An index for diagnostic accuracy in the multiple disease setting

Area of Science:

• Medical Imaging Analysis
• Diagnostic Accuracy Assessment
• Receiver Operating Characteristic (ROC) Curve Analysis

Background:

• Evaluating diagnostic accuracy in clinical settings requires assessing performance in patients with multiple co-existing abnormalities.
• Traditional Receiver Operating Characteristic (ROC) curves are effective for single diseases but not easily generalizable to complex cases with multiple abnormalities.

Purpose of the Study:

• To propose and describe a novel measure for diagnostic accuracy that extends the concept of the area under the ROC curve.
• To provide a method for evaluating imaging system performance in patients presenting with multiple abnormalities.

Main Methods:

• The proposed diagnostic accuracy measure is a weighted average incorporating areas under individual ROC curves for single diseases.
• It also includes components representing areas under ROC curves specifically constructed for patients with multiple diseases.
• The method discusses various options for scoring the presence of abnormalities in cases with two or more conditions.

Main Results:

• The developed methods for estimating diagnostic accuracy were demonstrated using a dataset.
• This dataset included a significant proportion (over one third) of abnormal cases with multiple chest disease abnormalities.
• The application showed the practical utility of the proposed method.

Conclusions:

• An accessible method is presented for estimating diagnostic accuracy in the context of multiple abnormalities.
• This new approach facilitates the inclusion of complex cases with multiple abnormalities in the assessment of imaging system diagnostic accuracy.
• It enhances the comprehensive evaluation of medical imaging technologies.