A Novel Quality Parameter for Stratus OCT

Abstract

Purpose:: To develop a more representative quality assessment parameter for optical coherence tomography (Stratus OCT; Carl Zeiss Meditec, Inc., Dublin, CA) and to compare it with the signal strength (SS), the current standard for quality assessment.

Methods:: Stratus OCT images with three fast scan types (macular, nerve fiber layer, and optic nerve head scans) were obtained. SS was calculated for each image using Stratus 4.0 software. Four OCT experts evaluated these images in a randomized fashion and labeled them as either acceptable or unacceptable images, and the majority opinion scores were recorded for each image. Three individual quality parameters quantified three different aspects of the OCT image: pixel intensity histogram pattern (PI), visible speckle noise density, and visible speckle noise intensity (SpD and SpI, respectively). Pearson’s correlation was used to compare SS with the PI, SpD, and SpI. To predict OCT experts’ assessment, an optimized formula utilizing three individual parameters was generated with a logistic regression analysis with interactions. Areas under the receiver operating characteristic curves (AROC) were computed to assess the performance of SS, individual parameters, and the combined parameter, using the OCT experts’ assessment.

Results:: 270 images of 90 subjects (30 each for normal, early, and advanced glaucoma) were enrolled in this study. Each parameter showed strong correlation with SS (R² 0.62 (PI), 0.59 (SpD), 0.37 (SpI), all p<0.0001, Pearson correlation). SpD and PI showed no significant difference in AROC in comparison with SS (SpD 0.83, PI 0.80, SS 0.83, p>0.35). AROC of SpI was significantly smaller than that of SS (0.72 vs 0.83, respectively, p=0.025). The combined paramter outperformed SS (0.88 vs 0.83, respectively, p=0.03) in the discrimination of acceptable versus unacceptable images. The addition of SS to the model did not improve the discriminating performance.

Conclusions:: The new combined quality parameter outperforms SS. This model may provide more robust way of assessing OCT image quality.

Clinical Trial:: www.clinicaltrials.gov NCT00343746

Keywords: image processing • imaging/image analysis: clinical • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)