Search

PurposeTo report an image analysis pipeline, DDLSNet, consisting of a rim segmentation (RimNet) branch and a disc size classification (DiscNet) branch to automate estimation of the disc damage likelihood scale (DDLS).DesignRetrospective observational.ParticipantsRimNet and DiscNet were developed with 1208 and 11 536 optic disc photographs (ODPs), respectively. DDLSNet performance was evaluated on 120 ODPs from the RimNet test set, for which the DDLS scores were graded by clinicians. Reproducibility was evaluated on a group of 781 eyes, each with 2 ODPs taken within 4 years apart.MethodsDisc damage likelihood scale calculation requires estimation of optic disc size, provided by DiscNet (VGG19 network), and the minimum rim-to-disc ratio (mRDR) or absent rim width (ARW), provided by RimNet (InceptionV3/LinkNet segmentation model). To build RimNet's dataset, glaucoma specialists marked optic disc rim and cup boundaries on ODPs. The "ground truth" mRDR or ARW was calculated. For DiscNet's dataset, corresponding OCT images provided "ground truth" disc size. Optic disc photographs were split into 80/10/10 for training, validation, and testing, respectively, for RimNet and DiscNet. DDLSNet estimation was tested against manual grading of DDLS by clinicians with the average score used as "ground truth." Reproducibility of DDLSNet grading was evaluated by repeating DDLS estimation on a dataset of nonprogressing paired ODPs taken at separate times.Main outcome measuresThe main outcome measure was a weighted kappa score between clinicians and the DDLSNet pipeline with agreement defined as ± 1 DDLS score difference.ResultsRimNet achieved an mRDR mean absolute error (MAE) of 0.04 (± 0.03) and an ARW MAE of 48.9 (± 35.9) degrees when compared to clinician segmentations. DiscNet achieved 73% (95% confidence interval [CI]: 70%, 75%) classification accuracy. DDLSNet achieved an average weighted kappa agreement of 0.54 (95% CI: 0.40, 0.68) compared to clinicians. Average interclinician agreement was 0.52 (95% CI: 0.49, 0.56). Reproducibility testing demonstrated that 96% of ODP pairs had a difference of ≤ 1 DDLS score.ConclusionsDDLSNet achieved moderate agreement with clinicians for DDLS grading. This novel approach illustrates the feasibility of automated ODP grading for assessing glaucoma severity. Further improvements may be achieved by increasing the number of incomplete rims sample size, expanding the hyperparameter search, and increasing the agreement of clinicians grading ODPs.

PurposeAccurate neural rim measurement based on optic disc imaging is important to glaucoma severity grading and often performed by trained glaucoma specialists. We aim to improve upon existing automated tools by building a fully automated system (RimNet) for direct rim identification in glaucomatous eyes and measurement of the minimum rim-to-disc ratio (mRDR) in intact rims, the angle of absent rim width (ARW) in incomplete rims, and the rim-to-disc-area ratio (RDAR) with the goal of optic disc damage grading.DesignRetrospective cross sectional study.ParticipantsOne thousand and twenty-eight optic disc photographs with evidence of glaucomatous optic nerve damage from 1021 eyes of 903 patients with any form of primary glaucoma were included. The mean age was 63.7 (± 14.9) yrs. The average mean deviation of visual fields was -8.03 (± 8.59).MethodsThe images were required to be of adequate quality, have signs of glaucomatous damage, and be free of significant concurrent pathology as independently determined by glaucoma specialists. Rim and optic cup masks for each image were manually delineated by glaucoma specialists. The database was randomly split into 80/10/10 for training, validation, and testing, respectively. RimNet consists of a deep learning rim and cup segmentation model, a computer vision mRDR measurement tool for intact rims, and an ARW measurement tool for incomplete rims. The mRDR is calculated at the thinnest rim section while ARW is calculated in regions of total rim loss. The RDAR was also calculated. Evaluation on the Drishti-GS dataset provided external validation (Sivaswamy 2015).Main outcome measuresMedian Absolute Error (MAE) between glaucoma specialists and RimNet for mRDR and ARW.ResultsOn the test set, RimNet achieved a mRDR MAE of 0.03 (0.05), ARW MAE of 31 (89)°, and an RDAR MAE of 0.09 (0.10). On the Drishti-GS dataset, an mRDR MAE of 0.03 (0.04) and an mRDAR MAE of 0.09 (0.10) was observed.ConclusionsRimNet demonstrated acceptably accurate rim segmentation and mRDR and ARW measurements. The fully automated algorithm presented here would be a valuable component in an automated mRDR-based glaucoma grading system. Further improvements could be made by improving identification and segmentation performance on incomplete rims and expanding the number and variety of glaucomatous training images.

