Qianzhong Cao

Localization and diagnosis framework for pediatric cataracts based on slit-lamp images using deep features of a convolutional neural network

Slit-lamp images play an essential role for diagnosis of pediatric cataracts. We present a computer vision-based framework for the automatic localization and diagnosis of slit-lamp images by identifying the lens region of interest (ROI) and employing a deep learning convolutional neural network (CNN). First, three grading degrees for slit-lamp images are proposed in conjunction with three leading ophthalmologists. The lens ROI is located in an automated manner in the original image using two successive applications of Candy detection and the Hough transform, which are cropped, resized to a fixed size and used to form pediatric cataract datasets. These datasets are fed into the CNN to extract high-level features and implement automatic classification and grading. To demonstrate the performance and effectiveness of the deep features extracted in the CNN, we investigate the features combined with support vector machine (SVM) and softmax classifier and compare these with the traditional representative methods. The qualitative and quantitative experimental results demonstrate that our proposed method offers exceptional mean accuracy, sensitivity and specificity: classification (97.07%, 97.28%, and 96.83%) and a three-degree grading area (89.02%, 86.63%, and 90.75%), density (92.68%, 91.05%, and 93.94%) and location (89.28%, 82.70%, and 93.08%). Finally, we developed and deployed a potential automatic diagnostic software for ophthalmologists and patients in clinical applications to implement the validated model.

A Novel Congenital Cataract Category System Based on Lens Opacity Locations and Relevant Anterior Segment Characteristics.

Purpose We compared the anterior segment characteristics of congenital cataract (CC) patients with lens opacities in different locations and proposed a modified, simple CC category system. Methods Cataractous eyes of CC patients were classified into four groups based on the locations of lens opacities shown in slit-lamp examinations and by a 3-dimensional anterior segment imaging system as follows: total, anterior, interior, and posterior cataracts. The mean keratometry value, corneal astigmatism (CA), central corneal thickness (CCT), and anterior chamber depth (ACD) of eyes in different groups were compared. Results We included a total of 428 CC patients. Half of the patients with an anterior cataract had the complication of a pupillary residual membrane. Among the patients with posterior lentiglobus cataracts, 90.38% had unilateral involvement. Patients with total, anterior, or interior cataracts had larger keratometry values than those with either posterior cataracts or clear lens. Congenital cataract patients had greater CA and CCT values than those with a clear lens. The largest CA was presented in patients with anterior cataracts, and the value decreased gradually with more posterior locations of lens opacities. Eyes with total and anterior cataracts had smaller ACDs, and eyes with interior and posterior cataracts had greater ACDs than eyes with a clear lens. Conclusions Cataractous eyes in CC patients with lens opacities in different locations presented distinct anterior segment characteristics. The modified CC category system, based on the relationships among the locations of lens opacities and anterior segment characteristics, may be beneficial for CC diagnosis and treatment.

Comparative analysis of image classification methods for automatic diagnosis of ophthalmic images

There are many image classification methods, but it remains unclear which methods are most helpful for analyzing and intelligently identifying ophthalmic images. We select representative slit-lamp images which show the complexity of ocular images as research material to compare image classification algorithms for diagnosing ophthalmic diseases. To facilitate this study, some feature extraction algorithms and classifiers are combined to automatic diagnose pediatric cataract with same dataset and then their performance are compared using multiple criteria. This comparative study reveals the general characteristics of the existing methods for automatic identification of ophthalmic images and provides new insights into the strengths and shortcomings of these methods. The relevant methods (local binary pattern +SVMs, wavelet transformation +SVMs) which achieve an average accuracy of 87% and can be adopted in specific situations to aid doctors in preliminarily disease screening. Furthermore, some methods requiring fewer computational resources and less time could be applied in remote places or mobile devices to assist individuals in understanding the condition of their body. In addition, it would be helpful to accelerate the development of innovative approaches and to apply these methods to assist doctors in diagnosing ophthalmic disease.

