PAP-SMEAR Image Analysis

A tétel rövid nézete
dc.contributor.advisorKovács, László
dc.contributor.authorUllah, Abaid
dc.contributor.departmentDE--Informatikai Karhu_HU
dc.date.accessioned2019-04-29T09:48:45Z
dc.date.available2019-04-29T09:48:45Z
dc.date.created2019-04-29
dc.description.abstractIt has been demonstrated that nucleus highlights for automated PAP - smear picture order can be very instructive. Therefore, the detection of the cell nucleus pursued by the division of the nucleus in single - cell PAP - smear pictures might be the essential advance for automated screening. We propose a patch based approach utilizing U-net for a division of nuclei in single cell images. We at that point offer the conversation starter of the particle of division for arrangement utilizing portrayal learning with CNN, and whether low-level U-net highlights might be valuable for an order. We propose a U-Net-based component level examination and an exchange learning based methodology for grouping utilizing both sectioned as well as full single cell pictures. We additionally propose a choice tree based methodology for arrangement. Trial results show the viability of the proposed calculations separately (with low-level U-net OR CNN highlights), and at the same time demonstrating the adequacy of cell-nucleus location (as opposed to exact division) for grouping. Along these lines, we propose a framework for an examination of multi-cell PAP-smear pictures comprising of a basic nucleus identification calculation pursued by arrangement utilizing transfer learning. Cervical cancer remains one of the deadliest cancers among women in the world, particularly since it is the most common cause of death in developing countries. Approximately 500,000 new cases were reported annually, of which 85 percent occur in developing countries, along with around 270,000 deaths worldwide. Cervical cancer precancerous lesions take nearly a decade to convert into cancerous lesions. Thus, unlike many other cancers, despite the above facts, it can be completely cured if detected early.hu_HU
dc.description.correctorN.I.
dc.description.courseComputer Sciencehu_HU
dc.description.degreeMSc/MAhu_HU
dc.format.extent37hu_HU
dc.identifier.urihttp://hdl.handle.net/2437/266353
dc.language.isoenhu_HU
dc.subjectPap-smear image analysishu_HU
dc.subjectDigital Colposcopyhu_HU
dc.subjectCervical cell classificationhu_HU
dc.subject.dspaceDEENK Témalista::Informatikahu_HU
dc.titlePAP-SMEAR Image Analysishu_HU
Fájlok
Gyűjtemények
Theses (IK)