Purpose: To report an image analysis pipeline, DDLSNet, consisting of a rim segmentation (RimNet) branch and a disc size classification (DiscNet) branch to automate estimation of the disc damage likelihood scale (DDLS). Design: Retrospective observational. Participants: RimNet and DiscNet were developed with 1208 and 11 536 optic disc photographs (ODPs), respectively. DDLSNet performance was evaluated on 120 ODPs from the RimNet test set, for which the DDLS scores were graded by clinicians. Reproducibility was evaluated on a group of 781 eyes, each with 2 ODPs taken within 4 years apart. Methods: Disc damage likelihood scale calculation requires estimation of optic disc size, provided by DiscNet (VGG19 network), and the minimum rim-to-disc ratio (mRDR) or absent rim width (ARW), provided by RimNet (InceptionV3/LinkNet segmentation model). To build RimNet’s dataset, glaucoma specialists marked optic disc rim and cup boundaries on ODPs. The ''ground truth'' mRDR or ARW was calculated. For DiscNet’s dataset, corresponding OCT images provided ''ground truth'' disc size. Optic disc photographs were split into 80/10/10 for training, validation, and testing, respectively, for RimNet and DiscNet. DDLSNet estimation was tested against manual grading of DDLS by clinicians with the average score used as ''ground truth.'' Reproducibility of DDLSNet grading was evaluated by repeating DDLS estimation on a dataset of nonprogressing paired ODPs taken at separate times. Main Outcome Measures: The main outcome measure was a weighted kappa score between clinicians and the DDLSNet pipeline with agreement defined as ± 1 DDLS score difference. Results: RimNet achieved an mRDR mean absolute error (MAE) of 0.04 (± 0.03) and an ARW MAE of 48.9 (± 35.9) degrees when compared to clinician segmentations. DiscNet achieved 73% (95% confidence interval [CI]: 70%, 75%) classification accuracy. DDLSNet achieved an average weighted kappa agreement of 0.54 (95% CI: 0.40, 0.68) compared to clinicians. Average interclinician agreement was 0.52 (95% CI: 0.49, 0.56). Reproducibility testing demonstrated that 96% of ODP pairs had a difference of ≤ 1 DDLS score. Conclusions: DDLSNet achieved moderate agreement with clinicians for DDLS grading. This novel approach illustrates the feasibility of automated ODP grading for assessing glaucoma severity. Further improvements may be achieved by increasing the number of incomplete rims sample size, expanding the hyperparameter search, and increasing the agreement of clinicians grading ODPs. Financial Disclosure(s): Proprietary or commercial disclosure may be found after the references.

