EARLY DETECTION OF DIABETIC NEUROPATHY USING RETINAL IMAGE

Purpose:

Diabetic neuropathy is one of the most common complications found in diabetes patients with a prevalence ranging from 7%-68%. Estimation from a multi-hospital survey in China indicated that 18% of diabetes patients and 6.4% of newly diagnosed patients had perception defects. Moreover, only one-third of the studied subjects had previously gone through neuropathy screening, and more than two third among those diagnosed with diabetic neuropathy undertook formal treatment. 

Method(s):

Several reasons may account for the low screening rate small proportion receiving treatment for this disease.  One of which could be related to the complexity of current screening process, which not only requires various equipments but also trained and experienced doctors. Secondly, lack of objective measurements in clinical or family care's setting affect the ability to identify the problems. Also, as the disease progresses slowly with bearable discomfort, patients usually ignore the symptoms until a late stage.  Thus, an assessment tool that is objective, easy to operate and accurate will not only help increase the screening and diagnostic capability but also raise patients' awareness of the disease. 

Microvascular abnormality has been studied extensively over the past decades. Both animal models and pathological anatomy had shown the abnormal structural changes in the microcirculation as the earliest pathological sign around the impaired nerve. Thus, it was proposed that vessel assessment be used as a potential tool for early risk assessment and detection of the neuropathy.  Retina was viewed as the most promising site for microcirculation evaluation because it is the only place where microcirculation can be view noninvasively.  Large population-based cohort studies, including The Atherosclerosis Risk in Communities Study (ARIC), The Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR) and EURODIAB Prospective Complications Study all reported a moderately positive association between diabetic retinopathy and neuropathy. Our study with 2,127 type 2 diabetes subjects collected from the U.S Nutrition and Health Examination Study (NHANES) dataset also confirmed these findings.

Result(s):

The development of sophisticated image analysis technology made it possible to extract detail information from retinal vessels such as caliber, tortuosity, and bifurcation angles. These measurements can provide accurate and objective assessment of the vessel abnormity.  Studies using WESDR, ARIC data found an association between retinal vessel caliber and diabetic retinopathy, nephropathy and even early metabolic symptom.  We further conducted a pilot study to investigate the association between retinal vessel measurements and neuropathy severity. The statistics from our pilot study showed a U-shape change of the vessel caliber in the early stage of neuropathy. Interestingly, similar trend was found in a study investigating retinal vessel caliber and diabetic peripheral neuropathy using 608 patients from Malaysia eye study.

Conclusion(s):

Based on all the information, we proposed that retinal vessel measurement could be used as a promising tool for the early detection of diabetic neuropathy.