A deep network DeepOpacityNet for detection of cataracts from color fundus photographs.

BACKGROUND: Cataract diagnosis typically requires in-person evaluation by an ophthalmologist. However, color fundus photography (CFP) is widely performed outside ophthalmology clinics, which could be exploited to increase the accessibility of cataract screening by automated detection. METHODS: DeepOpacityNet was developed to detect cataracts from CFP and highlight the most relevant CFP features associated with cataracts. We used 17,514 CFPs from 2573 AREDS2 participants curated from the Age-Related Eye Diseases Study 2 (AREDS2) dataset, of which 8681 CFPs were labeled with cataracts. The ground truth labels were transferred from slit-lamp examination of nuclear cataracts and reading center grading of anterior segment photographs for cortical and posterior subcapsular cataracts. DeepOpacityNet was internally validated on an independent test set (20%), compared to three ophthalmologists on a subset of the test set (100 CFPs), externally validated on three datasets obtained from the Singapore Epidemiology of Eye Diseases study (SEED), and visualized to highlight important features. RESULTS: Internally, DeepOpacityNet achieved a superior accuracy of 0.66 (95% confidence interval (CI): 0.64-0.68) and an area under the curve (AUC) of 0.72 (95% CI: 0.70-0.74), compared to that of other state-of-the-art methods. DeepOpacityNet achieved an accuracy of 0.75, compared to an accuracy of 0.67 for the ophthalmologist with the highest performance. Externally, DeepOpacityNet achieved AUC scores of 0.86, 0.88, and 0.89 on SEED datasets, demonstrating the generalizability of our proposed method. Visualizations show that the visibility of blood vessels could be characteristic of cataract absence while blurred regions could be characteristic of cataract presence. CONCLUSIONS: DeepOpacityNet could detect cataracts from CFPs in AREDS2 with performance superior to that of ophthalmologists and generate interpretable results. The code and models are available at https://github.com/ncbi/DeepOpacityNet ( https://doi.org/10.5281/zenodo.10127002 ).

Cataracts are cloudy areas in the eye that impact sight. Diagnosis typically requires in-person evaluation by an ophthalmologist. In this study, a computer program was developed that can identify cataracts from specialist photographs of the eye. The computer program successfully identified cataracts and was better able to identify these than ophthalmologists. This computer program could be introduced to improve the diagnosis of cataracts in eye clinics.

Can we offer additional BCG therapy for three-month BCG refractory high grade/T1, Tis bladder cancer patients?

Background: We lack tools to predict treatment and survival outcomes in patients receiving additional BCG therapy as a bladder-preserving therapy in high grade/T1, Tis NMIBC patients who showed persistent/recurrent tumors at three-month follow-up. Objectives: To assess the predictors of additional BCG response in patients who experienced persistent/recurrent tumors at three-month follow-up after BCG induction. Patients and methods: We retrospectively analyzed database for NMIBC. Between 2000 and 2019, 231 patients with high-grade T1/Tis NMIBC showed persistent/recurrent tumors at 3-month after BCG-induction, refused or were unfit to radical cystectomy (RC) and were offered additional intravesical BCG as bladder-preserving treatment. Predictors of the outcome after additional BCG were studied using univariate and multivariate logistic regression analysis. Kaplan Meier curve was utilized to estimate the recurrence-free survival (RFS) and progression-free survival (PFS). COX regression analysis was performed to identify independent predictors or RFS and PFS. Results: During a median (range) of 148 (24-224) months, poor response to additional BCG (tumor recurrence and/or progression) was noted in 112 (48.5%) patients. On multivariate logistic regression analysis, 3-month tumor features (persistent T stage, persistent grade and persistent/new CIS) significantly predicted poor response to additional BCG (OR: 3.4, 95%CI: 1.3-10.8, p = 0.021, OR: 2.1, 95%CI: 1.1-4.1, p = 0.02 and OR: 16.6, 95%CI: 4.5-109, p=<0.001, respectively). The mean RFS was 26 (9-152) months with identified 3-month tumor features (persistent T stage and persistent/new CIS) as independent predictors of RFS (HR = 11.5, 95%CI = 2.7-48.3, p = 0.001 and HR = 2.5, 95%CI = 1.5-4.1, p=<0.001, respectively) on multivariate COX regression analysis. In addition, 3-month tumor features (persistent/new CIS, non-papillary shape and bladder neck involvement) were identified to significantly predict PFS (HR = 6.2, 95%CI = 3.4-11.5, p=<0.001 and HR = 2.3, 95%CI = 1.3-4.3 p = 0.001 and HR = 2.1, 95%CI = 1.2-3.8, p=<0.005, respectively). Conclusions: Three-month tumor features could be utilized as a tool to predict treatment outcomes and survival benefits when additional intravesical BCG is utilized as a bladder-preserving treatment in patients with recurrent/persistent tumors at three-month follow-up.