Purpose: Accurate neural rim measurement based on optic disc imaging is important to glaucoma severity grading and often performed by trained glaucoma specialists. We aim to improve upon existing automated tools by building a fully automated system (RimNet) for direct rim identification in glaucomatous eyes and measurement of the minimum rim-to-disc ratio (mRDR) in intact rims, the angle of absent rim width (ARW) in incomplete rims, and the rim-to-disc-area ratio (RDAR) with the goal of optic disc damage grading. Design: Retrospective cross sectional study. Participants: One thousand and twenty-eight optic disc photographs with evidence of glaucomatous optic nerve damage from 1021 eyes of 903 patients with any form of primary glaucoma were included. The mean age was 63.7 (± 14.9) yrs. The average mean deviation of visual fields was −8.03 (± 8.59). Methods: The images were required to be of adequate quality, have signs of glaucomatous damage, and be free of significant concurrent pathology as independently determined by glaucoma specialists. Rim and optic cup masks for each image were manually delineated by glaucoma specialists. The database was randomly split into 80/10/10 for training, validation, and testing, respectively. RimNet consists of a deep learning rim and cup segmentation model, a computer vision mRDR measurement tool for intact rims, and an ARW measurement tool for incomplete rims. The mRDR is calculated at the thinnest rim section while ARW is calculated in regions of total rim loss. The RDAR was also calculated. Evaluation on the Drishti-GS dataset provided external validation (Sivaswamy 2015). Main Outcome Measures: Median Absolute Error (MAE) between glaucoma specialists and RimNet for mRDR and ARW. Results: On the test set, RimNet achieved a mRDR MAE of 0.03 (0.05), ARW MAE of 31 (89)°, and an RDAR MAE of 0.09 (0.10). On the Drishti-GS dataset, an mRDR MAE of 0.03 (0.04) and an mRDAR MAE of 0.09 (0.10) was observed. Conclusions: RimNet demonstrated acceptably accurate rim segmentation and mRDR and ARW measurements. The fully automated algorithm presented here would be a valuable component in an automated mRDR-based glaucoma grading system. Further improvements could be made by improving identification and segmentation performance on incomplete rims and expanding the number and variety of glaucomatous training images.

Clinically, there are many systems of visual field imaging systems. The agreement between glaucoma function and appearance is still uncertain. This study was conducted to compare visual field staging classifications (as function staging) with Disc Damage Likelihood Scale (DDLS as appearance staging) in Primary Open Angle Glaucoma (POAG). A cross-section study was performed using 55 patients (90 eyes) examined and treated by the Vietnam National Institute of Ophthalmology from 9/2016 - 9/2017. The MD (Mean Deviation), PSD (Pattern Standard Deviation), and VFI (Visual Field Index) indexes were recorded and were used to classify glaucoma staging according to categories of eGSS (enhanced Glaucoma Severity Staging), mGSS (modified Glaucoma Severity Staging), HPA (Hodapp- Parrish-Aderson), AGIS (Advanced Glaucoma Intervention Study), and DDLS. Using SPSS 16.0 software, statistical significance was tested with proper tests of Chi-square test, Fisher exact test, Kappa (to assess the agreement-disagreement), and Spearman Rank (to assess the level of correlation). Mean VFI were -11.49 dB (MD), 5.85 dB (PSD), and 74.2 % (VFI). Compared to DDLS, all systems had slight agreement (K < 0.2). The disagreement percentage was highest in eGSS (70 %), followed by mGSS (50 %) and HPA (48.9 %). In terms of definitive diagnosis, HPA and mGSS showed a substantial agreement (K > 0.6) with AGIS, which was higher than that between eGSS and AGIS (0.773 and 0.75 vs 0.399), with p < 0.001. In terms of staging detection, the agreement between HPA, mGSS, and AGIS was substantial (K > 0.6), while that between eGSS and AGIS was fair (K < 0.4). In conclusion, mGSS and HPA tended to show stronger agreement with standard classifications than eGSS. mGSS should be used in clinical practice and research. [ABSTRACT FROM AUTHOR]

Glaucoma is a retinal disease caused due to increased intraocular pressure in the eyes. It is the second most dominant cause of irreversible blindness after cataract, and if this remains undiagnosed, it may become the first common cause. Ophthalmologists use different comprehensive retinal examinations such as ophthalmoscopy, tonometry, perimetry, gonioscopy and pachymetry to diagnose glaucoma. But all these approaches are manual and time-consuming. Thus, a computer-aided diagnosis system may aid as an assistive measure for the initial screening of glaucoma for diagnosis purposes, thereby reducing the computational complexity. This paper presents a deep learning-based disc cup segmentation glaucoma network (DC-Gnet) for the extraction of structural features namely cup-to-disc ratio, disc damage likelihood scale and inferior superior nasal temporal regions for diagnosis of glaucoma. The proposed approach of segmentation has been tested on RIM-One and Drishti-GS dataset. Further, based on experimental analysis, the DC-Gnet is found to outperform U-net, Gnet and Deep-lab architectures. [ABSTRACT FROM AUTHOR]

