Comparison of Auto-Contouring Tools for Delineation of Normal Organs at Risk in Paediatric Patients Undergoing...

Area of Science:

• Radiation Oncology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Manual contouring of organs at risk (OARs) in paediatric patients undergoing cranial-spinal radiation therapy (CSI) is time-consuming.
• Automated contouring tools aim to improve efficiency and consistency in radiation therapy planning.

Purpose of the Study:

• To evaluate the accuracy and clinical acceptability of auto-contours generated by Siemens DirectORGANS software.
• To assess the software's performance on paediatric patient datasets for CSI treatment.

Main Methods:

• Retrospective analysis of 20 paediatric CSI patient datasets.
• Comparison of auto-contours with manual contours using quantitative (Dice, Hausdorff, MDA) and qualitative (Likert, Turing test) methods.

Main Results:

• 72.8% of auto-contours were clinically acceptable without edits, compared to 91.7% for manual contours.
• Auto-contours performed poorly on low-contrast structures (breasts, oesophagus, brainstem) and brain structures.
• Liver auto-contours outperformed manual contours.

Conclusions:

• Siemens DirectORGANS software shows promise for paediatric CSI contouring but requires critical assessment and editing.
• The software, designed for adults, can be cautiously applied to paediatric patients.
• Further validation is recommended for specific OARs and patient populations.