A Reproducible Method for Donor Site Computed Tomography Measurements in Abdominally Based Autologous Breast Reconstruction.

Abstract

We present an approach for evaluating abdominal computed tomography (CT) scans that generates reproducible measures relevant to donor site morbidity after abdominally based breast reconstruction. Seventeen preoperative CT metrics were measured in 20 patients with software: interanterior superior iliac spine distance; abdominal wall protrusion; interrectus distance; rectus abdominis width, thickness, and width-to-thickness ratio; abdominal wall thickness; subcutaneous fat volume; visceral fat volume; right/left psoas volumes and densities; and right/left rectus abdominis volumes and densities. Two operators performed measures to determine interrater reliability (n = 10). Interclass coefficients (ICCs) were calculated, and Bland-Altman plots were fashioned. Intrarater reliability was excellent (ICC > 0.9, 0.958-1) for 15 measures, and good (0.75 < ICC < 0.9, 0.815-0.853) for 2 measures. Interrater reliability was excellent (ICC > 0.9, 0.912-0.995) for 12 measures and good (0.75 < ICC < 0.9, 0.78-0.896) for 5 measures. Bland-Altman plots confirmed intra/interrater agreement. Our study meets its objective of establishing a protocol for obtaining abdominal CT measurements with high reproducibility and intrarater and interrater reliability. Although this study is not meant to weigh the particular influences of various CT measurements on clinical outcomes, we are now actively studying this with the intention of reporting our findings in the near future. Larger patient cohorts must be leveraged to determine correlations between abdominal CT scan findings and donor site outcomes using machine learning algorithms that generate models for predicting abdominal donor site complications.