subjects
dc:subject × 1Rights
dc:rights- Statement dc:rights
-
- Copyright © 2024 the author(s). Theses may be used for non-commercial research, educational, or related academic purposes only. Such uses include personal study, distribution to students, research and scholarship. Theses may only be shared by linking to the Carleton University Institutional Repository and no part may be copied without proper attribution to the author; no part may be used for commercial purposes directly or indirectly via a for-profit platform; no adaptation or derivative works are permitted without consent from the copyright owner.
- Language dc:language.iso
- en