Smart Neuropathy Detection using Machine Intelligence: Filling the Void Between Clinical Practice and Early Diagnosis

An important aspect of Smart healthcare involves the development of innovative non-intrusive procedures for early detection and diagnosis of specific illnesses. Essentially, diseases for which very complicated and invasive clinical and pathological tests are required could benefit from such innovations. Neuropathy or nerve loss is a characteristic of several commonly occurring longterm or terminal illnesses such as Diabetes, Multiple Sclerosis, Parkinson's disease, and Alzheimer. The detection of a degree of neuropathy is usually done through biopsy-type procedures in which tissue samples are collected from the body for further tests and visualization. In this paper, an innovative approach is presented by using one of the neuronal-representative organ in the body, the eye, to detect nerve loss. Specifically, small nerves are affected in the eye as the disease progresses. While this is currently a clinical practice in neurology and ophthalmology, it is still highly subjective and depends upon the clinician's perspective and experiences. The presented technique in this paper, attempts to make this procedure highly objective through translating human knowledge of neuropathy grading into a data-centric classification system using the corneal images through Confocal microscope. Clinicians use this common, non-invasive procedure for detecting a variety of anomalies in the eye. However, using the nerve-level resolution of the microscope, the corneal nerve canvas can be visualized and accordingly segmented and classified for possible neuropathy grading, which in turn, give the insight into the disease progression. The technique used in this paper is Adaptive Neuro Fuzzy Inference System (ANFIS) that translates the human understanding of the underlying pathology into numerical decisions that can be utilized as a tool into predicting and early diagnosis of the disease.