Abstract: Objective To reconstruct the original three-dimensional conformation of tumor resection tissue through the study of breast-conserving surgery  excision specimens for part-mount sub-serial section  and pathological three-dimensional (3D) reconstruction, to establish a new margin assessment model, and to guide tumor bed delineation individually for radiotherapy. Methods From February 2016 to February 2017， thirty-three eligible breast cancer patients underwent breast-conserving surgery in Breast Cancer Center of Shandong Cancer Hospital were recruited. The excision specimens were prepared with part-mount sub-serial section, and residual tumors were microscopically outlined, scanned and registered by Photoshop software. The 3D model of residual tumors was reconstructed with 3DDOCTOR software to evaluate margin status and record pathological type, tumor length and 3D negative margin distance. The gross tumor volume (GTV) was delineated based on clips placed in the lumpectomy cavity. CTV1 and CTV2 were defined by adding uniform 1.00 cm and 1.50 cm margin based on GTV respectively. CTV3 and CTV4 were defined by adding 1.00 cm and 1.50 cm margin based on 3D boundary of excision tumor respectively, and compared the volume differences of CTV1 and CTV3, CTV2 and CTV4. Results Based on the marginal assessment results of 3D pathological reconstruction, the rates of false negatives during the intraoperative rapid pathological examination and postoperative routine pathological margin evaluation were 6.7% (2/30) and 3.4% (1/29) respectively. The pathological type of pathological large slice and routine pathological examination was consistent with rate of 93.9%（31/33). The M(QR) tumor lengths of routine pathological and pathological 3D reconstruction were 1.90 (1.502.40)cm and 2.00 (1.602.70)cm respectively, with statistical difference between the two groups (Z=-2.438,P=0.015). The M(QR) volumes for CTV1, CTV2, CTV3, CTV4 were 70.76 (49.8478.07)cm3, 110.11 (83.38126.17）cm3, 23.85 (16.4631.49）cm3 and 38.74 (30.4750.58）cm3 respectively. There were statistical differences between CTV1 and CTV3, CTV2 and CTV4 （Z=-4.372, P<0.001; Z=-4.372, P<0.001）. Conclusion The application of pathological 3D reconstruction technology can largely compensate for the shortcomings of the traditional margin assessment model, make the decisions of adjuvant treatment after breastconserving surgery  more accurate, and guide the tumor bed delineation individually for radiotherapy.

Key words: Breast neoplasms, Radiotherapy, Imaging, three-dimensional, Target delineation