Conference Program

Return to main conference page

All Times EDT

Wednesday, September 21
Wed, Sep 21, 4:15 PM - 5:30 PM
Salon FG
Assess, Analyze, and Incorporate Reader Variation in Studies with Medical Images

Study Design and Statistical Considerations for a Clinical Study Involving Readers in Medical Imagings (303702)

View Presentation

*Jessie Moon, FDA 

Keywords: Medical imaging, Reader study, SaMD, Oncology, MRMC, AI/ML, CAD

Medical imaging devices, including software devices and computer-aided devices, are often used for diagnosis, assessing treatment effects, or providing useful information to clinicians in standard care. However, validating the performance of these devices in a clinical study can be challenging because their performance is often estimated based on interpretations from a panel of readers. Not only these readers’ readings are subjective and have variability, but also various biases can be introduced in the clinical study when readers interpret the images. Therefore, a clinical study needs to be carefully designed and analysis needs to be carefully done. I will discuss study design and statistical considerations when performing a clinical study involving readers.