Purpose: We investigated the utility of measuring vascular cloud (VC) to exclude poor-quality laser speckle flowgraphy (LSFG) images.
Subjects and methods: LSFG images were collected from healthy patients and patients with glaucoma. High-, medium-, and low-quality images were scored as 3,2, or 1, respectively, as image quality indices. Correlations with VC and other LSFG parameters were examined using Spearman rank correlation coefficient. Next, we classified the images into two groups: good (medium- or high-quality) and poor (low-quality) image quality groups. We analyzed the ability of VC to differentiate the two groups.
Results: For high-, medium-, and low-quality LSFG images, 44,28, and 23 images were collected from healthy eyes and 39,29, and 36 images were collected from eyes with glaucoma, respectively. In the two image quality groups, VC was positively correlated with image quality index (r=0.75, respectively, p< 0.001) and showed high area under the curve (AUC) (healthy eyes: cut-off value=0.24, AUC=0.90; eyes with glaucoma: cut-off value=0.28, AUC=0.91) for discriminating the poor image quality group. In the good image quality group, VC was lower in healthy eyes than in eyes with glaucoma (p< 0.05). In both groups, vessel-area mean blur rate was most strongly correlated with image quality index (r=0.65-0.78, p< 0.001). Patient age was higher in the poor image quality group than in the good image quality group (healthy eyes: 68.4±8.6 vs. 53.9±13.9 years, eyes with glaucoma: 72.7±8.9 vs. 62.0±8.6 years, p< 0.001).
Conclusion: VC is useful for differentiating LSFG images with poor image quality.
Nippon Ganka Gakkai Zasshi (J Jpn Ophthalmol Soc) 127: 549-556,2023.