Purpose: To assess the accuracy of maximum-intensity volumes (MIV) for fast contouring of lung tumors including respiratory motion.
Methods and materials: Four-dimensional computed tomography (4DCT) data of 10 patients were acquired. Maximum-intensity volumes were constructed by assigning the maximum Hounsfield unit in all CT volumes per geometric voxel to a new, synthetic volume. Gross tumor volumes (GTVs) were contoured on all CT volumes, and their union was constructed. The GTV with all its respiratory motion was contoured on the MIV as well. Union GTVs and GTVs including motion were compared visually. Furthermore, planning target volumes (PTVs) were constructed for the union of GTVs and the GTV on MIV. These PTVs were compared by centroid position, volume, geometric extent, and surface distance.
Results: Visual comparison of GTVs demonstrated failure of the MIV technique for 5 of 10 patients. For adequate GTV(MIV)s, differences between PTVs were <1.0 mm in centroid position, 5% in volume, +/-5 mm in geometric extent, and +/-0.5 +/- 2.0 mm in surface distance. These values represent the uncertainties for successful MIV contouring.
Conclusion: Maximum-intensity volumes are a good first estimate for target volume definition including respiratory motion. However, it seems mandatory to validate each individual MIV by overlaying it on a movie loop displaying the 4DCT data and editing it for possible inadequate coverage of GTVs on additional 4DCT motion states.