Background: To evaluate the accuracy of the combined maximum and minimum intensity projection-based internal target volume (ITV) delineation in 4-dimensional (4D) CT scans for liver malignancies.
Methods: 4D CT with synchronized IV contrast data were acquired from 15 liver cancer patients (4 hepatocellular carcinomas; 11 hepatic metastases). We used five approaches to determine ITVs: (1). ITVAllPhases: contouring gross tumor volume (GTV) on each of 10 respiratory phases of 4D CT data set and combining these GTVs; (2). ITV2Phase: contouring GTV on CT of the peak inhale phase (0% phase) and the peak exhale phase (50%) and then combining the two; (3). ITVMIP: contouring GTV on MIP with modifications based on physician's visual verification of contours in each respiratory phase; (4). ITVMinIP: contouring GTV on MinIP with modification by physician; (5). ITV2M: combining ITVMIP and ITVMinIP. ITVAllPhases was taken as the reference ITV, and the metrics used for comparison were: matching index (MI), under- and over-estimated volume (Vunder and Vover).
Results: 4D CT images were successfully acquired from 15 patients and tumor margins were clearly discernable in all patients. There were 9 cases of low density and 6, mixed on CT images. After comparisons of metrics, the tool of ITV2M was the most appropriate to contour ITV for liver malignancies with the highest MI of 0.93 ± 0.04 and the lowest proportion of Vunder (0.07 ± 0.04). Moreover, tumor volume, target motion three-dimensionally and ratio of tumor vertical diameter over tumor motion magnitude in cranio-caudal direction did not significantly influence the values of MI and proportion of Vunder.
Conclusion: The tool of ITV2M is recommended as a reliable method for generating ITVs from 4D CT data sets in liver cancer.