Deep-GA-Net for Accurate and Explainable Detection of Geographic Atrophy on OCT Scans.

Objective: To propose Deep-GA-Net, a 3-dimensional (3D) deep learning network with 3D attention layer, for the detection of geographic atrophy (GA) on spectral domain OCT (SD-OCT) scans, explain its decision making, and compare it with existing methods. Design: Deep learning model development. Participants: Three hundred eleven participants from the Age-Related Eye Disease Study 2 Ancillary SD-OCT Study. Methods: A dataset of 1284 SD-OCT scans from 311 participants was used to develop Deep-GA-Net. Cross-validation was used to evaluate Deep-GA-Net, where each testing set contained no participant from the corresponding training set. En face heatmaps and important regions at the B-scan level were used to visualize the outputs of Deep-GA-Net, and 3 ophthalmologists graded the presence or absence of GA in them to assess the explainability (i.e., understandability and interpretability) of its detections. Main Outcome Measures: Accuracy, area under receiver operating characteristic curve (AUC), area under precision-recall curve (APR). Results: Compared with other networks, Deep-GA-Net achieved the best metrics, with accuracy of 0.93, AUC of 0.94, and APR of 0.91, and received the best gradings of 0.98 and 0.68 on the en face heatmap and B-scan grading tasks, respectively. Conclusions: Deep-GA-Net was able to detect GA accurately from SD-OCT scans. The visualizations of Deep-GA-Net were more explainable, as suggested by 3 ophthalmologists. The code and pretrained models are publicly available at https://github.com/ncbi/Deep-GA-Net. Financial Disclosures: The author(s) have no proprietary or commercial interest in any materials discussed in this article.

Novel urine-based DNA methylation biomarkers for urothelial bladder carcinoma detection in patients with hematuria.

Background: Urothelial bladder carcinoma (UBC) is usually detected during work-up for hematuria. Cystoscopy and/or contrast-enhanced imaging are the gold standard tools for UBC diagnosis, despite limited by being invasive, expensive and low yield in small flat tumors. Objectives: To assess the diagnostic performance of urine-based DNA methylation of six genes (GATA4, P16, P14, APC, CDH1 and CD99) for UBC detection in patients with hematuria. Patients and methods: Voided urine was collected from consecutive patients presented with hematuria for urine cytology and DNA methylation assay of the assigned genes using methylation-specific Polymerase Chain Reaction (PCR). Further assessment by office cystoscopy and imaging with subsequent inpatient cystoscopic biopsy for positive findings was done. The diagnostic characteristics of DNA methylation and urine cytology were assessed based on its capability to predict UBC. Results: We included 246 patients in the study with identified macroscopic hematuria in 204 (82.9%) patients. Positive cytology was found in 78 (31.7%) patients. DNA methylation of GATA4, P16, P14, APC, CDH1 and CD99 genes was identified in 127 (51.6%), 52 (21.1%), 117 (47.6%), 106 (43.1%), 90 (36.6%) and 71 (28.9%) patients, respectively. The sensitivity of the assigned genes for UBC detection ranges from 35% (95%CI: 31-39) to 83% (95%CI: 79-87). Optimal specificity (SP) (100%) was noted for P16, APC and CDH1 genes. While for the other genes (GATA4, P14 and CD99), the SP was 95% (95%CI: 92-98), 96% (95%CI: 92-99) and 97% (95%CI: 93-99), respectively. On multivariate logistic regression analysis, all genes exclusively demonstrated independent prediction of UBC. On receiver operator characteristic (ROC) analysis, all tested genes methylation showed superior area under the curve (AUC) when compared to urine cytology. Conclusions: We have developed a novel urine-based DNA methylation assay for detection of UBC in patients with hematuria with superior diagnostic performance and independent predictive capacity over urine cytology.

PIPE-Net: A pyramidal-input-parallel-encoding network for the segmentation of corneal layer interfaces in OCT images.

Segmentation of corneal layer interfaces in optical coherence tomography (OCT) images is necessary to generate thickness maps used for cornea diagnosis. In this paper, we propose PIPE-Net, a fully convolutional neural network with a pyramidal input, parallel encoders, and a densely connected decoder to segment four corneal layer interfaces. The pyramidal input is encoded using parallel encoders, which allows the network to process a larger receptive field. The encoders are connected level-wise to the decoder through residual summations. The decoder is densely connected using residual summations between its levels to enhance the gradient flow. We use a linear growth rate for the number of feature maps to limit the network parameters, which allows the network to be trained using a small dataset. A dataset of 295 OCT images was obtained and manually segmented by experienced and trained operators. We implemented other related networks in the literature for comparison with our proposed network. We performed k-fold cross-validation to evaluate all the networks, and their performance was evaluated using precision-recall curves and average precision. PIPE-Net outperformed the other networks with an average precision of 0.95. The layer interfaces were detected and smoothed using the Savitzky-Golay filter, and they were closer to the expert.

