ESCRS - PO0248 - The Assessment Of Relationship Between Front And Back Nuclear Lens Features Of As-Oct Imaging With The Locs Iii Criteria

The Assessment Of Relationship Between Front And Back Nuclear Lens Features Of As-Oct Imaging With The Locs Iii Criteria

Published 2023 - 41st Congress of the ESCRS

Reference: PO0248 | DOI: 10.82333/wa2w-m625

Authors: Lingxi Hu 1 , Jiang Liu* 2 , Risa Higashita 3 , Zunjie Xiao 4 , Xiaoling Li 5 , Jin Yuan 6 , Chen Lin 7 , Yepang Liu 4 , Menglan Zhou 8

1Research Institute of Trustworthy Autonomous Systems,Southern university of Science and Technology,Shenzhen,China;School of Computer Science,University of Birmingham,Birmingham,United Kingdom, 2Research Institute of Trustworthy Autonomous Systems,Southern university of Science and Technology,Shenzhen,China;Computer Science and Engineering,Southern university of Science and Technology,Shenzhen,China, 3Research Institute of Trustworthy Autonomous Systems,Southern university of Science and Technology,Shenzhen,China;Tomey Corporation,Nagoya,Japan, 4Computer Science and Engineering,Southern university of Science and Technology,Shenzhen,China, 5School of Ophthalmology and Optometry,Wenzhou Medical University,Wenzhou,China, 6State Key Laboratory of Ophthalmology,Sun Yat-sen University,Guangzhou,China, 7Ophthalmology,Shenzhen People's Hospital,Shenzhen,China, 8Chengdu E Area Aier Ophthalm. Hosp.,Chengdu,China

To investigate the quantitative image attribute-based features of the front and back nuclear regions in lens AS-OCT imaging and to explore their association with LOCS III grading levels.

The dataset was collected at Zhong-shan Ophthalmic Center, Zhongshan, China

This was a cross sectional study performed in 451 eyes of 308 patients. All patients underwent AS-OCT(Tomey, Japan) and slit-lamp examination with LOCS III grading. Patients with Corneal disease or cortical cataract were excluded. The nuclear lens in AS-OCT was bisected along the middle axis into front and back regions. Image attribute-based features was extracted, where pixel_mean indicates an average opacity of the target area, and pixel_entropy indicates the distribution randomness. The correlation between extracted features and LOCS III were examined, logistic regression analysis was performed to investigate whether the features were associated with the increasing LOCS III level. Statistically significant was considered at P < 0.05.

The image attribute-based features showed significant correlation with LOCS III grading, the front region pixel_entropy is slightly higher(r=0.79 p<0.05) correlated with LOCS III grading than other regions(front_mean/ back_mean/back_entropy, r=0.72/0.71/0.76 p<0.05). Logistic regression suggests a unit increase in pixcel_mean would increase the odds of the LOCS III level by a factor of around 0.84(front/back 0.84/0.84). However, higher pixcel_entropy does not have such evident effect on the odds of LOCS III level (front/back 3.97e-4/1.61e-4). Similar performance was shown in the parameters taken from whole lens, where an increase in whole_mean significantly correlates with LOCS III and increase the odds of severity by 0.84 identically.

We found a greater extent of pixel_mean of nuclear region lens measured by AS-OCT, was significantly associated with a greater odds of higher LOCS III grading, where sensitivity of front and back region measurements are identical. This can in some extent provide more comprehensive evidence for objectifying the cataractous grade, and further extend the use of AS-OCT in preoperative planning.