A tensor-based population value decomposition to explain rectal toxicity after prostate cancer radiotherapy

Area of Science:

• Radiation Oncology
• Medical Imaging Analysis
• Biostatistics

Background:

• The relationship between radiation dose distribution and side effects in prostate cancer radiotherapy remains unclear.
• Understanding these associations is crucial for minimizing patient toxicity.
• Current methods lack the ability to effectively compare dose distributions across patient populations.

Purpose of the Study:

• To develop and validate a tensor-based method for population analysis of dose distributions in prostate cancer radiotherapy.
• To identify specific irradiated zones correlated with rectal bleeding.
• To elucidate dose patterns associated with rectal toxicity.

Main Methods:

• A tensor-based approach was generalized from 2D image analysis to compare dose distributions.
• The method was applied to a cohort of 63 prostate cancer patients.
• Population analysis was performed by comparing dose distributions between patients with and without rectal bleeding.

Main Results:

• The proposed method successfully highlighted over-irradiated zones.
• A distinct dose pattern was identified that characterizes patients experiencing rectal bleeding.
• The analysis revealed correlations between specific dose distributions and rectal toxicity.

Conclusions:

• The tensor-based method is effective for population analysis of dose distributions in radiotherapy.
• This approach can help elucidate dose-response relationships for rectal toxicity in prostate cancer patients.
• Identifying high-risk dose patterns can inform treatment planning to reduce side effects.