SU-E-J-189: The Kullback-Leiber Divergence for Quantifying Changes in Radiotherapy Treatment Response

Area of Science:

• Medical Imaging
• Radiotherapy
• Computational Analysis

Background:

• Repeated imaging is crucial in radiotherapy for tumor detection and treatment assessment.
• Change detection algorithms from remote sensing can quantify modifications in time-series data.
• Automated quantification of clinical changes in repeated radiotherapy imaging is needed.

Purpose of the Study:

• To propose and evaluate change detectors for automated quantification of clinical changes in repeated radiotherapy imaging.
• To explore the Kullback-Leiber divergence as an indicator of tumor change and treatment response.

Main Methods:

• Utilized Kullback-Leiber divergence, based on likelihood theory, to measure statistical distribution differences.
• The method accommodates noise and variations in imaging acquisition parameters.
• Compared Kullback-Leiber divergence with simple difference maps for change detection.

Main Results:

• Kullback-Leiber divergence effectively discerned noise by considering regional voxel statistics.
• Unlike difference maps, it consistently detected low-intensity and high-contrast changes.
• The proposed operator marked both low and high contrast changes accurately.

Conclusions:

• Statistical comparison using Kullback-Leiber divergence reliably quantifies changes in repeated radiotherapy imaging.
• This method enhances automated assessment of tumor response and personalized treatment.