Automated Fat-tissue Extraction for Preoperative Estimation of Profunda Artery Perforator Flap Weight in Breast...

This study aims to establish an objective and reproducible method for estimating profunda artery perforator (PAP) flap weight by incorporating automated fat-tissue extraction using computed tomography (CT) attenuation thresholds. Thirty-two women who underwent breast reconstruction using PAP flaps were retrospectively analyzed. Preoperative CT data were processed using the Synapse Vincent system to generate 3-dimensional models. Adipose tissue in the flap region was automatically segmented using predefined Hounsfield unit thresholds (-150 to -50 HU). Predicted flap weight was calculated by assuming a fat density of 0.9 g/cm³ and compared with the intraoperative weight using Pearson correlation and linear regression analyses. The mean predicted and actual flap weights were 294.9 ± 73.1 and 272.4 ± 69.8 g, respectively, and demonstrated a strong correlation (r = 0.9149, P < 0.0001). The flap weight estimation formula obtained by linear regression was as follows: actual weight = 0.8628 × predicted weight + 30.39. The coefficient of determination (R² = 0.837) and root-mean-square percentage error (10.33%) indicated high predictive accuracy. Bland-Altman analysis showed a bias of -6.99 g (limits, -62.34 to 48.36 g). Automated fat-tissue extraction using 3-dimensional CT modeling is a rapid and noninvasive technique for reliable estimation of PAP flap weight and may improve preoperative planning for perforator flap breast reconstruction.