Organ sample generator for expected treatment dose construction and adaptive inverse planning optimization

Area of Science:

• Medical Physics
• Radiotherapy
• Computational Biology

Background:

• Adaptive radiotherapy requires accurate modeling of organ variations.
• Patient-specific organ motion significantly impacts treatment planning and dose delivery.
• Current methods may not fully capture dynamic organ changes during treatment.

Purpose of the Study:

• To develop an organ sample generator (OSG) for adaptive radiotherapy.
• To enable patient-specific organ variation modeling for dose construction.
• To optimize adaptive inverse planning using generated organ samples.

Main Methods:

• Utilized principal component analysis (PCA) and time-varying least-squares regression (LSR).
• Analyzed daily volumetric images from 11 head and neck cancer patients.
• Modeled geometric variations of 18 organs (15 organs at risk, 3 targets).

Main Results:

• Organ variations during radiotherapy can be represented by 3-4 principal eigenvectors.
• The OSG accurately generates samples from patient-specific organ variation probability density functions (PDFs).
• Estimated PDF accuracy improved recursively with daily image feedback, with deviations <2mm and <1mm for mean and std dev after two weeks.

Conclusions:

• A patient-specific OSG is feasible for adaptive radiotherapy.
• The OSG supports accurate treatment dose construction and adaptive inverse planning.
• Continuous improvement of OSG accuracy is possible with daily volumetric image feedback.