Impact of B-Scan Averaging on Spectralis Optical Coherence Tomography Image Quality before and after Cataract Surgery

Figure 3

Quantitative measures of image quality. The graphs show signal-to-noise ratio (SNR; (a)), the Cohen’s value (b), and intraretinal layer segmentation performance (c). SNR and Cohen’s are computed relative to the baseline at two averaged frames. The SNR compares the level of a desired signal in the image to the level of noise. Cohen’s value describes the dissimilarity of individual retinal layers based on their grey values and is inversely proportional to the overlap of distributions. It is computed for two neighboring layers. The ability to differentiate between the individual retinal layers based on their appearance was further assessed by measuring the performance of an automated computer algorithm on correctly segmenting each layer based on pixel classification. Cohen’s value and the automated layer segmentation performance are demonstrated as relative values. Results are shown as mean for each averaging frame increment, separately for pre- and postoperative results overall patients and all layers.

(a)

(b)

(c)