Oncologic Outcomes of Squamous Cell Carcinoma Versus Urothelial Carcinoma With Squamous Differentiation After Radical Cystectomy for Bladder Carcinoma.

INTRODUCTION: In this study we aim to compare clinicopathological characteristics and cancer specific survival between patients treated with radical cystectomy for pure squamous cell carcinoma (SCC) and urothelial carcinoma with squamous differentiation (SqD). PATIENTS AND METHODS: We reviewed data of 1737 consecutive patients treated with radical cystectomy and urinary diversion between January 2004 and February 2014. Only patients with pure SCC or SqD were included in the analysis. Squamous differentiation was defined as intercellular bridges or keratinization in the tumor. Clinicopathological data and recurrence free survival (RFS) were compared between patients diagnosed with SCC and SqD. RESULTS: SCC and SqD were found in 318 and 223 patients, respectively. Mean age was 57 ± 8.3 years in SCC and 58.8 ± 7.8 in SqD (P = .008). A higher proportion of female patients was observed in SCC group compared to SqD (31.8% vs. 22% P < .0001). Patients with SqD were more likely to have extravesical (58.3% vs. 46.2%: P = .006) and nodal positive disease (34.5% vs. 14.5%: P < .0001) than pure SCC patients. Bilharzial eggs were found in 61% of SCC vs. 46% of SqD (P = .001).; The median (IQR) follow up period for SCC and SqD was 63 (12-112) months and 23 months (9-74.7), respectively. The 5-year RFS for SCC and SqD were 77% and 59.8 %, respectively (P < .0001).; Multivariate cox regression analysis identified advanced pT stage (OR: 1.9, 95% CI: 1.3-2.86, P = .0001), nodal positive disease (OR: 1.6, 95% CI: 1.1-2.48, P = .01) and SqD histology (OR: 1.6, 95% CI: 1.14-2.31, P = .007 as independent predictors of 5-year RFS. CONCLUSION: Patient with SCC had significantly higher 5-year RFS in comparison to SqD. The higher rate of extravesical disease and lymph node metastasis in SqD patients is indicative of aggressive behavior of this histologic type.

Identification of Different miRNAs and Their Relevant miRNA Targeted Genes Involved in Sister Chromatid Cohesion and Segregation (SCCS)/chromatin Remodeling Pathway on T1G3 Urothelial Carcinoma (UC) Response to BCG Immunotherapy.

BACKGROUND: Till now, no definite clinical or laboratory marker can predict the recurrence or progression of T1 G3 urothelial carcinoma (UC). Genetic aberrations of the chromatin remodeling genes and sister chromatid cohesion and segregation (SCCS) were identified in UC. Here we investigated the impact of novel miRNAs and their targeted expressed SCCS and chromatin remodeling genes on T1G3 UC response to Bacillus Calmette-Guérin (BCG) therapy. METHODS: One hundred tissue samples were obtained from NMIBC patients. Gene expression and immunohistochemical assay of STAG2, ARID1A, NCOR1and UTX were assessed. MiRNA analysis for their targeting miRNAs (miR-21, miR-31, Let7a and miR-199a) was carried out. Assessed genes were compared between responders and no responders to BCG. Univariate and multivariate analysis of predictors of disease recurrence and progression were performed using cox regression analysis. RESULTS: Thirty-two and 22 patients developed recurrence and progression to MIBC (BCG non-responders). BCG non-responders showed statistically significant higher expression of miR-21 and their targeted STAG2, miR-199a and NCOR1 gene (P < .001), and lower expression of miR-31, Let7a, ARID1A and UTX genes (P < .001). Higher miR-199a (P = .006) and lower miR-31 (P = .01), ARID1A (P = .008) and UTX (P = .03) were independent predictor of higher tumor recurrence. Recurrent disease (P = .003), higher expression of STAG2 (P = .01), NCOR1 (P = .01) and miR-21 (P = .03) genes and lower expression of miR-31 (P = .02), Let7a (P = .04) and ARID1A (P = .04) genes were the independent predictor of disease progression. CONCLUSION: Upregulation of STAG2 and NCOR1 and down regulation of ARID1A and UTX genes and their targeting miRNAs were associated with UC non-response to BCG.

