Voxel-based analysis in radiation oncology: A methodological cookbook

Area of Science:

• Radiation Oncology
• Medical Physics
• Radiotherapy Research

Background:

• Traditional Dose Volume Histogram (DVH) methods analyze radiation dose based on organ-level statistics.
• Pixel- or voxel-based (VB) methods offer a more localized approach to dose distribution analysis.
• VB methods are emerging in radiation oncology, inspired by neuroimaging techniques.

Purpose of the Study:

• To critically review methods for evaluating radiation therapy toxicity using spatial dose distribution information.
• To discuss the challenges and solutions in building and comparing dose maps for toxicity analysis.
• To explore novel phenomenological approaches for Normal Tissue Complication Probability (NTCP) modeling.

Main Methods:

• Review of existing literature on pixel- or voxel-based (VB) dose distribution analysis in radiation oncology.
• Discussion of spatial normalization techniques for comparing dose distributions across patients.
• Examination of statistical analysis options and potential pitfalls for VB dose map comparison.

Main Results:

• VB methods enable comparison of spatial dose patterns between patients with and without toxicity.
• Critical aspects include spatial normalization of dose distributions and robust statistical analysis.
• Practical applications illustrate the theoretical and technical issues in VB dose map analysis.

Conclusions:

• VB methods provide a powerful framework for understanding localized dose-response relationships in radiation therapy.
• Addressing challenges in dose map construction and analysis is crucial for clinical implementation.
• Future directions include phenomenological NTCP models accounting for inhomogeneous organ radiosensitivity.