Timing and approaches in congenital cataract surgery: a randomised controlled trial

BACKGROUND The optimum timing of surgery and the approach that should be used during congenital cataract surgery are controversial and present challenges for paediatric cataract surgeons in China. The aim of this study was to compare visual prognoses and postoperative adverse effects of paediatric cataract surgery done at different times and using different surgical approaches. METHODS In this prospective, randomised controlled trial, we recruited Chinese patients aged 3 months or younger who were diagnosed with bilateral diffuse or total cataracts, but who did not have coexisting ocular, systemic or neurological diseases. Patients were randomly assigned into two surgical timing groups: a 3-month-old and a 6-month-old. Each eye had one of three randomly assigned surgical procedures: lens aspiration, lens aspiration with posterior continuous curvilinear capsulorhexis, and lens aspiration with posterior continuous curvilinear capsulorhexis and anterior vitrectomy. Long-term visual acuity and incidence of postoperative secondary glaucoma and severe posterior capsular opacification were the primary outcomes. χ2 tests, Fishers exact tests and Kruskal-Wallis tests were used to compare the outcomes between groups. Generalised Estimating Equation with a working covariance matrix of unstructured correlations was integrated into the multivariate linear regression to explore the relationship between long-term visual acuity and the incidence of complications. The research protocol was approved by the Institutional Review Board and Ethics Committee of Sun Yat-sen University. Informed written consent was obtained from at least one family member of each participating patient, and the study was done according to the Declaration of Helsinki. This trial is registered with ClinicalTrials.gov, number NCT02581046. FINDINGS We recruited 61 patients (30 girls, 31 boys) to our study. 4 children were excluded from the final analysis because their parents declined participation. A total of 57 participants (114 eyes) with a mean follow-up of 48·7 months were included in the final analysis. Long-term visual acuity in patients who underwent surgery aged 6 months (0·81 [SD 0·28]) was better than that of patients who had surgery aged 3 months (0·96 [0·30], p=0·02) and incidence of posterior capsular opacification during the 6 months following surgery was positively correlated with long-term visual acuity (β=-0·080, p=0·036). Two (2/95, 2%) eyes with a confirmed diagnosis of secondary glaucoma were in the 3-month-old group, and additional surgical interventions were done in these patients to lower the intraocular pressure. For patients who had surgery aged 6 months, the incidence of severe posterior capsular opacification was significantly higher in the lens aspiration group than other two groups (lens aspiration, 8/15 [53%]; lens aspiration with posterior continuous curvilinear capsulorhexis, 3/21 [14%]; and lens aspiration with posterior continuous curvilinear capsulorhexis and anterior vitrectomy, 3/16 [19%]; p=0·03). INTERPRETATION It was safer and more effective for Chinese bilateral diffuse congenital cataract patients to have surgery of lens aspiration with posterior continuous curvilinear capsulorhexis with or without anterior vitrectomy at 6 months of age than at 3 months of age. FUNDING National Basic Research Program (973; 2015CB964600); Natural Science Foundation of Guangzhou (91546101, 81300750); Clinical Research and Translational Medical Center for Pediatric Cataract in Guangzhou, China.

Visual Outcome and Related Factors in Bilateral Total Congenital Cataract Patients: A Prospective Cohort Study.

This study is to evaluate the visual outcome and identify its crucial related factors in children undergoing cataract surgery for bilateral total congenital cataract (CC). This prospective study included consecutive bilateral total cataract patients undergoing primary surgery at Zhongshan Ophthalmic Center (ZOC), Guangzhou, China from Jan 2010 to May 2014. Visual outcome was estimated by best-corrected visual acuity (BCVA) at last follow-up. Potential related factors, including gender, age at last follow-up, age at primary surgery, surgical procedure, postoperative complications (PCs), frequency of follow-up and changes in spectacles were evaluated. Eighty-eight children (176 eyes) were included in the cohort. The mean post-operative BCVA (logMAR) was 1.07 ± 0.53 at the mean follow-up duration 31.07 ± 19.36 months. Multivariable generalized estimating equations (GEEs) showed BCVA was significantly associated with PCs, age at last follow-up and age at primary surgery. Partial correlation analysis indicated age at primary surgery was positively correlated with BCVA controlling for the other factors, both for the whole age range (R = 0.415, P < 0.001) and age >6 months (R = 0.867, P < 0.001). Better visual acuity was related to early primary surgery and low PC occurrence in children with bilateral total CC. Timely surgical intervention and strict control of PCs would be potential steps to achieving better visual outcome.