Predictive value of immunological markers after bacille Calmette-Guérin induction in bladder cancer.

OBJECTIVES: To investigate the predictive value of different immunological markers on treatment outcomes after bacille Calmette-Guérin (BCG) induction in high-risk non-muscle-invasive bladder cancer (NMIBC). PATIENTS AND METHODS: Patients who underwent transurethral resection of bladder tumour for NMIBC were assessed for study eligibility. Urine and blood samples were taken from patients at baseline (immediately before first dose of induction) and after induction (4 h after last [sixth] dose). Urine samples were evaluated for interleukin (IL)-2 and IL-10 by solid-phase enzyme-linked immunosorbent assay. Blood samples were evaluated for tumour necrosis factor α (TNF-α), cytotoxic T-lymphocyte antigen 4 (CTLA-4) and transcription factors (TFs) (GATA-binding protein 3 [GATA3], T-box expressed in T cells [T-bet], and forkhead box protein 3 [FoxP3]) using quantitative reverse transcriptase-polymerase chain reaction analysis. Change pattern and fold change of each evaluable marker was assessed in relation to different treatment outcomes (initial complete response [ICR]/recurrence/progression). RESULTS: Between July 2013 and May 2019, 204 patients were included. Among evaluable markers, urinary IL-2 and serum TNF-α increased in all patients, serum CTLA-4 and FoxP3+ showed a predominant decreased pattern in 188 (92.2%) and 192 (94.1%) patients, respectively. An ICR was achieved in 186 (91.2%) patients. Serum TNF-α fold change and urinary IL-10 change pattern were significantly associated with an ICR (P = 0.001 and P = 0.03, respectively). At a median (range) follow-up of 37 (20-88) months, 104 (56%) patients developed recurrence. Urinary IL-10, serum CTLA-4, T-bet+ , FoxP3+ change patterns and GATA3+ /T-bet+ ratio were significantly associated with tumour recurrence (P = 0.001, P = 0.001, P = 0.02, P = 0.009 and P = 0.001, respectively). Tumour progression occurred in 34 (18.3%) patients. Urinary IL-10, serum CTLA-4, serum T-bet+ change patterns and GATA3+ /T-bet+ ratio were independent predictors of tumour progression (P = 0.001, P = 0.001, P = 0.02 and P = 0.001, respectively). CONCLUSIONS: Urinary IL-10 and serum TNF-α can significantly predict ICR. Moreover, change pattern of urinary IL-10, serum CTLA-4, TFs (GATA3, T-bet and FoxP3) and GATA3+ /T-bet+ ratio after BCG induction can independently predict further BCG response. These markers could be implemented in clinical practice when management options are discussed or in systems with severe BCG shortage.

Predictors and survival benefit of achieving pentafecta in a contemporary series of open radical cystectomy.

BACKGROUND: Pentafecta provides a comprehensive approach for standardized reporting of surgical and oncologic outcomes after radical cystectomy and urinary diversion. We aimed to report the rate, predictors of achieving pentafecta and its impact on long-term survival in a contemporary series of open radical cystectomy (ORC). METHODS: A retrospective analysis of a computerized database of patients treated with ORC between 2004 till 2014 was performed. Pentafecta criteria included negative soft tissue surgical margin (STSM), retrieval of ≥16 lymph nodes, absence of clinical recurrence within 12 months after surgery, absence of high-grade complication (GIII-V) within 90 days after surgery, and absence of urinary diversion related complications at 12 months follow-up. Multivariate analysis was used to identify predictors of achieving pentafecta. RESULTS: Pentafecta was achieved in 545 (33.6%) patients out of 1624 included in the study. Absence of ≥16 LN yield was the first cause of missing pentafecta (49.5%). Multivariate analysis identified: ASA Score grades ≥III (OR=0.7, 95%CI 0.6-0.9, P=0.04), BMI≥35 (OR=0.5, 95%CI 0.3-0.8, P=0.007), perioperative blood transfusion (≥4 units) (OR=0.5, 95%CI 0.3-0.7, P=0.001), and ileal conduit (OR=0.7, 95%CI 0.5-0.9, P= 0.01) as independent predictors of missing pentafecta. Patients who achieved pentafecta had higher estimated 5-year RFS than their counterparts (81.7% vs. 62.5%; P<0.0001). CONCLUSIONS: Pentafecta was achieved in nearly one third of patients after ORC. Achievement of pentafecta was associated with better long-term recurrence-free survival. Obesity (class II, III), perioperative blood transfusion (>4 units), associated comorbidities, and ileal conduit were independent predictors of missing pentafecta.

Comparison of Autonomous AS-OCT Deep Learning Algorithm and Clinical Dry Eye Tests in Diagnosis of Dry Eye Disease.