Background: Clinically, there are many systems of visual field staging parallel exist. The agreement between glaucoma function and appearance is still uncertain. This study was to compare visual field staging classifications (as function staging) with disc damage likelihood scale (DDLS as appearance staging) in primary open angle glaucoma (POAG). Methods: Cross-section study on 55 patients (90 eyes) examined and treated in Vietnam National Institute of Ophthalmology from 9/2016 – 9/2017. The MD (mean deviation), PSD (pattern standard deviation), VFI (visual field index) indexes were recorded and were used to classify glaucoma staging according to categories of eGSS (enhance glaucoma severity staging), mGSS (modified glaucoma severity staging), HPA (Hoddaps – Palmela – Adersons), AGIS (advance glaucoma intervention study) and DDLS. Using SPSS 16.0 software, statistically significance was tested with proper tests as Chi – square test, Fisher exact test, Kappa (to assess the agreement - disagreement), Spearman Rank (to assess the level of correlation). Results: Mean visual field indexes were -11.49 dB (MD), 5.85 dB (PSD), 74.2% (VFI). Compared to the DDLS, all systems had slight agreement (K < 0.2). The disagreement percentage was highest in eGSS (70%) then followed by mGSS (50%) and HPA (48.9%). In terms of definitive diagnosis, HPA and mGSS showed a substantial agreement (K > 0.6) with AGIS which is higher than that between eGSS and AGIS (retrospectively 0.773 and 0.75 vs 0.399) with p < 0.001. In terms of staging detection, the agreement between HPA &mGSS and AGIS was substantial (K > 0.6) while that between eGSS and AGIS was fair (K < 0.4). Conclusion: mGSS and HPA tend to show the stronger agreement with standard classifications than eGSS. mGSS should be used in clinical practice and research.

Dedifferentiated liposarcoma (DDLS) is characterized at the molecular level by amplification of genes within 12q13-15 including MDM2 and CDK4 . However, other than FNCLCC grade, prognostic markers are limited. We aim to identify molecular prognostic markers for DDLS to help risk stratify patients. To this end, we studied 49 cases of DDLS in our institutional archives and performed cytogenomic microarray analysis on 47 cases. Gene copy numbers for 12 loci were evaluated and correlated with outcome data retrieved from our institutional electronic medical records. Using cut point analysis and comparison of Kaplan-Meier survival curves by log rank tests, high amplification levels of MDM2 (>38 copies) and CDK4 (>30 copies) correlated with decreased disease free survival (DFS) ( P = . 0168 and 0.0169 respectively) and disease specific survival (DSS) ( P = . 0082 and 0.0140 respectively). Additionally, MDM2 and CDK4 showed evidence of a synergistic effect so that each additional copy of one enhances the effect on prognosis of each additional copy of the other for decreased DFS ( P = . 0227, 0.1% hazard). High amplification of JUN (>16 copies) also correlated with decreased DFS ( P = . 0217), but not DSS. The presence of copy number alteration at 3q29 correlated with decreased DSS ( P = . 0192). The presence of >10 mitoses per 10 high power fields and FNCLCC grade 3 also correlated with decreased DFS ( P = . 0310 and 0.0254 respectively). MDM2 and CDK4 gene amplification levels, along with JUN amplification and copy alterations at 3q29, can be utilized for predicting outcome in patients with DDLS. [ABSTRACT FROM AUTHOR]

Purpose: To assess the correlation between the disc damage likelihood scale (DDLS) objectively measured by a non-mydriatic fundus camera, Heidelberg Retina Tomograph 3, and optic coherence tomography in preperimetric glaucoma. Methods: One-hundred-twenty-five patients with preperimetric primary open-angle glaucoma (POAG) and pseudoexfoliation glaucoma (n=30) were included (mean age 58.9±15.9 years). All three devices graded the optic disc topography: Diagnosis 1 was defined as "outside normal limits", while Diagnosis 2 as "borderline or outside normal limits". Results: For Diagnosis 1, a significant correlation was shown between DDLS and Moorfields regression analysis (P=0.022), and for Diagnosis 2 with glaucoma probability score analysis (P=0.024), in POAG. In pseudoexfoliation glaucoma, DDLS did not correlate significantly with Heidelberg Retina Tomograph 3 and optic coherence tomography. Regarding the area under the curve the highest predictive power was demonstrated by the objective DDLS (0.513--0.824) compared to Burk (0.239--0.343) and Mikelberg (0.093--0.270) coefficients. Conclusions: The DDLS showed a significant correlation to the Moorfields regression analysis in preperimetric POAG. The objective DDLS showed the highest predictive power and thus is an additive tool in diagnosing preperimetric glaucoma. [ABSTRACT FROM AUTHOR]

