Interobserver Variability in a Spanish Society of Radiation Oncology (SEOR) Head and Neck Course. Is Current Contouring Training Sufficient?

Abstract

Purpose

External beam radiation therapy has grown significantly, incorporating advanced techniques like intensity modulation or stereotactic treatments, which enhance precision and accuracy. Nevertheless, variability in target volume delineation by radiation oncologists remains a challenge, influencing dose distribution. This study analyzes an online training course by the Spanish Society of Radiation Oncology, focusing on head and neck tumor contouring, to evaluate interobserver variability.

Material and Methods

Eight instructors provided clinical directives for 8 head and neck pathologies. Participants contoured structures using their own treatment planning systems, emphasizing gross tumor volume and high-, medium-, and low-risk clinical target volumes (CTV) contouring. Delineation variability was evaluated using the Dice similarity coefficient and volume relative change.

Results

The results reveal significant variability in contouring, with mean Dice similarity coefficient values ranging from 0.57 to 0.69. High-risk CTV demonstrated higher variability compared with medium-risk CTV. The presence of a gross tumor volume and supporting positron emission tomography/computed tomography or magnetic resonance imaging studies did not significantly improve the concordance. Parotid cases exhibited the greatest differences.

Conclusions

Despite the introduction of new automatic tools, this study points to the need for uniform contouring criteria. Training and standardization efforts are essential to enhance radiation therapy treatment consistency and quality.