ABSTRACT
Objective:To investigate the diagnostic value of an intelligent assisted grading algorithm for nuclear cataract using anterior segment optical coherence tomography (AS-OCT) images.Methods:A diagnostic test study was conducted.AS-OCT image data were collected from 939 cases of 1 608 eyes of nuclear cataract patients at the Shanghai Tenth People's Hospital of Tongji University from November 2020 to September 2021.The data were obtained from the electronic case system and met the requirements for clinical reading clarity.Among them, there were 398 cases of 664 male eyes and 541 cases of 944 female eyes.The ages of the patients ranged from 18 to 94 years, with a mean age of (65.7±18.6) years.The AS-OCT images were labelled manually from one to six levels according to the Lens Opacities Classification System Ⅲ (LOCS Ⅲ grading system) by three experienced clinicians.This study proposed a global-local cataract grading algorithm based on multi-level ranking, which contains five basic binary classification global local network (GL-Net).Each GL-Net aggregates multi-scale information, including the cataract nucleus region and original image, for nuclear cataract grading.Based on ablation test and model comparison test, the model's performance was evaluated using accuracy, precision, sensitivity, F1 and Kappa, and all results were cross-validated by five-fold.This study adhered to the Declaration of Helsinjki and was approrved by Shanghai Tenth People's Hospital of Tongji University (No.21K216).Results:The model achieved the results with an accuracy of 87.81%, precision of 88.88%, sensitivity of 88.33%, F1 of 88.51%, and Kappa of 85.22% on the cataract dataset.The ablation experiments demonstrated that ResNet18 combining local and global features for multi-level ranking classification improved the accuracy, recall, specificity, F1, and Kappa metrics.Compared with ResNet34, VGG16, Ranking-CNN, MRF-Net models, the performance index of this model were improved.Conclusions:The deep learning-based AS-OCT nuclear cataract image multi-level ranking classification algorithm demonstrates high accuracy in grading cataracts.This algorithm may help ophthalmologists in improving the diagnostic accuracy and efficiency of nuclear cataract.
ABSTRACT
Objective To construct an objective analysis system of corneal nerve tortuosity and detect the changes of corneal subbasal nerve tortuosity in patients with dry eye and diabetes. Methods GradeⅠtoⅣnerve tortuosity were evaluated and 80 photos of each grade were randomly chosen from the in vivo confocal microscopy library. Nerve fibers were extracted,segmented and then analyzed by 6 tortuosity related parameters including L C, Seg L C mean,Cur mean,Specific p,ICM and SCC mean. After verifying the validaty of parameters above,a cross-sectional study was conducted. Subjects were collected from June,2018 to February,2019 in Peking University Third Hospital,and were divided into healthy control group (28 persons 56 eyes),dry eye without diabetes group (28 patients 56 eyes),diabetes without dry eye group(24 patients 48 eyes),diabetes with dry eye group (23 patients 46 eyes) . Basic and dry eye information includes sex,age,ocular surface disease index ( OSDI) ,tear film break-up time (TBUT),Schirmer Ⅰ test (SⅠt) and corneal fluorescence staining (CFS) score. Fasting plasma glucose (FPG) and hemoglobin A1c (HbA1c) were detected in diabetic patients. Cochet-Bonnet examination (C-BE) was detected to evaluate corneal sensation and 2 corneal subbasal nerve photos of each eye were selected for effective tortuosity and density related parameters analysis. Data was analyzed by SPSS and diagnostic test were perfomed by MedCalc. This study followed the Declaration of Helsinki. This study protocol was approved by Ethic Committee of Peking University Third Hospital ( No. IRB00006761-M2017354 ) . Written informed consent was obtained from each subject prior to entering study cohort. Results L C,Seg L C mean,Cur mean,Specific p,ICM and SCC mean increased as the nerve tortuosity increased from Grade Ⅰ to Grade Ⅳ,with an overall significance among 4 groups (F=39. 100, 36. 367,57. 743,4. 043,6. 818,33. 493;all at P<0. 01). Among the above 6 parameters,Cur mean and L C of any two groups were of significant difference (all at P<0. 01). Twenty three to twenty eight persons were enrolled in each group of the cross-sectional study. Sex and age were comparable among 4 groups. Diagnostic criteria were met in dry eye and diabetes. Corneal sensation parameter C-BE decreased in diabetes without dry eye group and diabetes with dry eye group compared with healthy control group ( all at Adj P<0. 05 ) , other than in dry eye without diabetes group (AdjP≥0. 05). Nerve density of diabetes without dry eye group and diabetes with dry eye group was lower compared with healthy control group(all at P<0. 001),while no significant difference between dry eye without diabetes group and healthy control group(P≥0. 05). Among the effective parameters of tortuosity,L C,Cur mean,Seg L C mean and SCC mean of dry eye without diabetes group,diabetes without dry eye group,diabetes with dry eye group were higher compared with healthy control group ( all at P<0. 05 ) . Diagnostic tests of tortuosity related parameters all showed an area under curve (AUC) from 0. 5 to 0. 7. Conclusions L C and Cur mean can be used to analyze corneal nerve curvature more reliably. Compared with normal volunteers,patients of dry eye or diabetes show higher corneal subbasal nerve tortuosity.