Purpose: To report an image analysis pipeline, DDLSNet, consisting of a rim segmentation (RimNet) branch and a disc size classification (DiscNet) branch to automate estimation of the disc damage likelihood scale (DDLS)., Design: Retrospective observational., Participants: RimNet and DiscNet were developed with 1208 and 11 536 optic disc photographs (ODPs), respectively. DDLSNet performance was evaluated on 120 ODPs from the RimNet test set, for which the DDLS scores were graded by clinicians. Reproducibility was evaluated on a group of 781 eyes, each with 2 ODPs taken within 4 years apart., Methods: Disc damage likelihood scale calculation requires estimation of optic disc size, provided by DiscNet (VGG19 network), and the minimum rim-to-disc ratio (mRDR) or absent rim width (ARW), provided by RimNet (InceptionV3/LinkNet segmentation model). To build RimNet's dataset, glaucoma specialists marked optic disc rim and cup boundaries on ODPs. The "ground truth" mRDR or ARW was calculated. For DiscNet's dataset, corresponding OCT images provided "ground truth" disc size. Optic disc photographs were split into 80/10/10 for training, validation, and testing, respectively, for RimNet and DiscNet. DDLSNet estimation was tested against manual grading of DDLS by clinicians with the average score used as "ground truth." Reproducibility of DDLSNet grading was evaluated by repeating DDLS estimation on a dataset of nonprogressing paired ODPs taken at separate times., Main Outcome Measures: The main outcome measure was a weighted kappa score between clinicians and the DDLSNet pipeline with agreement defined as ± 1 DDLS score difference., Results: RimNet achieved an mRDR mean absolute error (MAE) of 0.04 (± 0.03) and an ARW MAE of 48.9 (± 35.9) degrees when compared to clinician segmentations. DiscNet achieved 73% (95% confidence interval [CI]: 70%, 75%) classification accuracy. DDLSNet achieved an average weighted kappa agreement of 0.54 (95% CI: 0.40, 0.68) compared to clinicians. Average interclinician agreement was 0.52 (95% CI: 0.49, 0.56). Reproducibility testing demonstrated that 96% of ODP pairs had a difference of ≤ 1 DDLS score., Conclusions: DDLSNet achieved moderate agreement with clinicians for DDLS grading. This novel approach illustrates the feasibility of automated ODP grading for assessing glaucoma severity. Further improvements may be achieved by increasing the number of incomplete rims sample size, expanding the hyperparameter search, and increasing the agreement of clinicians grading ODPs., (© 2022 Published by Elsevier Inc. on behalf of American Academy of Ophthalmology.)

