Inflammatory demyelinating diseases of the central nervous system, including multiple sclerosis (MS), neuromyelitis optica spectrum disorder (NMOSD), and myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD).
Conditions
Interventions
Gold Standard:1. Upon signing by the principal investigator, this form is deemed to indicate the investigator's commitment that the research project does not involve personal privacy or commercial int
Index test:Brain magnetic resonance imaging
Sponsors
The first affiliated hostipal of nanchang university
Eligibility
Sex/Gender
All
Age
18 Years to 75 Years
Inclusion criteria
Inclusion criteria: 1. Complete conventional MRI examination at the study site, with imaging data available for post-de-identification analysis. 2. Have a definite clinical diagnosis or diagnostic evidence that can serve as background/simulated disease category; diagnosis of MS, NMOSD, and MOGAD should be based on international or domestic diagnostic criteria applicable during the study period and comprehensive clinical judgment. 3. MRI sequence quality meets the minimum requirements for model training, validation, or physician reading evaluation, and at least includes the routine non-contrast MRI sequences required for the study. 4. Necessary clinical data must be obtainable, including diagnosis, examination date, basic information such as age/sex, and data related to model label determination.
Exclusion criteria
Exclusion criteria: 1. Severe artifacts, missing sequences, or format abnormalities in MRI images that prevent standardized preprocessing or image interpretation. 2. Missing key clinical diagnostic information, or diagnosis category cannot be confirmed after investigator review. 3. Repeated examinations cannot confirm independence, or multiple examinations from the same subject do not meet the preset inclusion rules. 4. Other conditions that, in the investigator's judgment, may significantly affect the reliability of imaging labels.
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| AUC, Accuracy, Sensitivity, Specificity, and F1 Score for AI Model Discrimination of MS, NMOSD, MOGAD vs. Similar Diseases in Target Case Identification Task;Overall Accuracy, Class-wise Sensitivity, Specificity, Weighted Average F1 Score, and External Validation Performance of AI Model in Multiclass Differential Diagnosis Task for MS, NMOSD, and MOGAD; | — |
Secondary
| Measure | Time frame |
|---|---|
| Radiologist Read Evaluation: Diagnostic Accuracy, Confidence Scores, Reading Time, and Between-Group Differences Before and After AI Assistance for Junior and Senior Radiologists;Enhanced Prediction Task: AUC, Accuracy, Sensitivity, Specificity, and F1 Score for Predicting Active vs. Non-active Disease Status Based on Non-enhanced MRI;Explainability Metrics: Overlap Rate between SHAP Heatmaps and Expert Diagnostic Rationale, Precision-like Alignment, Recall-like Alignment, and F1-like Alignment; | — |
Countries
China
Contacts
Public ContactWu Lin
The first affiliated hostipal of nanchang university
Outcome results
None listed