Iterative Variance Stabilizing Transformation Denoising of Spectral Domain Optical Coherence Tomography Images Applied to Retinoblastoma

Background: Due to the presence of speckle Poisson noise, the interpretation of spectral domain-optical coherence tomography (SD-OCT) images frequently requires the use of data averaging to improve the signal-to-noise ratio. This implies long acquisition times and requires patient sedation in some cases. Iterative variance stabilizing transformation (VST) is a possible approach by which to remove speckle Poisson noise on single images. Methods: We used SD-OCT images of human and murine (LH Beta-Tag mouse model) retinas with and without retinoblastoma acquired with 2 different imaging devices (Bioptigen and Micron IV). These images were processed using a denoising workflow implemented in Matlab. Results: We demonstrated the presence of speckle Poisson noise, which can be removed by a VST-based approach. This approach is robust as it works in all used imaging devices and in both human and mouse retinas, independently of the tumor status. The implemented algorithm is freely available from the authors on demand. Conclusions: On a single denoised image, the proposed method provides results similar to those expected from the SD-OCT averaging. Because of the friendly user interface, it can be easily used by clinicians and researchers in ophthalmology.