OBJECTIVE: To evaluate a deep learning-based method to autonomously detect dry eye disease (DED) in anterior segment optical coherence tomography (AS-OCT) images compared to common clinical dry eye tests. METHODS: In this study, 27,180 AS-OCT images were prospectively collected from 151 eyes of 91 patients. Images were used to train and test the deep learning model. Masked cornea specialist ophthalmologist diagnoses were used as the gold standard. Clinical dry eye tests were performed on patients in the DED group to compare the results of the model. The dry eye tests performed were tear break-up time (TBUT), Schirmer's test, corneal staining, conjunctival staining, and Ocular Surface Disease Index (OSDI). RESULTS: Our deep learning model achieved an accuracy of 84.62%, sensitivity of 86.36%, and specificity of 82.35% in the diagnosis of DED. The positive likelihood ratio was 4.89, and the negative likelihood ratio was 0.17. The mean DED probability score was 0.81 ± 0.23 in the DED group and 0.20 ± 0.27 in the healthy group (P < 0.01). The deep learning model accuracy in the diagnosis of DED was significantly better than that of corneal staining, conjunctival staining, and Schirmer's test (P < 0.05). There was no significant difference between the deep learning diagnostic accuracy and that of the OSDI and TBUT. CONCLUSION: Based on preliminary results, reliable autonomous diagnosis of DED with our deep learning model was achieved, when compared with standard dry eye clinical tests that correlated significantly more or similarly to diagnoses made by cornea specialist ophthalmologists.

Prediction of corneal graft rejection using central endothelium/Descemet's membrane complex thickness in high-risk corneal transplants.

To determine whether measurements of Endothelium/Descemet complex thickness (En/DMT) are of predictive value for corneal graft rejection after high-risk corneal transplantation, we conducted this prospective, single-center, observational case series including sixty eyes (60 patients) at high risk for corneal graft rejection (GR) because of previous immunologic graft failure or having at least two quadrants of stromal vascularization. Patients underwent corneal transplant. At 1st, 3rd, 6th, 9th, and 12th postoperative month, HD-OCT imaging of the cornea was performed, and the corneal status was determined clinically at each visit by a masked cornea specialist. Custom-built segmentation tomography algorithm was used to measure the central En/DMT. Relationships between baseline factors and En/DMT were explored. Time dependent covariate Cox survival regression was used to assess the effect of post-operative En/DMT changes during follow up. A longitudinal repeated measures model was used to assess the relationship between En/DMT and graft status. Outcome measures included graft rejection, central Endothelium/Descemet's complex thickness, and central corneal thickness (CCT). In patients with GR (35%), the central En/DMT increased significantly 5.3 months (95% CI: 2, 11) prior to the clinical diagnosis of GR, while it remained stable in patients without GR. During the 1-year follow up, the rejected grafts have higher mean pre-rejection En/DMTs (p = 0.01), compared to CCTs (p = 0.7). For En/DMT ≥ 18 µm cut-off (at any pre-rejection visit), the Cox proportional hazard ratio was 6.89 (95% CI: 2.03, 23.4; p = 0.002), and it increased to 9.91 (95% CI: 3.32, 29.6; p < 0.001) with a ≥ 19 µm cut-off. In high-risk corneal transplants, the increase in En/DMT allowed predicting rejection prior to the clinical diagnosis.

A Novel Network With Parallel Resolution Encoders for the Diagnosis of Corneal Diseases.

OBJECTIVE: To propose a deep-learning network for the diagnosis of two corneal diseases: Fuchs' endothlelial dystrophy and keratoconus, based on optical coherence tomography (OCT) images of the cornea. METHODS: In this paper, we propose a novel network with parallel resolution-specific encoders and composite classification features to directly diagnose Fuchs' endothelial dystrophy and keratoconus using OCT images. Our proposed network consists of a multi-resolution input, multiple parallel encoders, and a composite of convolutional and dense features for classification. The purpose of using parallel resolution-specific encoders is to perform multi-resolution feature fusion. Also, using composite classification features enhances the dense feature learning. We implemented other related networks for comparison with our network and performed k-fold cross-validation on a dataset of 16,721 OCT images. We used saliency maps and sensitivity analysis to visualize our proposed network. RESULTS: The proposed network outperformed other networks with an image classification accuracy of 0.91 and a scan classification accuracy of 0.94. The visualizations show that our network learned better features than other networks. SIGNIFICANCE: The proposed methods can potentially be a step towards the early diagnosis of corneal diseases, which is necessary to prevent their progression, hence, prevent loss of vision.

