Thoughts concerning the application of thermogram images for automated diagnosis of dry eye - A review

Abstract Dry eye disease (DED) is a multifactorial condition of the tear and ocular surface that is characterized by loss of homeostasis and symptoms of tear instability. Some symptoms of DED include blurring of vision, crusting of eyelids, and irritation to the eyes. An array of clinical methods, such as Schirmer’s test, is used to classify DED. Yet, these approaches are often invasive, need to be performed manually by clinicians, and/or they do not have reproducible results. They are also prone to interobserver variation among clinicians. Thus, computer-aided detection (CAD) systems are preferred for DED diagnosis. This paper reviews the existing CAD techniques used to automatically diagnose DED, and focuses on the benefits of thermographic CAD systems. CAD systems for DED using thermography are found to be highly sensitive, specific, accurate, minimally invasive, convenient, and satisfactory. Also, deep learning techniques are discussed to precede conventional machine learning techniques in the development of a CAD system. It is concluded that the use of thermographic CAD systems coupled with a deep learning technique is likely to be useful for DED assessment in future work.