ROC Curve & Diagnostic Performance Analysis
Enter the gold standard (0=negative / 1=positive) and a continuous measured value, and it automatically computes AUC (with 95% CI), draws the ROC curve, and gives the optimal cut-off by the Youden index with the corresponding sensitivity, specificity, predictive values, and likelihood ratios. A core tool for imaging diagnostic-performance studies.
① Input data
One record per row: gold standard (0 or 1) + measured value, separated by space/Tab/comma. You can copy two columns from Excel.
How to use & methodology
What can ROC curve analysis do?
The ROC (receiver operating characteristic) curve evaluates how well a continuous indicator distinguishes two states (e.g. benign vs malignant, diseased vs not). The closer the area under the curve (AUC) is to 1, the stronger the discrimination. In imaging studies it is often used to assess the diagnostic value of a quantitative indicator (e.g. ADC value, enhancement rate).
What AUC counts as valuable?
Usually 0.5 means no discrimination (like a coin flip), 0.7–0.8 is moderate, 0.8–0.9 good, and above 0.9 excellent. Report the AUC with its 95% CI; if the CI's lower bound is near or below 0.5, the diagnostic value is uncertain.
How is the optimal cut-off determined?
This tool selects the optimal cut-off by maximising the Youden index (sensitivity + specificity − 1). But the clinical choice should weigh the costs of missed vs false diagnoses — screening favours high sensitivity, confirmation favours high specificity — so it need not be the Youden-maximising point.
How do I prepare the data?
One record per row: the first column is the gold standard (0 = negative/normal, 1 = positive/diseased), the second is the subject's measured value. If the indicator is 'lower suggests abnormal' (as in some ADC scenarios), switch the decision direction. The positive and negative groups each need at least 2 cases.