NIDEK CO., LTD., a global leader in the design, manufacture, and distribution of ophthalmic, optometric, and lens edging equipment recently announced the launch of the B-scan Denoising Software for the NIDEK OCT series.
NIDEK has been creating competitive products with our technological prowess by combining optical technology and electronic technology. At present, NIDEK is pursuing global business activities based on three fields, Ophthalmology & Optometry, Lens Edging, and Coating.
These activities encompass everything from development, manufacture, sales, and customer service to quickly respond to customer needs.
This software converts a B-scan retinal OCT image to a high definition image by a denoising technique using deep learning.
A large data set of images that are each averaged from 120 images are used as teaching data for deep learning to create clear retinal images comparable to images generated by averaging 120 images.
Different from image averaging which is an ordinal technique to create clear B-scan images, this unique denoising technique does not require capturing multi-frame images to remove speckle noise. This feature provides better patient flow and increased patient comfort.
Providing exquisite image quality, this software is particularly useful when capturing multiple B-scans for wide-area screening.
When a B-scan image is captured, a denoised image is automatically displayed by the denoising technique using deep learning.
A large data set of images that are each averaged from 120 images were used as teaching data for deep learning to create clear images comparable to images generated by averaging 120 images. A generative adversarial network, which is reported as an effective approach to improve image clarity, was adopted as the deep learning method.
The B-scan Denoising Software removes speckle noise from a single scan, producing an image that is as clear as an image generated by averaging 120 images. The denoising technique allows for easier image acquisition for obtaining high-quality images.