Can repeat biopsy be skipped after initial complete resection of T1 bladder cancer? The role of a novel urinary mRNA biomarker.

OBJECTIVES: To prospectively investigate the role of a urinary mRNA biomarker (Xpert Test) after initial complete resection of T1 bladder cancer (BC) for the prediction of positive repeat biopsy for malignancy. METHODS: Patients who underwent TURBT for NMIBC between September 2018 and April 2020 were included. Patients with benign pathology, incomplete resection, concomitant CIS/upper tract urothelial tumor or muscle invasive BC, were excluded. 2 to 6 weeks after primary TURBT, voided urine sample was retrieved for Xpert analysis and patients were scheduled for repeat biopsy. The primary outcome was to determine the role of positive Xpert test to predict positive repeat biopsy for malignancy. RESULTS: During the study period, 254 patients met the study inclusion criteria of which 61 (24%) patients had recurrent NMIBC. Complete resection was censured by the presence of detrusor muscle in the specimen with documented T1 disease in all study participants. Xpert test was positive in 128 patients; of whom 85 (66.4%) showed positive repeat biopsy (HR=6.2, 95%CI=3.46-9.4, P = 0.002). The sensitivity, specificity, positive and negative predictive values of Xpert test for repeat biopsy were 85.9% (95%CI: 82-89), 72.3% (95%CI: 68-76), 66.4% (95%CI: 62-71) and 88.9% (95%CI: 85-94), respectively. On median (range) follow up of 12(3-25) months, tumor recurrence was encountered in 84 (35%) patients. On multivariate Cox regression analysis, Xpert test was significantly associated with tumor recurrence (HR= 9.7, 95%CI=5-18, P <0.001). CONCLUSIONS: Positive Xpert test after primary complete resection of T1 BC is significantly associated with positive repeat biopsy for malignancy. In addition, Xpert test is an independent predictor of early tumor recurrence. Xpert test might be applied after initial complete resection of NMIBC to minimize unnecessary repeat biopsy with potential saving of healthcare costs and reduction in patient morbidity.

Multidisease Deep Learning Neural Network for the Diagnosis of Corneal Diseases.

PURPOSE: To report a multidisease deep learning diagnostic network (MDDN) of common corneal diseases: dry eye syndrome (DES), Fuchs endothelial dystrophy (FED), and keratoconus (KCN) using anterior segment optical coherence tomography (AS-OCT) images. STUDY DESIGN: Development of a deep learning neural network diagnosis algorithm. METHODS: A total of 158,220 AS-OCT images from 879 eyes of 478 subjects were used to develop and validate a classification deep network. After a quality check, the network was trained and validated using 134,460 images. We tested the network using a test set of consecutive patients involving 23,760 AS-OCT images of 132 eyes of 69 patients. The area under receiver operating characteristic curve (AUROC), area under precision-recall curve (AUPRC), and F1 score and 95% confidence intervals (CIs) were computed. RESULTS: The MDDN achieved eye-level AUROCs >0.99 (95% CI: 0.90, 1.0), AUPRCs > 0.96 (95% CI: 0.90, 1.0), and F1 scores > 0.90 (95% CI: 0.81, 1.0) for DES, FED, and KCN, respectively. CONCLUSIONS: MDDN is a novel diagnostic tool for corneal diseases that can be used to automatically diagnose KCN, FED, and DES using only AS-OCT images.

Prospective Validation of Clinical Usefulness of a Novel mRNA-based Urine Test (Xpert® Bladder Cancer Monitor) for surveillance in Non Muscle Invasive Bladder Cancer.

OBJECTIVES: To assess the clinical performance characteristics of Xpert Monitor test for recurrence detection during surveillance of patients with non muscle invasive bladder cancer (NMIBC). PATIENTS AND METHODS: Patient with previous history of NMIBC were included in the study. Voided urine specimens were collected for Xpert monitor analysis and cytology. Office cystoscopy was performed for all study participants with in patient biopsy specimen retrieval for positive or suspicious cases. Test characteristics were calculated based on cystoscopy/biopsy results and compared between Xpert and cytology. RESULTS: Between March 2018 and May 2019, 181 patients including 168 (92.8%) males fulfilled the study criteria with median age 61 years, Primary tumors were low, intermediate, high risk in 2.8%, 22.7% and 74.5% of patients respectively. Biopsy proven recurrence was detected in 19 patients (10.4%). Xpert Monitor had a sensitivity of 73.7% with a negative predictive value (NPV) of 96.3%. Xpert Monitor was positive in all cases with high grade tumors (9 patients). Urine cytology showed sensitivity of 47% and an NPV of 93.2%. During follow up surveillance, out of 162 cystoscopy negative patients (CNP), 9.3% developed recurrence within 8 months. Xpert Monitor was found to be an independent predictor of early recurrence in CNP (HR=2.8, 95%CI=1.1-7.2, p=0.01). CONCLUSIONS: Xpert Monitor urine test has a superior diagnostic performance for recurrence detection in NMIBC patients compared to urine cytology. It might be a helpful tool not only for excluding bladder cancer recurrence in those patients, but also for prediction of possible future early recurrence.

