PURPOSE: Usefulness of three dimensional (3D) multidetector-row CT (MDCT) images for preoperative evaluation of tumor extension was studied in primary breast cancer patients. METHODS: 3D-MDCT tumor images of 143 tumors in 143 patients with primary breast cancer were created with the volume rendering method. The transverse tumor size (TS) and vertical tumor size (VS) were then measured in an anterior-posterior view of the 3D-MDCT images. The pathological tumor size was determined according to a map of the tumor spread prepared by pathologists using multi-sliced (3-5 mm intervals) surgical specimens and compared with the tumor size on 3D-MDCT images. RESULTS: First, the optimal method for creating 3D-MDCT tumor images was determined for the first 40 patients (learning set), resulting in a fairly good correlation of tumor size on 3D-MDCT images with pathological tumor size (r = 0.983 for TS and r = 0.958 for VS). We then carried out a validation study on the next 103 patients (validation set). The 3D-MDCT tumor size's strong correlation with the pathological tumor size demonstrated a high rate of accuracy (r = 0.974 for TS and r = 0.977 for VS). Subset analyses according to histological type showed that correlation coefficients were r = 0.979 for TS and r = 0.981 for VS of invasive ductal carcinomas (n = 88), r = 0.948 for TS and r = 0.970 for VS of ductal carcinomas in situ (n = 10), and r = 0.984 for TS and r = 0.976 for VS of invasive lobular carcinomas (n = 5). CONCLUSION: 3D-MDCT images can assess breast cancer tumor extension highly accurately, and thus seems to be useful for planning the extent of resection in breast conserving surgery.
PURPOSE: Usefulness of three dimensional (3D) multidetector-row CT (MDCT) images for preoperative evaluation of tumor extension was studied in primary breast cancerpatients. METHODS: 3D-MDCT tumor images of 143 tumors in 143 patients with primary breast cancer were created with the volume rendering method. The transverse tumor size (TS) and vertical tumor size (VS) were then measured in an anterior-posterior view of the 3D-MDCT images. The pathological tumor size was determined according to a map of the tumor spread prepared by pathologists using multi-sliced (3-5 mm intervals) surgical specimens and compared with the tumor size on 3D-MDCT images. RESULTS: First, the optimal method for creating 3D-MDCT tumor images was determined for the first 40 patients (learning set), resulting in a fairly good correlation of tumor size on 3D-MDCT images with pathological tumor size (r = 0.983 for TS and r = 0.958 for VS). We then carried out a validation study on the next 103 patients (validation set). The 3D-MDCT tumor size's strong correlation with the pathological tumor size demonstrated a high rate of accuracy (r = 0.974 for TS and r = 0.977 for VS). Subset analyses according to histological type showed that correlation coefficients were r = 0.979 for TS and r = 0.981 for VS of invasive ductal carcinomas (n = 88), r = 0.948 for TS and r = 0.970 for VS of ductal carcinomas in situ (n = 10), and r = 0.984 for TS and r = 0.976 for VS of invasive lobular carcinomas (n = 5). CONCLUSION: 3D-MDCT images can assess breast cancer tumor extension highly accurately, and thus seems to be useful for planning the extent of resection in breast conserving surgery.