A Multinoise Removal Algorithm for OCT Imaging

Benefiting from the high imaging resolution and deep penetration depth, Optical coherence tomography (OCT) is extensively applicable in ophthalmology, dermatology, and other clinical fields. However, the imaging quality is usually compromised by some noises such as horizontal coherence stripes, periodic background noise, and speckle noise. This paper proposes a multi-noise removal algorithm that combines spatial and transform-domain methods with optimized wavelet threshold denoising. This algorithm eliminates horizontal coherence stripes by generating a denoising mask through image segmentation and connected-domain filtering of superimposed B-scan images, utilizing the mask to remove these stripes. Besides, the periodic noise is removed by using frequency domain filters, while the speckle noise is also suppressed with the optimized wavelet threshold denoising method. We performed skin imaging using the SS-OCT system, processed the images, and evaluated the algorithm by quantifying the parameters such as signal-to-noise ratio, contrast-to-noise ratio, and equivalent number of looks. Results demonstrate that the proposed algorithm can effectively suppress multiple noises while retaining the original detailed information. This study offers an ideal solution for OCT image denoising, potentially extending its clinical applications.