Pathological-Corneas Layer Segmentation and Thickness Measurement in OCT Images.

Purpose: The purpose of this study was to propose a new algorithm for the segmentation and thickness measurement of pathological corneas with irregular layers using a two-stage graph search and ray tracing. Methods: In the first stage, a graph, with only gradient edge-cost, is used to segment the air-epithelium and endothelium-aqueous boundaries. In the second stage, a graph, with gradient, directional, and multiplier edge-cost, is used to correct segmentation. The optical coherence tomography (OCT) image is flattened using the air-epithelium boundary and a graph search is used to segment the epithelium-Bowman's and Bowman's-stroma boundaries. Then, the OCT image is flattened using the endothelium-aqueous boundary and a graph search is used to segment the Descemet's membrane. Ray tracing is used to correct the inter-boundary distances, then the thickness is measured using the shortest distance. The proposed algorithm was trained and evaluated using 190 OCT images manually segmented by trained operators. Results: The mean and standard deviation of the unsigned errors of the algorithm-operator and inter-operator were 0.89 ± 1.03 and 0.77 ± 0.68 pixels in segmentation and 3.62 ± 3.98 and 2.95 ± 2.52 µm in thickness measurement. Conclusions: Our proposed algorithm can produce accurate segmentation and thickness measurements compared with the manual operators. Translational Relevance: Our algorithm could be potentially useful in the clinical practice.

An Investigation of Fibulin-2 in Hypertrophic Cardiomyopathy.

Hypertrophic cardiomyopathy (HCM) is the most common inherited heart muscle disease, with a prevalence of at least 1 in 500 in the general population. The disease is pleiotropic and is characterized by an increased stiffness of the myocardium, partly due to changes in the extracellular matrix (ECM), with elevated levels of interstitial fibrosis. Myocardial fibrosis is linked to impaired diastolic function and possibly phenotypic heterogeneity of HCM. The ECM consists of a very large number of proteins, which actively interact with each other as well as with myocardial cells. The role of other multiple components of the ECM in HCM has not been defined. Fibulin-2 is a glycoprotein component of the ECM, which plays an important role during embryogenesis of the heart; however, its role in adult myocardium has not been adequately studied. We here describe, for the first time, abnormal expression of fibulin-2 in the myocardium in patients with HCM as compared to normal controls. This abnormal expression was localized in the cytoplasm of myocardial cells and in the interstitial fibroblasts. In addition, fibulin-2 levels, measured by ELISA, were significantly elevated in the serum of patients with HCM as compared to normal controls.

Parastomal hernia after ileal conduit urinary diversion: re-visiting the predictors radiologically and according to patient-reported outcome measures.

PURPOSE: To evaluate the predictors of post-ileal conduit (IC) parastomal hernia (PSH) based on a standard grading methodology and according to the patients reported outcome measures (PROM). METHODS: A prospective evaluation for patients with IC attending their scheduled follow-up was conducted between December 2013 and October 2015. The hernia stage was determined according to the European Hernia Society (EHS) classification as types I and II included defect size < 5 cm without and with a concomitant incisional hernia, respectively. Types III and IV included defect size > 5 cm without and with a concomitant incisional hernia (high-grade hernia). The evaluation was performed by a non-contrast CT scan. PROM were defined as symptomatic if there were hernia-related abdominal discomfort, appliance problems, and/or bowel complications. Perioperative parameters were modeled for prediction of high-grade and PROM outcomes. RESULTS: PSH was diagnosed in 138 (39.9%) patients, symptomatic in 119 (34.4%) and high-grade in 59 (17%). Independent predictors of radiologically diagnosed PSH were hypoalbuminemia (odds ratio [OR]: 1.7; 95% Confidence interval [CI]: 1.1-2.7; p = 0.02), localised disease (OR: 0.6; 95% CI: 0.3-0.9; p = 0.04) and negative lymphadenopathy (OR: 0.4; 95%CI: 0.2-0.8; p = 0.004). Predictors of symptomatic PSH were hypoalbuminemia (OR: 2; 95%CI: 1.2-2.3: p = 0.003) and previous hernia surgery (OR: 2.1; 95%CI: 1.1-4.2; p = 0.024). CONCLUSIONS: Only a small proportion of patients with PSH were asymptomatic. Preoperative hypoalbuminemia was the most significant factor contributing to the development and symptomatizing of PSH. Previous hernia surgery further contributed to the patient complaint.

