Automated Morphological Analysis of Meibomian Glands

Meibomian glands are essential for maintaining ocular surface health and integrity secrete various lipid components to forms a lipid layer to prevent excessive tear evaporation. Functional disorders of the meibomian glands, referred to today as meibomian gland dysfunction (MGD), are increasingly recognized as a high incidence disease commonly characterized by terminal duct obstruction and/or abnormal glandular secretion, often results in ocular surface epithelium damage, chronic blepharitis and dry eye disease that significantly reduces quality of life. A wide variation of the prevalence of MGD were reported from 0.39% to 69.3%, which is likely due to lack of diagnostic methods. To identify which clinical features are likely to be predictive of progressive disease in MGD may indicate the early diagnosis and proper treatment strategies.

Histologic section through the normal meibomian glands and the obstructed human meibomian gland revealed that obstruction of orifice in MGD could lead to dilation of the central duct, damage of the secretory meibocytes and finally result in atrophy of dilated meibomian glands and glands drop-out. It was thus be accepted that detailed changes of meibomian glands morphology are key signs to diagnose and evaluate the severity of MGD. The detailed changes including dilation, distortion, shortening and loss of visualisation of glands which can be directly observed and visual assessment by the developed of non-contact meibomian gland infrared imaging technology. Quantitative evaluations of meibomian glands were obtain by developing imaging processing techniques. The most common use is the image editing software Image J (National Institute of Health; http://imagej.nih.gov/ij) which can identify the gland region on the image manually by the users and may lead to inter-observer variability. Koh et al., first applied original algorithms to automatically analysed gland loss in meibography images with a manually pre-processing. Reiko et al., then develop an objective and automatic system to measure the meibomian gland area. However, the existing methods of meibomian gland analysis have been limited to clinical use where large number of images needs to be analyzed efficiently due to the inter-observer variability or time-consuming process.

Meanwhile, the existing quantitative morphological parameters obtain by those imaging processing techniques, including percentage of MG drop-out and gland atrophy area, were suggested to not only be advanced stages or terminal changes in MGD, but also occurs as an age-related atrophic process. The early findings of MGD induced by the primary pathologic obstruction including degenerative gland dilation, irregularly shapes of gland and change of meibum quality are still difficult to be evaluated automatically and quantitively from the image. Moreover, the meibomian gland drop-out is still an approximate assessment without specific pattern. Whether the atrophy or loss occur in upper or lower eyelids, central, distal or proximal, total loss of gland or partial loss of gland has the greatest effect on the pathology progress of MGD will be important to identify. Thus, a comprehensive analysis technique to automatically detect multi-information of meibomian gland morphology will benefit the future early diagnosis and management of MGD.

Recently, an automated quantitative meibomian gland analyzer based on all kinds of infrared meibomian gland images was develop to obtain more detail in meibomian gland, including width, length, area, signal intensity correlated to the quality of meibum, deformation index and ratio of area of each visible specific gland. The purpose of this study is present as separate sections the following points: (1) to compared the detailed characteristics of meibomian glands in normal subjects and MGD patients by the automated quantitative analyzer; (2) to identify the inter-examiner and intra-examiner repeatability of the new technique; (3) to explore the correlation among morphological and functional parameters of meibomian gland and risk factors，clinical symptoms and signs; (4) to explore the sensitivity and specificity of meibomian gland morphological and functional parameters in MGD diagnosis. (5) using morphological and functional parameters as new assessment of MGD severity and efficacy indicators for treatment.