A pilot study included 12 randomly selected type-2 diabetic patients followed at the Department of Endocrinology of the Coimbra University Hospitals, and 8 age-matched non-diabetic control individuals. Diabetic patients were classified in 3 groups, according to the severity of neuropathy (absent, mild, moderate).
All subjects were assessed with surface electromyography (EMG) for evaluating the severity and monitoring significant injuries and progressions of DPN. All measurements were performed at the Department of Neurology of Coimbra University Hospitals. EMG measurements comprised nerve conduction evaluation, assessing motor (peroneal) and sensory (sural) NCV and amplitudes, as well as cutaneous sympathetic response measurement.
Corneal confocal microscopy (CCM) images were obtained at the Department of Ophthalmology of the Coimbra University Hospitals.
Images were analyzed by our automatic algorithms and also through manual segmentation of the nerves. The following parameters were measured: NFD, nerve width (NFW), NFL, nerve tortuosity (TC), NBD, NBP and nerve angle (NBA).
The study determined that CCM can be successfully used as a complementing technique for the clinical and electrophysiological diagnosis of DPN using nerve conduction studies (NCS).
NCS examinations have validated that DPN is mainly axonal with a predominantly large diameter nerve fiber involvement. CCM quantified small fiber disorders (gradual reduction in NFD, NBD and NFL), showing that DPN also occur with mixed large - and small diameter nerve fiber involvement. We were able to distinguish the controls from diabetics, to differentiate individuals without DPN from those with DPN, and most important, to stratify the patients according to the severity of neuropathy.
We proposed an innovative approach for the assessment of DPN using CCM images: using image statistics, without nerve segmentation, for detecting and staging DPN patients. To demonstrate the concept we analyzed the images obtained during the pilot clinical study using texture analysis, based on image statistics obtained from the Gray Level Co-occurrence Matrix (GLCM) and from the Gray Level Run Length Matrix (GLRLM), and classification techniques.
Cornea Module for Slit Lamp
Automatic Corneal Nerves Segmentation Algorithm
Pilot trial with diabetic neuropathy patients