Abstract
In this work we evaluated a postprocessing, customized automatic retinal OCT B-scan enhancement software for noise reduction, contrast enhancement and improved depth quality applicable to Heidelberg Engineering Spectralis OCT devices. A trained deep neural network was used to process images from an OCT dataset with ground truth biomarker gradings. Performance was assessed by the evaluation of two expert graders who evaluated image quality for B-scan with a clear preference for enhanced over original images. Objective measures such as SNR and noise estimation showed a significant improvement in quality. Presence grading of seven biomarkers IRF, SRF, ERM, Drusen, RPD, GA and iRORA resulted in similar intergrader agreement. Intergrader agreement was also compared with improvement in IRF and RPD, and disagreement in high variance biomarkers such as GA and iRORA.
Highlights
In this work we evaluated a postprocessing, customized automatic retinal OCT B-scan enhancement software for noise reduction, contrast enhancement and improved depth quality applicable to Heidelberg Engineering Spectralis OCT devices
Higher scores were achieved for the post processed, enhanced OCT images compared to the original images (p < 0.001) illustrates the improvement of perceived quality of up to 4 points in the enhanced images
In this paper we show that our proposed image enhancing algorithm significantly increases the image quality based on annotations from two graders on a large dataset of OCT B-scan slices
Summary
In this work we evaluated a postprocessing, customized automatic retinal OCT B-scan enhancement software for noise reduction, contrast enhancement and improved depth quality applicable to Heidelberg Engineering Spectralis OCT devices. We identify two main areas in which OCT denoising has been evaluated, the first one considers spatial denoising methods, where image enhancement happens either via local image filtering such as median[3] or mean Gaussian filters[4], or at global OCT volume scale. Strategies in spatial denoising considered using machine learning to target specific speckle noise distributions[11], but the results were limited by the need for large amounts of data to train neural networks. To solve this issue, Lethinen et al.[12] proposed a solution that in natural images that corrupted observations can be used to clean signals by including additional noise (i.e. Gaussian or Poisson). Deep learning methods have been successfully applied on reducing speckle noise of OCT images[13,14,15]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