Purpose: Accurate neural rim measurement based on optic disc imaging is important to glaucoma severity grading and often performed by trained glaucoma specialists. We aim to improve upon existing automated tools by building a fully automated system (RimNet) for direct rim identification in glaucomatous eyes and measurement of the minimum rim-to-disc ratio (mRDR) in intact rims, the angle of absent rim width (ARW) in incomplete rims, and the rim-to-disc-area ratio (RDAR) with the goal of optic disc damage grading., Design: Retrospective cross sectional study., Participants: One thousand and twenty-eight optic disc photographs with evidence of glaucomatous optic nerve damage from 1021 eyes of 903 patients with any form of primary glaucoma were included. The mean age was 63.7 (± 14.9) yrs. The average mean deviation of visual fields was -8.03 (± 8.59)., Methods: The images were required to be of adequate quality, have signs of glaucomatous damage, and be free of significant concurrent pathology as independently determined by glaucoma specialists. Rim and optic cup masks for each image were manually delineated by glaucoma specialists. The database was randomly split into 80/10/10 for training, validation, and testing, respectively. RimNet consists of a deep learning rim and cup segmentation model, a computer vision mRDR measurement tool for intact rims, and an ARW measurement tool for incomplete rims. The mRDR is calculated at the thinnest rim section while ARW is calculated in regions of total rim loss. The RDAR was also calculated. Evaluation on the Drishti-GS dataset provided external validation (Sivaswamy 2015)., Main Outcome Measures: Median Absolute Error (MAE) between glaucoma specialists and RimNet for mRDR and ARW., Results: On the test set, RimNet achieved a mRDR MAE of 0.03 (0.05), ARW MAE of 31 (89)°, and an RDAR MAE of 0.09 (0.10). On the Drishti-GS dataset, an mRDR MAE of 0.03 (0.04) and an mRDAR MAE of 0.09 (0.10) was observed., Conclusions: RimNet demonstrated acceptably accurate rim segmentation and mRDR and ARW measurements. The fully automated algorithm presented here would be a valuable component in an automated mRDR-based glaucoma grading system. Further improvements could be made by improving identification and segmentation performance on incomplete rims and expanding the number and variety of glaucomatous training images., (© 2022 Published by Elsevier Inc. on behalf of American Academy of Ophthalmology.)

Abstract: The goal of the treatment of patients with glaucoma is to prevent disability or, if disability already exists, to repair the disability or at the least to prevent further disability from developing. To achieve these goals requires knowing what will happen if there is no treatment and what will happen if there is treatment. That is, one must know the possible benefits from the therapy in comparison to the damage caused by the therapy. As we will demonstrate, the classic risk factors do not provide accurate estimates of the development of disability and do not answer these two questions. The most helpful clues are provided by what the patient says, by whether the disc is damaged and by whether the disc is deteriorating. The severity of disease is best estimated by the nature of the optic disc and how it changes. This report is primarily focused on increasing the skill of the physician in being able to estimate the nature of the optic disc and how that changes. This does not, however, lessen the tremendous importance of careful history-taking and of gonioscopy. In this report, however, we focus on the evaluation of the disc. Currently disc evaluation is often not adequate because of poor examination techniques, reliance on cup/disc ratios, and reliance on the results of image analyzers. We will present here the Disc Damage Likelihood Scale, which is a user-friendly method which correlates accurately with visual field changes. [Copyright &y& Elsevier]

Background: Differentiating large lipomas from atypical lipomatous tumors (ALT) is challenging, and preoperative management guidelines are not well defined. The diagnostic ambiguity leads many surgeons to refer all patients with large lipomatous masses to an oncologic specialist, perhaps unnecessarily., Study Design: In this retrospective cohort study of patients with nonretroperitoneal lipomatous tumors, preoperative characteristics discernible without invasive diagnostic procedures were evaluated for diagnostic predictive value., Results: We identified 319 patients (256 with lipomas, 63 with ALTs) treated between 1994 and 2012. Patients with ALTs were older (60.5 vs 53.5 years, p < 0.0001), had larger tumors (16.0 vs 8.3 cm, p < 0.0001), had tumors more often located on an extremity (88.9% vs 60.5% torso, p < 0.0001), and more frequently had a history of previous operations at the same site, exclusive of excision leading to diagnosis and referral (20.6% vs 5.9%, p = 0.001). Local recurrence was observed in 2 patients with lipomas (0.8%) vs 14 with ALTs (22.6%, p < 0.0001). No patients with ALTs developed distant metastases or disease-specific mortality, with a median follow-up of 27.4 months (range 0 to 164.6 months). On multivariate analysis, age ≥ 55 years, tumor size ≥ 10 cm, extremity location, and history of previous resections were predictors for diagnosis of ALT (p < 0.05)., Conclusions: Characteristics of lipomatous masses associated with a diagnosis of ALT include patient age ≥ 55 years, tumor size ≥ 10 cm, previous resection, and extremity location (vs torso). These easily identifiable traits may guide surgical management or referral to a specialist., (Copyright © 2013 American College of Surgeons. Published by Elsevier Inc. All rights reserved.)