Comparability of two fundus photograph reading centers in grading cytomegalovirus retinitis progression

Fundus image diagnostic agreement in uveitis utilizing free and open source software

OBJECTIVE

To assess the adequacy of image agreement regarding uveitis based on color fundus and fluorescein angiography images alone, and to use free and open source applications to conduct an image agreement study.

DESIGN

Cross-sectional agreement study.

PARTICIPANTS

Baseline fundus and fluorescein images of patients with panuveitis, posterior, or intermediate uveitis enrolled in the Multi-center Uveitis Steroid Treatment (MUST) trial.

METHODS

Three fellowship-trained specialists in uveitis independently reviewed patient images using ClearCanvas™ and responded using Epi Info™. The diagnoses of the 3 reviewers were compared with the MUST clinician as a gold standard. A rank transformation adjusted for the possible variation in number of responses per patient. Chance-corrected interobserver agreement among the 3 reviewers was estimated with the ι coefficient. Confidence interval (CI) and SE were bootstrapped.

RESULTS

Agreement between the diagnoses of the respondents and the baseline MUST clinician's diagnosis was poor across all diagnostic categories, ι = 0.09 (95% CI, 0.07-0.11). The agreement among respondents alone also was poor, ι = 0.11 ± 0.02 (95% CI, 0.08-0.13). The specialists requested more patient historical and clinical information to make a diagnosis on all patients.

CONCLUSIONS

The role in distinguishing the multiple conditions in uveitis appears to be limited when based on fundus imaging alone. Future studies should investigate different categories of clinical data to supplement image data. Freely available applications have excellent utility in ophthalmic imaging agreement studies.

A domain model of a clinical reading center - Design and implementation

In clinical trials huge amounts of raw data are generated. Often these data are submitted to reading centers for being analyzed by experts of that particular type of examination. Although the installment of a reading center can raise the overall quality, they also introduce additional complexity to the management and conduction of a clinical trial. Software can help to handle this complexity. Domain-driven-design is one concept to tackle software development in such complex domains. Here we present our domain model for a clinical reading center, as well as its actual implementation utilizing the Nuxeo enterprise content management system.

Cytomegalovirus retinitis and the acquired immunodeficiency syndrome--bench to bedside: LXVII Edward Jackson Memorial Lecture

PURPOSE

To update information on cytomegalovirus (CMV) retinitis in patients with acquired immunodeficiency syndrome (AIDS) and to integrate information on its pathogenesis and clinical outcomes.

DESIGN

Literature review.

METHODS

Selected articles from the medical literature, particularly large epidemiologic studies, including the Johns Hopkins Cytomegalovirus Retinitis Cohort Study, the Longitudinal Study of the Ocular Complications of AIDS, and the Cytomegalovirus Retinitis and Viral Resistance Study, were reviewed. Clinical information is discussed in light of knowledge on CMV, its pathogenesis, and its interactions with human immunodeficiency virus (HIV).

RESULTS

Cytomegalovirus uses several mechanisms to evade the immune system and establish latent infection in immunologically normal hosts. With immune deficiency, such as late-stage AIDS, CMV reactivates, is disseminated to the eye, and establishes a productive infection, resulting in retinal necrosis. HIV and CMV potentiate each other: CMV accelerates HIV disease, and CMV retinitis is associated with increased mortality. Randomized clinical trials have demonstrated the efficacy of treatments for CMV retinitis. Systemically administered treatment for CMV retinitis decreases AIDS mortality. Highly active antiretroviral therapy (HAART) effectively suppresses HIV replication, resulting in immune recovery, which, if sufficient, controls retinitis without anti-CMV therapy. Resistant CMV, detected in the blood, correlates with resistant virus in the eye and is associated with worse clinical outcomes, including mortality. Host factors, including host genetics and access to care, play a role in the development of CMV retinitis.

CONCLUSIONS

Clinical outcomes of CMV retinitis in patients with AIDS are dependent on characteristics of the virus and host and on HIV-CMV interactions.

Evaluation of optical coherence tomography findings in age-related macular degeneration: a reproducibility study of two independent reading centres

BACKGROUND/AIMS

To determine the reproducibility among readers of two independent certified centres, the Vienna Reading Center (VRC) and the University of Wisconsin-Madison Reading Center (UW-FPRC) for optical coherence tomography (OCT) images in age-related macular degeneration (AMD).

METHODS

Fast macular thickness scans and 6 mm cross hair scans were obtained from 100 eyes with all subtypes of AMD using Stratus OCT. Consensus readings were performed by two certified OCT readers of each reading center using their grading protocol. Common variables of both grading protocols, such as presence of cystoid spaces, subretinal fluid, vitreomacular traction and retinal pigment epithelial detachment, were compared using κ statistics. In addition, the intraclass correlation coefficient (ICC) was calculated for centre point thickness (CPT) of values re-measured manually in the presence of alignment errors.

RESULTS

The reproducibility was dependent on the variable measured with a κ value of 0.81 for the presence of cystoid spaces, 0.78 for the presence of subretinal fluid and 0.795 for the presence of vitreomacular traction. The lowest reproducibility was found for the presence of retinal pigment epithelial detachment with a κ value of 0.51. The CPT was re-measured in 29 out of 100 scans at both sites with an ICC of the re-measured thicknesses of 0.92.

CONCLUSION

OCT scan data are crucial in monitoring treatment efficacy in AMD clinical trials. For comparison of results obtained by different reading centers, the inter-reading center reproducibility is essential. Although the reproducibility is generally high, the reliability depends on the selected morphological parameters.