THESIS
2022
1 online resource (114 pages) : illustrations (some color)
Abstract
Glaucoma is one of the most prevalent eye diseases in the world that causes blindness. Early-stage glaucoma is often exhibited through changes of optic nerve and the ischemia grows progressively. It is traditionally believed that optic nerve ischemia is mainly attributed to the elevated intraocular pressure (IOP) and/or the blockage of the aqueous humor flowing inside the eyeball. However, current assessments of ocular biomechanics preceding optic nerve ischemia cannot discriminate glaucomatous eyes and sometimes even cause misdiagnosis. The current gold standard of IOP measurement, Goldmann Applanation Tonometry (GAT) could hardly guarantee the accuracy due to the corneal geometric and biomechanical variations of individuals. Besides, a practical, non-invasive method to characterize th...[
Read more ]
Glaucoma is one of the most prevalent eye diseases in the world that causes blindness. Early-stage glaucoma is often exhibited through changes of optic nerve and the ischemia grows progressively. It is traditionally believed that optic nerve ischemia is mainly attributed to the elevated intraocular pressure (IOP) and/or the blockage of the aqueous humor flowing inside the eyeball. However, current assessments of ocular biomechanics preceding optic nerve ischemia cannot discriminate glaucomatous eyes and sometimes even cause misdiagnosis. The current gold standard of IOP measurement, Goldmann Applanation Tonometry (GAT) could hardly guarantee the accuracy due to the corneal geometric and biomechanical variations of individuals. Besides, a practical, non-invasive method to characterize the ocular drainage of aqueous humor has not yet been developed. In our study, a corneal indentation methodology is adapted to measure the ocular biomechanical properties. The experimental method was verified on porcine eyes in vitro. The analysis using biomechanics model and empirical data showed that quasi-static corneal modulus and scleral viscosity as a proxy of drainage had been successfully determined. Further, the quasi-static corneal modulus can be applied to correct the GAT IOP. A corrected cornea-independent IOP (ci-IOP) was demonstrated more than 30% improvement compared with the GAT method.
Similar to IOP, these comprehensive geometric and biomechanical properties such as corneal thickness, stiffness and modulus are highly correlated to glaucomatous damages. However, each single indicator has a wide range of overlap among glaucomatous and healthy eyes in the population. This indicates each of them is not able to serve as a precious diagnostic criterion for glaucoma. A combined diagnostic model using machine learning was therefore proposed to classify glaucomatous damages and healthy eyes. A preliminary result with small clinical data demonstrated even simple models can achieve higher accuracy and sensitivity in glaucoma detection than the traditional diagnostic criterion. The sophisticated machine learning models were then trained by a bigger clinical study, and unlike a black box, the explainable models revealed the relationship between biomechanical properties and the glaucomatous damages with deeper insight. This may help the community better screen out the disease and help the clinicians better understand the progression of glaucoma. Patients would also benefit from precious treatments with quantifiable glaucomatous defects in probability.
Post a Comment