Automatic diagnosis of imbalanced ophthalmic images using a cost-sensitive deep convolutional neural network

BackgroundOcular images play an essential role in ophthalmological diagnoses. Having an imbalanced dataset is an inevitable issue in automated ocular diseases diagnosis; the scarcity of positive samples always tends to result in the misdiagnosis of severe patients during the classification task. Exploring an effective computer-aided diagnostic method to deal with imbalanced ophthalmological dataset is crucial.MethodsIn this paper, we develop an effective cost-sensitive deep residual convolutional neural network (CS-ResCNN) classifier to diagnose ophthalmic diseases using retro-illumination images. First, the regions of interest (crystalline lens) are automatically identified via twice-applied Canny detection and Hough transformation. Then, the localized zones are fed into the CS-ResCNN to extract high-level features for subsequent use in automatic diagnosis. Second, the impacts of cost factors on the CS-ResCNN are further analyzed using a grid-search procedure to verify that our proposed system is robust and efficient.ResultsQualitative analyses and quantitative experimental results demonstrate that our proposed method outperforms other conventional approaches and offers exceptional mean accuracy (92.24%), specificity (93.19%), sensitivity (89.66%) and AUC (97.11%) results. Moreover, the sensitivity of the CS-ResCNN is enhanced by over 13.6% compared to the native CNN method.ConclusionOur study provides a practical strategy for addressing imbalanced ophthalmological datasets and has the potential to be applied to other medical images. The developed and deployed CS-ResCNN could serve as computer-aided diagnosis software for ophthalmologists in clinical application.

Developmental profile of ocular refraction in patients with congenital cataract: a prospective cohort study.

BACKGROUND Refractive status is crucial to visual function and can change throughout life. For patients with congenital cataract, the developmental profile of refraction is complicated owing to the potential influence of diverse clinical manifestations and various treatments, and has not yet been fully characterised. METHODS In this prospective cohort study, we enrolled patients with congenital cataract from the Childhood Cataract Program of the Chinese Ministry of Health (CCPMOH). Two experienced cataract surgeons (YL and WC) did cataract removal or intraocular lens (IOL) implantation at Zhongshan Ophthalmic Center. Patients were followed up at 1 week, 1 month, and 3 months postoperatively, then every 3 months thereafter. Follow-up refraction assessments were conducted with objective retinoscopy and cycloplegia, and were performed by experienced independent optometrists. We compared refraction status (presented as spherical equivalent), and the rate of myopic shift between bilateral and unilateral congenital cataract with linear mixed model (LMM) analysis, then compared the range of myopic shift between bilateral and unilateral congenital cataract with multivariate linear regression Ethical approval was provided by the institutional review board of the Zhongshan Ophthalmic Center (Guangzhou, China). Parents of participating children provided written informed consent. FINDINGS Between Jan 1, 2010, and Oct 31, 2013, we enrolled 1258 participants, all of whom underwent cataract removal and/or intraocular lens (IOL) implantation. By study completion on Oct 31, 2015, 1164 (93%) participants had attended at least one follow-up visit and had refraction status recorded (6585 follow-ups total). Median follow-up was 3 years (IQR 2·5-4·5 years). Spherical equivalent measurements decreased by 1·36 D (SD 0·06) per year between 0·5 years and 4 years of age in bilateral and unilateral congenital cataract aphakic patients. The spherical equivalent of unilateral aphakia was 0·67 D (SD 0·31) smaller than that of bilateral aphakia after correcting for age and sex (p=0·0325). Spherical equivalents in congenital cataract pseudophakia decreased by 0·28 D (SD 0·02) between 1·5 years and 18 years of age (p<0·0001). Spherical equivalents in unilateral pseudophakia did not differ from bilateral pseudophakia (p=0·2996). With the current strategy for IOL calculation, the spherical equivalent in congenital cataract pseudophakia reached emmetropia around 6 years of age, while the healthy eyes of unilateral congenital cataract reached emmetropia at 9-10 years. The rate of myopic shift peaked in both young childhood and early adolescence in all participants. The range of myopic shift in the unilaterally congenital cataract affected eyes was correlated with age at refraction measurement, cataract removal, and IOL implantation, as well as with sex. INTERPRETATION Factors that dictate the range of myopic shift in unilateral congenital cataract seem to be different from those in bilateral congenital cataract, indicating differences in the aetiopathogenesis and refractive prognosis between bilateral and unilateral congenital cataract. Refraction data from patients with congenital cataract are of clinical significance to the guidelines for congenital cataract treatment, especially for IOL calculation concerning the laterality of congenital cataract. FUNDING National Natural Science Foundation of China (number 81300750), the Ministry of Science and Technology of China Grants (973 program, 2015CB964600), the Key Research Plan for the National Natural Science Foundation of China (number 91546101) and the Guangdong Provincial Natural Science Foundation for Distinguished Young Scholars of China (number 2014A030306030).