Automated diagnosis and staging of Fuchs' endothelial cell corneal dystrophy using deep learning.

BACKGROUND: To describe the diagnostic performance of a deep learning algorithm in discriminating early-stage Fuchs' endothelial corneal dystrophy (FECD) without clinically evident corneal edema from healthy and late-stage FECD eyes using high-definition optical coherence tomography (HD-OCT). METHODS: In this observational case-control study, 104 eyes (53 FECD eyes and 51 healthy controls) received HD-OCT imaging (Envisu R2210, Bioptigen, Buffalo Grove, IL, USA) using a 6 mm radial scan pattern centered on the corneal vertex. FECD was clinically categorized into early (without corneal edema) and late-stage (with corneal edema). A total of 18,720 anterior segment optical coherence tomography (AS-OCT) images (9180 healthy; 5400 early-stage FECD; 4140 late-stage FECD) of 104 eyes (81 patients) were used to develop and validate a deep learning classification network to differentiate early-stage FECD eyes from healthy eyes and those with clinical edema. Using 5-fold cross-validation on the dataset containing 11,340 OCT images (63 eyes), the network was trained with 80% of these images (3420 healthy; 3060 early-stage FECD; 2700 late-stage FECD), then tested with 20% (720 healthy; 720 early-stage FECD; 720 late-stage FECD). Thereafter, a final model was trained with the entire dataset consisting the 11,340 images and validated with a remaining 7380 images of unseen AS-OCT scans of 41 eyes (5040 healthy; 1620 early-stage FECD 720 late-stage FECD). Visualization of learned features was done, and area under curve (AUC), specificity, and sensitivity of the prediction outputs for healthy, early and late-stage FECD were computed. RESULTS: The final model achieved an AUC of 0.997 ± 0.005 with 91% sensitivity and 97% specificity in detecting early-FECD; an AUC of 0.974 ± 0.005 with a specificity of 92% and a sensitivity up to 100% in detecting late-stage FECD; and an AUC of 0.998 ± 0.001 with a specificity 98% and a sensitivity of 99% in discriminating healthy corneas from all FECD. CONCLUSION: Deep learning algorithm is an accurate autonomous novel diagnostic tool of FECD with very high sensitivity and specificity that can be used to grade FECD severity with high accuracy.

Diagnostic performance and predictive capacity of early urine cytology after transurethral resection of nonmuscle invasive bladder cancer: A prospective study.

OBJECTIVES: To prospectively evaluate the value of early urine cytology (EUC) after the primary transurethral resection of bladder tumor (TURBT) of nonmuscle invasive bladder cancer (NMIBC) for the prediction of positive biopsy findings on repeat TURBT. METHODS: After approval of institutional review board, patients who underwent TURBT for NMIBC between February 2014 and July 2018 were included in the study. Patients with concomitant Carcinoma in Situ (CIS), upper tract urothelial tumors, biopsy proven muscle invasion, or low-risk NMIBC (single, primary, Ta, and G1 tumor) were excluded. Forty-eight hours after primary TURBT, EUC was retrieved and patients were scheduled for repeat TURBT 2 to 6 weeks later according to the predetermined protocol. The primary outcome was to determine the role of positive EUC to predict positive biopsy findings on repeat TURBT. RESULTS: During the study period, 198 patients fulfilled the study inclusion criteria of which 49 (25%) had recurrent NMIBC. Primary TURBT pathology results showed T1 stage in 195 (98.5%) patients and high-grade malignancy in 158 (79.8%). Intermediate- and high-risk NMIBC were defined in 49 (25%) and 149 (75%) patients, respectively. EUC was positive in 114 patients; of whom 78 (68.4%) showed positive biopsy findings on repeat TURBT (P = 0.001). The sensitivity, specificity, negative, and positive predictive values of EUC for biopsy findings at repeat TURBT were 90% (95%CI: 87-94), 75% (95%CI: 71-79), 89% (95%CI: 85-94), and 68% (95%CI: 62-74), respectively. On mean (±SD) follow-up of 42(±13) months, tumor recurrence was encountered in 101 (53%) patients. On multivariate Cox regression analysis, EUC was significantly associated with tumor recurrence (HR = 4.6, 95%CI: 2.37-8.9, P < 0.001). CONCLUSIONS: Positive EUC after primary TURBT for NMIBC is significantly associated with positive repeat TURBT for malignancy. EUC is an independent predictor of tumor recurrence. EUC might be implemented after primary TURBT to help refining indications of repeat biopsy and planning of further intervention.