Abstract
Purpose: :
To present a method to reduce speckle noise and increase region contrast of optical coherence tomography (OCT) imaging through automatic registration of multiple B-scans.
Methods: :
Twenty young adult subjects each had fifty consecutive chorio-retinal B-scans obtained within one measurement session with a spectral domain OCT. Speckle noise, which is present in each of the single B-scans, significantly limits the information provided by OCT imaging by reducing the contrast between regions and thereby limiting the effective spatial resolution of the measurement. If the speckle is uncorrelated between measurements, then averaging consecutive measurements can reduce this problem. Despite the high acquisition rates of current OCT systems, small eye movements can misalign consecutive B-scans and therefore should be also compensated. Therefore to address these issues, each subject’s 50 B-scans were exported from the instrument and analysed with custom written software. This software registered and aligned each of the 50 B-scan images, and excluded any B-scans that did not exhibit a high level of alignment, in order to calculate an average B-scan image for each subject. The resulting ‘average’ images were compared to the original individual B-scan images using a range of speckle-reduction performance metrics including: signal to noise ratio (SNR), contrast-to-noise ratio (CNR), equivalent number of looks (ENL) and edge preservation.
Results: :
An average of 40±10 individual B-scans were used in each subject’s average B-scan image. For all 20 subjects, SNR, CNR and ENL metrics showed statistically significant improvement (p<0.01) in the averaged B-scan versus the single B-scan. The edge preservation did not show statistically significant changes (p>0.01), thus demonstrating the conservation of the original anatomical information after the method is applied.
Conclusions: :
The proposed automatic algorithm for registration and alignment of OCT B-scan images significantly improves the OCT image quality by enhancing the contrast between regions and chorio-retinal layer boundaries.
Keywords: image processing • imaging/image analysis: clinical • topography