Dr. Arushi Garg, G15746, Dr.Krishnapoojita Vunnava, Dr. Rohit Shetty
Authors:
ARUSHI GARG, ROHIT SHETTY, KRISHNAPOOJITA VUNNAVA
(Degree and designation, institutional affliation, name and address of corresponding author)
ABSTRACT :
INTRODUCTION: Meibomian glands are an integral component of the ocular surface. Meibomian gland dysfunction or MGD,according to the International Workshop on Meibomian Gland Dysfunction1has been defined as a chronic, diffuse abnormality of the meibomian glands, characterized by terminal duct obstruction and/or qualitative orquantitative changes in the glandular secretion. The prevalence of MGD in Indian population has been reported to be as high as 30%2. The conventional dry eye workup comprises a host of table-side tests including the Ocular Surface Disease Index (OSDI) scoring, tear meniscus height measurement, tear film breakup time (TBUT), vital staining of the ocular surface, Schirmer’s test with and without anaesthesia, and a detailed slit lamp examination of the ocular surface. Out of these, meibomian gland function can be empirically evaluated only on slit lamp examination of the gland orifices at the lid margin and by the duration of tear film breakup time. Meibography is the technique of in vivo visualization and documentation of gland morphology3which is rapidly becoming an important component of a comprehensive dry eye clinic.Meibography can be either contact or non-contact,The various non-contact methods can be confocal-based, OCT-based or infra-red based meibography. Of these, the infrared based meibographers are currently the most popular meibographers in use as they are useful for the morphological characterization of the meibomian glands. This study was conducted to compare the quantitative and qualitative performance of a novel infra-red hand-held camera with the current gold standard instrument for meibography. We also evaluated the efficacy of a novel automated image analysis algorithm (MENTION THE NAME OF THE ALGORITHM) in deriving the metrics (EXPLAIN METRICS) in meibomian gland disease.
METHODS: This prospective cross sectional observational study was conducted on200 eyes of patientsin the age group of 30-60 years, presenting to a tertiary care hospital in South India, with or without symptoms of dry eye disease (DED).(KINDLY MENTION CLEARANCE FROM ETHICS COMMITTEE AND INSTITUTIONAL REVIEW BOARD) Patients who had any corneal, conjunctivalor lid abnormalities other than dry eye and MGD were excluded from the study. All subjects underwent meibomian gland imaging using a prototype infrared hand-held camera (Robert Bosch Engineering and business solutions) and a standard meibographer (Oculus Keratograph®). The Oculus Keratograph 4 (K4; OCULUS, Wetzlar, Germany) is primarily a corneal topographer with an infrared camera, originally meant for pupillometry,and adapted for meibography. The hand-held infra-red camera, on the other hand is equipped with 5 megapixel infrared camera (IR range of 875 nm wavelength) and has both autofocus and manual focus mode.
The image acquired on the hand-held camera is enhanced and then subjected to adaptive histogram neutralization technique (KINDLY EXPLAIN THE TERM ADAPTIVE HISTOGRAM NEUTRALIZATION TECHNIQUE) to highlight the meibomian glands. Gland marking was done by trained doctors using Fiji Image software. Edges of the lid area with the glands were marked with polygon tool and then edges of the individual glands were marked (ground truth manual marking) (PLEASE PROVIDE IMAGES IF POSSIBLE). The automated algorithm developed by Robert Bosch Engineering analyses the image by edge detection and morphological operations to segment the glands and then computes the metrics to quantify MGD, in the form of percentage of lid area without glands and number of tortuous glands.These images obtained by manual marking by trained doctors and the automated image analysis algorithm, are then compared for quantitative disease metrics. KINDLY MENTION THAT CLEARANCE WAS TAKE FROM ETHICS COMMITTEE AND IRB)
RESULTS: Out of 200 eyes, 100 had meibomian gland dysfunction (MGD), 100 were healthy eyes. The average percentage of lid area without glands,called drop-out, wascomputed using manual marking by trained health professionals and compared with the drop-out obtained using the automated algorithm.In healthy eyes, on the hand-held camera, the computed percentage lid area without glands from manual analysis was 58.3% (Standard error of mean 0.0698), and 58.16% (SEM 0.0887) using the novel automated software (P = 0.435), thereby showing comparability. For healthy eyes, using the standard meibographer, we observed an average drop-out of 59.0% (SEM 0.09) from manual marking and 56.4% (SEM 0.105) (P=0.205).In diseased eyes, the automated analysis showed the computed percentage lid area without glands as 65.8% (SEM 0.115), as compared to manual analysis which has 67.1 % (SEM 0.099). Using the standard meibographer, the mean drop-out is 64.6% (SEM 0.103) from manual markings, and from the automated analysis it is 65.5% (SEM 0.088). From the mean drop-out values, we can observe that the hand-held infrared camera was comparable to the standard meibographer in detecting diseased meibomian glands.
DISCUSSION: The standard meibographer is an expensive and bulky device that provides only the image of the glands. The hand-held infrared camera on the other hand, is a cheaper, portable device capable of image capturing and computation of metrics. The hand held infrared camera is able to capture adequate quality gland image in both diseased and healthy eyes. Metrics in the form of percentage of lid area without glands and number of tortuous glands were computed manually and using an automated algorithm. Manual method is time-consuming and tedious, and is subject to inter-observer variability. The automated software is as sensitive as manual marking that was done by health care professionals for the detection and computation of gland metrics in healthy eyes.
CONCLUSIONS: The hand-held device can be a cost-effective, portable alternative to diagnose MGD in clinics. The novel automated software is able to reliably detect the glands and compute metrics for better classification of meibomian gland disease.
REFERENCES:
- Nichols KK, Foulks GN, Bron AJ, et al. The International Workshop on Meibomian Gland Dysfunction: Executive Summary. Investigative Ophthalmology & Visual Science. 2011;52(4):1922-1929.
- Basak SK, Pal PP, Basak S, Bandyopadhyay A, Choudhury S, Sar S.Prevalence of dry eye diseases in hospital-based population in West Bengal, Eastern India. J Indian Med Assoc. 2012 Nov;110(11):789-94
- Jester J.V., Rife L., Nii D., Luttrull J.K., Wilson L., Smith R.E. Invivobiomicroscopy and photography of meibomian glands in a rabbit model of meibomian gland dysfunction. Invest Ophthalmol Vis Sci. 1982;22:660–677
Leave a Comment