Automated malarial retinopathy detection using transfer learning and multi-camera retinal images.

Cerebral malaria (CM) is a fatal syndrome found commonly in children less than 5 years old in Sub-saharan Africa and Asia. The retinal signs associated with CM are known as malarial retinopathy (MR), and they include highly specific retinal lesions such as whitening and hemorrhages. Detecting these lesions allows the detection of CM with high specificity. Up to 23% of CM, patients are over-diagnosed due to the presence of clinical symptoms also related to pneumonia, meningitis, or others. Therefore, patients go untreated for these pathologies, resulting in death or neurological disability. It is essential to have a low-cost and high-specificity diagnostic technique for CM detection, for which We developed a method based on transfer learning (TL). Models pre-trained with TL select the good quality retinal images, which are fed into another TL model to detect CM. This approach shows a 96% specificity with low-cost retinal cameras.

Automatic estimation of corneal nerves focused tortuosities.

The correlation between corneal nerve tortuosity and pathology has been shown multiple times. However, because there isn't any defacto definition of tortuosity, reproducibility is poor. Indeed, many studies still rely on the manual observation and judgment of tortuosity. Recently, two distinct forms of corneal nerve tortuosity have been identified, describing either short-range or long-range directional changes. In this study we were able to develop automatic corneal nerve tortuosity measurements that correctly and independently represent these two tortuosity definitions. Furthermore, we show that a combination of mathematical tortuosity measurements improves on single metric results.

A fully-automatic fast segmentation of the sub-basal layer nerves in corneal images.

Corneal nerves changes have been linked to damage caused by surgical interventions or prolonged contact lens wear. Furthermore nerve tortuosity has been shown to correlate with the severity of diabetic neuropathy. For these reasons there has been an increasing interest on the analysis of these structures. In this work we propose a novel, robust, and fast fully automatic algorithm capable of tracing the sub-basal plexus nerves from human corneal confocal images. We resort to logGabor filters and support vector machines to trace the corneal nerves. The proposed algorithm traced most of the corneal nerves correctly (sensitivity of 0.88 ± 0.06 and false discovery rate of 0.08 ± 0.06). The displayed performance is comparable to a human grader. We believe that the achieved processing time (0.661 ± 0.07 s) and tracing quality are major advantages for the daily clinical practice.