검색 상세

Standardizing Medical Imaging Data Using Radiology Common Data Model and Analysis of Retinal Thickness in Patients with Chronic Diseases with Standardized Optical Coherence Tomography Databases

  • 박철형 (아주대학교 일반대학원)

초록/요약

The Observational Medical Outcomes Partnership - Common Data Model (OMOP- CDM) is widely used for standardizing electronic medical records (EMR) across institutions globally. However, OMOP-CDM cannot standardize unstructured data like medical imaging, which limits integrating such data in collaborative, multi- institutional research efforts. To address this, the Radiology Common Data Model (R-CDM) was developed, allowing for the harmonization of imaging data alongside clinical records. This study seeks to showcase the efficacy of standardizing Optical Coherence Tomography (OCT) data using the R-CDM format to enable efficient and large-scale studies on retinal thickness changes in patients suffering from chronic diseases. For this study, OCT data from two major tertiary hospitals were standardized using R-CDM. The objective was to compare retinal thickness between patients with chronic diseases and those without. Individuals diagnosed with retinal or choroidal conditions that could affect retinal thickness were excluded. Clinical cohorts were created using OMOP-CDM, with 1:2 propensity score matching (PSM) applied for comparability. These cohorts were then linked to the R-CDM to extract relevant OCT imaging data to analyze central macular thickness (CMT) and retinal nerve fiber layer (RNFL) thickness. Finally, retinal thickness was analyzed using a linear mixed- effects model (LMM) to account for confounding variables such as age and sex. This model also adjusted for repeated measurements, ensuring that the variability in longitudinal data was properly addressed. Overall, 261,874 OCT images from Ajou University Medical Center (AUMC) and 475,626 images from Seoul National University Bundang Hospital (SNUBH) were standardized into the R-CDM format. Following PSM, the type 2 diabetes mellitus (T2DM) cohort included 957 patients, with the control group comprising 1,603 individuals. Statistical analysis revealed significant reductions in CMT in the T2DM groups from both institutions (P = 0.04 and P < 0.01, respectively), though no notable changes in RNFL thickness were observed (P = 0.56 and P = 0.39). Additionally, a significant reduction in CMT was noted in the hypertension (HTN) cohort at AUMC when compared to controls, while no other significant changes in retinal thickness were detected in the remaining analyses. This research highlights the utility of R-CDM in overcoming the limitations of current data integration practices, particularly when it comes to combining clinical and imaging datasets across institutions. By leveraging both OMOP-CDM and R- CDM, this study demonstrates that it is possible to conduct efficient, large-scale, multi-institutional research involving both clinical and imaging data, particularly in the context of long-standing chronic diseases. The findings underscore the potential for improving research methodologies and collaboration within the healthcare research community, opening the door for more comprehensive studies that integrate diverse types of data across multiple sites. This framework may serve as a model for future research initiatives, ensuring the standardization and availability of imaging data for robust clinical investigations. By utilizing the combined power of OMOP-CDM and R-CDM, researchers can draw on extensive, harmonized datasets, ultimately enabling more precise assessments of the impact of chronic diseases on retinal health, and facilitating the advancement of medical knowledge in this crucial area.

more

목차

I. Introduction 1
A. Background 1
1. Observational Medical Outcome Partnership Common Data Model 1
2. Efforts to Standardize Unstructured Data Using the OMOP-CDM Framework 3
3. Challenges in Utilizing Multi-Institutional Medical Imaging Data 4
4. The Potential Impact of R-CDM 6
5. Optical Coherence Tomography (OCT) 6
B. Objectives 9
II. Materials and Methods 10
A. Data sources 10
B. Designing a standard structure and terminology system for R-CDM 11
C. Constructing an R-CDM standardized OCT database 18
D. Study Design Using OMOP-CDM: Establishing Chronic Disease and Control Cohorts for OCT Imaging Analysis 21
E. Integration of R-CDM and OMOP-CDM: Extracting Specific OCT Data from Defined Cohorts 26
F. Analyzing Retinal Thickness Measurements from OCT Data Using Mixed-Effects Regression Models 28
III. Results 31
A. Composition of R-CDM Standardized OCT Databases 31
B. Study population 32
C. Clinical Outcomes 40
D. Sensitivity analyses 49
IV. Discussion 50
A. Main findings 50
1. The Global Impact and Application of R-CDM in Medical Imaging Research 51
2. Other Efforts to Standardize Medical Imaging Data 53
3. Retinal Thickness Alterations in Chronic Disease: A Comparative Review of Diabetic and Hypertensive Cohorts 56
4. Application of Large-Scale PSM and Multi-Institutional Analysis for Retinal Thickness Studies 57
B. Limitations 59
V. Conclusion 60
References 62
Appendix 68
국문요약 102

more
반출 Meta View 목록