Setu, Md Asif Khan ORCID: 0000-0003-4195-4217 (2022). Development of Novel Diagnostic Tools for Dry Eye Disease using Infrared Meibography and In Vivo Confocal Microscopy. PhD thesis, Universität zu Köln.
PDF (PhD thesis)
Md_Asif_Khan_Setu_IPHS_PhD_Thesis.pdf - Accepted Version Bereitstellung unter der CC-Lizenz: Creative Commons Attribution Non-commercial No Derivatives. Download (6MB) |
Abstract
Dry eye disease (DED) is a multifactorial disease of the ocular surface where tear film instability, hyperosmolarity, neurosensory abnormalities, meibomian gland dysfunction, ocular surface inflammation and damage play a dedicated etiological role. Estimated 5 to 50% of the world population in different demographic locations, age and gender are currently affected by DED. The risk and occurrence of DED increases at a significant rate with age, which makes dry eye a major growing public health issue. DED not only impacts the patient’s quality of vision and life, but also creates a socio-economic burden of millions of euros per year. DED diagnosis and monitoring can be a challenging task in clinical practice due to the multifactorial nature and the poor correlation between signs and symptoms. Key clinical diagnostic tests and techniques for DED diagnosis include tearfilm break up time, tear secretion – Schirmer’s test, ocular surface staining, measurement of osmolarity, conjunctival impression cytology. However, these clinical diagnostic techniques are subjective, selective, require contact, and are unpleasant for the patient’s eye. Currently, new advances in different state-of-the-art imaging modalities provide non-invasive, non- or semi-contact, and objective parameters that enable objective evaluation of DED diagnosis. Among the different and constantly evolving imaging modalities, some techniques are developed to assess morphology and function of meibomian glands, and microanatomy and alteration of the different ocular surface tissues such as corneal nerves, immune cells, microneuromas, and conjunctival blood vessels. These clinical parameters cannot be measured by conventional clinical assessment alone. The combination of these imaging modalities with clinical feedback provides unparalleled quantification information of the dynamic properties and functional parameters of different ocular surface tissues. Moreover, image-based biomarkers provide objective, specific, and non / marginal contact diagnosis, which is faster and less unpleasant to the patient’s eye than the clinical assessment techniques. The aim of this PhD thesis was to introduced deep learning-based novel computational methods to segment and quantify meibomian glands (both upper and lower eyelids), corneal nerves, and dendritic cells. The developed methods used raw images, directly export from the clinical devices without any image pre-processing to generate segmentation masks. Afterward, it provides fully automatic morphometric quantification parameters for more reliable disease diagnosis. Noteworthily, the developed methods provide complete segmentation and quantification information for faster disease characterization. Thus, the developed methods are the first methods (especially for meibomian gland and dendritic cells) to provide complete morphometric analysis. Taken together, we have developed deep learning based automatic system to segment and quantify different ocular surface tissues related to DED namely, meibomian gland, corneal nerves, and dendritic cells to provide reliable and faster disease characterization. The developed system overcomes the current limitations of subjective image analysis and enables precise, accurate, reliable, and reproducible ocular surface tissue analysis. These systems have the potential to make an impact clinically and in the research environment by specifying faster disease diagnosis, facilitating new drug development, and standardizing clinical trials. Moreover, it will allow both researcher and clinicians to analyze meibomian glands, corneal nerves, and dendritic cells more reliably while reducing the time needed to analyze patient images significantly. Finally, the methods developed in this research significantly increase the efficiency of evaluating clinical images, thereby supporting and potentially improving diagnosis and treatment of ocular surface disease.
Item Type: | Thesis (PhD thesis) | ||||||||||||||||
Translated abstract: |
|
||||||||||||||||
Creators: |
|
||||||||||||||||
URN: | urn:nbn:de:hbz:38-641831 | ||||||||||||||||
Date: | 22 November 2022 | ||||||||||||||||
Publisher: | Digital Express 24 | ||||||||||||||||
Place of Publication: | Köln | ||||||||||||||||
Language: | English | ||||||||||||||||
Faculty: | Faculty of Medicine | ||||||||||||||||
Divisions: | Faculty of Medicine > Augenheilkunde | ||||||||||||||||
Subjects: | Data processing Computer science Natural sciences and mathematics Life sciences Medical sciences Medicine |
||||||||||||||||
Uncontrolled Keywords: |
|
||||||||||||||||
Date of oral exam: | 22 November 2022 | ||||||||||||||||
Referee: |
|
||||||||||||||||
Refereed: | Yes | ||||||||||||||||
URI: | http://kups.ub.uni-koeln.de/id/eprint/64183 |
Downloads
Downloads per month over past year
Export
Actions (login required)
View Item |