Interobserver variability in the delineation of the tumour bed using seroma and surgical clips based on 4DCT scan for external-beam partial breast irradiation

BACKGROUND: To explore the interobserver variability in the delineation of the tumour bed using seroma and surgical clips based on the four-dimensional computed tomography (4DCT) scan for external-beam partial breast irradiation (EB-PBI) during free breathing. METHODS: Patients with a seroma clarity score (SCS) 3 ~ 5 and ≥5 surgical clips in the lumpectomy cavity after breast-conserving surgery who were recruited for EB-PBI underwent 4DCT simulation. Based on the ten sets of 4DCT images acquired, the tumour bed formed using the clips, the seroma, and both the clips and seroma (defined as TB(C), TB(S) and TB(C+S), respectively) were delineated by five radiation oncologists using specific guidelines. The following parameters were calculated to analyse interobserver variability: volume of the tumour bed (TB(C), TB(S), TB(C+S)), coefficient of variation (COV(C), COV(S), COV(C+S)), and matching degree (MD(C), MD(S), MD(C+S)). RESULTS: The interobserver variability for TB(C) and TB(C+S) and for COV(C) and COV(C+S) were statistically significant (p = 0.021, 0.008, 0.002, 0.015). No significant difference was observed for TB(S) and COV(S) (p = 0.867, 0.061). Significant differences in interobserver variability were observed for MD(C) vs MD(S), MD(C) vs MD(C+S), MD(S) vs MD(C+S) (p = 0.000, 0.032, 0.008), the interobserver variability of MD(S) was smaller than that of MD(C) and MD(C+S) (MD(S) > MD(C+S) > MD(C)). CONCLUSIONS: When the SCS was 3 ~ 5 points and the number of surgical clips was ≥5, interobserver variability was minimal for the delineation of the tumour bed based on seroma.