Retina

Advanced retinal imaging: Diagnostics, prognosis and outcomes

The technology in diagnostic imaging of retinal disease has impressively improved during the last years switching from 6 radial scans only with the earliest time domain optical coherence tomography (TD-OCT) to three-dimensional full-volume raster scanning by spectral domain (SD) OCT and now thousands of scans being taken in a short time by swept-source (SS-) OCT. This explosion of available information can only be adequately exploited by corresponding digital analysis tools being able to extract all relevant information from big data sets. Algorithmic methods are being used for computational analysis of OCT images including noise removal, motion correction, segmentation of individual retinal layers, quantification of intraretinal (IRF) and subretinal (SRF) fluid as well as pigment epithelial detachment (PED) and drusen. A comprehensive set of such automated registration methods enables longitudinal and inter-patient alignment of OCT scans and precise monitoring of progression or resolution. In addition, machine learning methodology can be used to identify a large spectrum of clinically relevant biomarkers for disease progression or therapeutic response.

Indications for advanced image analysis are unlimited ranging from monitoring of progression of early AMD in respect to drusen volume progression or regression, a decrease in IRF or SRF in diabetic macula edema (DME) as well as monitoring of progressive degeneration of neurosensory elements in diabetic retinal disease (DRP). Neovascular as well as atrophic AMD are important fields for computational analysis to even predict treatment response and the frequency of disease recurrence.

Normal morphologic features of relevance are subretinal hyperreflective material (SHRM) and fibrosis which may be detected by using selective OCT imaging tools such as polarization-sensitive (PS) OCT. Scientifically, novel targets for innovative therapeutic strategies can be identified and insight into the pathophysiology of retinal disease is greatly enhanced. Moreover, current and future disease management efforts require efficient risk prediction allowing cohort enrichment and streamlining of clinical trials.