Neurocornea

Publications

Susana F. Silva et al., Diabetic peripheral neuropathy assessment through texture based analysis of corneal nerve images, J. Phys.: Conf. Ser. 616 012002, 2015. doi:10.1088/1742-6596/616/1/012002.
Susana F. Silva et al. Corneal Nerve Morphometry for Diabetic Peripheral Neuropathy Assessment, The International Conference on Health Informatics, IFMBE Proceedings 42, 296-299, 2014. doi: 10.1007/978-3-319-03005-0_75.
Otel, Iulian et al., Diabetic peripheral neuropathy assessment through corneal nerve morphometry, Proceedings of 2013 IEEE 3rd Portuguese Meeting in Bioengineering (ENBENG), 20-23 Feb. 2013 doi: 10.1109/ENBENG.2013.6518436.
Susana F. Silva et al., Evaluation of corneal nerves morphology for diabetic peripheral neuropathy assessment, Proceedings of ENBENG 2012 IEEE 2nd Portuguese Meeting in Bioengineering, DOI: 10.1109/ENBENG.2012.6331376
Ana Ferreira, António Miguel Morgado and José Silvestre Silva, A Method for Corneal Nerves Automatic Segmentation and Morphometric Analysis, Computer Methods and Programs in Biomedicine. 107: 53–60, 2012.
Ana Ferreira, António Miguel Morgado and José Silvestre Silva, Corneal nerves segmentation and morphometric parameters quantification for early detection of diabetic neuropathy, Proceedings of MEDICON 2010 – 12th Mediterranean Conference on Medical and Biological Engineering and Computing, Springer Verlag IFBME Proceedings Series. pp 264-267. DOI: 10.1007/978-3-642-13039-7_66 (2010).
Ana Ferreira, António Miguel Morgado and José Silvestre Silva, Automatic corneal nerves identification in confocal microscopy images for earlier diagnosis and follow-up of diabetic neuropathy, Proceedings da ICIAR 2010 – 7th International Conference on Image Analysis and Recognition, Springer Lecture Notes in Computer Science series, pp 60-69 (2010).

Assessment of diabetic peripheral neuropathy through morphometric analysis of corneal nerves imaged in vivo by corneal confocal microscopy
Funding: FCT - Project PTDC/SAU-BEB/104183/2008: 75 k€01/06/2010 - 30/11/2012

Initial steps of the segmentation algorithm: (a) original Corneal Confocal Microscopy (CCM) image, (b) local equalization applied to CCM image, and (c) phase-shift applied to CCM image.

Background

The prevalence of Diabetes Mellitus is dramatically increasing worldwide. Consequently, there will be a substantial increase in the prevalence of chronic complications associated with diabetes. Diabetic peripheral neuropathy (DPN) is the major cause of chronic disability in diabetic patients. DPN is implicated in 50-75% of non-traumatic amputations.

The early diagnosis and accurate assessment of DPN are important to define the risk patients, decrease patient morbidity and assess the performance of new therapies. However, this diagnosis often fails or occurs only when patients became symptomatic due to the non-availability of a simple non-invasive method for early diagnosis.

There is a pressing need for a non-invasive technique, capable of accurately documenting nerve damage and being used for early diagnosis of DPN. The technique should be simple and not require expensive equipment, in order to ease its dissemination and to extend its reach to highest possible number of diabetic patients.

Team

In collaboration with the Departments of Ophthalmology, Neurology and Endocrinology of Coimbra's University Hospital

Results

A method for automatic segmentation and analysis of corneal nerves from images obtained in vivo through corneal confocal microscopy was developed. Tests showed that the method is capable of segmenting corneal nerves, with sensitivity near 90% and an average percentage of false recognitions of 5.3%.

We were able to distinguish controls from diabetics, to differentiate individuals without DPN from those with DPN, and to stratify patients according to the Michigan Neuropathy Screening Instrument (MNSI) classification for the severity of neuropathy.

The parameters that have shown a good quantification and diagnostic potential performance were Nerve Fiber Length (NFL), Nerve Fiber Density (NFD), and Nerve Branch Density (NBD). NFL and NFD revealed significant mean value differences between and within the control and diabetic groups, with high degrees of specificity and sensitivity.

NFD and NFL are significantly lower in diabetics compared to controls (p<0.05), establishing inverse correlations with severity degree of DPN. Both parameters revealed significant differences between controls and diabetics, between no DPN cases and DPN patients, and within diabetics (MNSI mild and moderate groups) and healthy controls.

NFL, NFD and NBD parameters allowed to identify patients at initial phase of DPN (mild degree) that are at subsequent risk of developing clinically significant DPN (moderate and severe DPN). The mean difference values between diabetics with mild and moderate degree of DPN were minimal, showing that all four parameters are reliable predictors for neuropathic complications.

Page updated

Google Sites

Report abuse