Construction and implications of structural equation modeling network for pediatric cataract: a data mining research of rare diseases

BackgroundThe majority of rare diseases are complex diseases caused by a combination of multiple morbigenous factors. However, uncovering the complex etiology and pathogenesis of rare diseases is difficult due to limited clinical resources and conventional statistical methods. This study aims to investigate the interrelationship and the effectiveness of potential factors of pediatric cataract, for the exploration of data mining strategy in the scenarios of rare diseases.MethodsWe established a pilot rare disease specialized care center to systematically record all information and the entire treatment process of pediatric cataract patients. These clinical records contain the medical history, multiple structural indices, and comprehensive functional metrics. A two-layer structural equation model network was applied, and eight potential factors were filtered and included in the final modeling.ResultsFour risk factors (area, density, location, and abnormal pregnancy experience) and four beneficial factors (axis length, uncorrected visual acuity, intraocular pressure, and age at diagnosis) were identified. Quantifiable results suggested that abnormal pregnancy history may be the principle risk factor among medical history for pediatric cataracts. Moreover, axis length, density, uncorrected visual acuity and age at diagnosis served as the dominant factors and should be emphasized in regular clinical practice.ConclusionsThis study proposes a generalized evidence-based pattern for rare and complex disease data mining, provides new insights and clinical implications on pediatric cataract, and promotes rare-disease research and prevention to benefit patients.

Proteomics analysis and proteogenomic characterization of different physiopathological human lenses

BackgroundThe aim of the present study was to identify the proteomic differences among human lenses in different physiopathological states and to screen for susceptibility genes/proteins via proteogenomic characterization.MethodsThe total proteomes identified across the regenerative lens with secondary cataract (RLSC), congenital cataract (CC) and age-related cataract (ARC) groups were compared to those of normal lenses using isobaric tagging for relative and absolute protein quantification (iTRAQ). The up-regulated proteins between the groups were subjected to biological analysis. Whole exome sequencing (WES) was performed to detect genetic variations.ResultsThe most complete human lens proteome to date, which consisted of 1251 proteins, including 55.2% previously unreported proteins, was identified across the experimental groups. Bioinformatics functional annotation revealed the common involvement of cellular metabolic processes, immune responses and protein folding disturbances among the groups. RLSC-over-expressed proteins were characteristically enriched in the intracellular immunological signal transduction pathways. The CC groups featured biological processes relating to gene expression and vascular endothelial growth factor (VEGF) signaling transduction, whereas the molecular functions corresponding to external stress were specific to the ARC groups. Combined with WES, the proteogenomic characterization narrowed the list to 16 candidate causal molecules.ConclusionsThese findings revealed common final pathways with diverse upstream regulation of cataractogenesis in different physiopathological states. This proteogenomic characterization shows translational potential for detecting susceptibility genes/proteins in precision medicine.

Impairments of Visual Function and Ocular Structure in Patients With Unilateral Posterior Lens Opacity

Purpose We investigate visual function impairment and ocular structure in patients with unilateral posterior lens opacity, a type of congenital cataract (CC) in our novel CC category system. Methods We studied patients aged 3 to 15 years who were diagnosed with unilateral posterior CC. Best corrected visual acuity (BCVA) and visual evoked potentials (VEP) were examined. Corneal astigmatism (CA), mean keratometry, central corneal thickness, anterior chamber depth (ACD), and axial length were measured by Pentacam and IOL-Master. Variations between two eyes were compared by paired t-tests. Results Among the 25 patients involved, BCVAs (logMAR) of cataractous and contralateral healthy eyes were 0.8 ± 0.4 (range, 0.1–1.7) and 0.1 ± 0.1 (range, −0.1 to 0.4). Compared to contralateral healthy eyes, larger CA (1.8 ± 1.2 vs. 0.9 ± 0.4 diopters [D], P = 0.002) and deeper ACD (3.7 ± 0.3 vs. 3.5 ± 0.4 mm, P = 0.009) were found in cataractous eyes. No significant positive or negative linear relationship was found between BCVA and parameters of VEP. Peak time of P100 of pattern VEP-60′ in cataractous eyes was longer than that in contralateral healthy eyes (114.9 ± 18.8 vs. 105.0 ± 12.4 ms, P = 0.013). Amplitudes of P100 of patterns VEP-60′ and -15′ in cataractous eyes were smaller than those in contralateral healthy eyes (PVEP-60′, 15.2 ± 5.3 vs. 19.9 ± 10.4 μV, P = 0.023; PVEP-15′, 10.4 ± 7.0 vs. 22.1 ± 11.9 μV, P = 0.012). Conclusions Impaired visual function and ocular structure were detected in patients with posterior lens opacities. Translational Relevance This study provides evidence-based clinical recommendations for unilateral posterior CC patients with controversial treatment options.