Adaptive designs in multi-reader multi-case clinical trials of imaging devices

Area of Science:

• Medical Imaging
• Biostatistics
• Clinical Trial Design

Background:

• Multiple-reader, multiple-case (MRMC) studies are crucial for evaluating medical imaging devices.
• Accurate sample size determination in MRMC studies is challenging due to reader and case variability.
• Existing adaptive methods for clinical trials do not fully address the complexities of MRMC data.

Purpose of the Study:

• To develop adaptive design methodologies for MRMC studies enabling study resizing.
• To achieve a target power and control the type I error rate when comparing diagnostic performance (AUC) of two imaging modalities.
• To address the correlation structure inherent in MRMC data during adaptive sizing.

Main Methods:

• Developed adaptive MRMC design methodologies for study resizing.
• Simultaneously adjusted study size and critical value for hypothesis testing after an interim analysis.
• Derived analytical results and conducted simulations to validate the methods.

Main Results:

• The proposed adaptive methods effectively control the type I error rate close to the nominal level.
• Simulations demonstrated that study power is adjusted towards the target value under various conditions.
• The methodology was successfully applied to a real-world case comparing breast cancer detection imaging modalities.

Conclusions:

• Adaptive MRMC design methodologies provide a robust framework for optimizing study size and power.
• These methods enhance the reliability of diagnostic performance evaluations in medical imaging.
• The developed techniques offer a valuable tool for future clinical trials involving MRMC designs.