Development of a Fully Automated 4D-CT Registration Method on Lung Studies

PURPOSE Current 4D-CT protocols use data from an external marker to register the acquired images. Breathing cycle variations affect registration, leading to banding artifacts (BAs) within the 3-D volume. We have developed a fully automatic 4D-CT registration method that is less affected by breath variation than the external marker method. METHOD AND MATERIALS 5 patients with lung cancer were scanned with a GE 4-Slice CT scanner using an overlapping ciné protocol: following a 4x2.5 mm ciné scan, the couch was translated 7.5 mm and another ciné scan was performed, with a common 2.5 mm overlapping slice linking the two adjacent ciné scans. A 4D-CT dataset was produced by aligning ciné image blocks into full 3D volumes for each breathing cycle phase: a starting couch position was selected, the ciné data at that position was truncated to 1 breathing cycle and then interpolated to 16 evenly spaced image block phases. Image block phase 1 was matched to an image block from the adjacent couch position by maximizing the Normalized Cross Correlation of the overlapping slice. The process continued to the next couch position using the matched image blocks until an entire 3D volume was selected. The algorithm was repeated for each of the 16 phases to complete a 4D-CT dataset. The data were also registered using a standard external marker registration method as referenc for comparison. The mean difference between overlapping slices from adjacent image blocks reflects the severity of BAs. The values obtained from the reference and our automated registration method were compared using t-statistics. RESULTS Visual inspection showed a decrease of banding artifacts with our automated method as compared to the reference method. Artifact reduction was accompanied by a significant decrease in mean image difference for 3 of the 5 patients (p \u3c 0.05). Two patients that showed no BAs also showed no significant difference in mean image difference. CONCLUSION An automatic 4D-CT registration method for lung studies has been developed and shown to perform as well or better than a reference external marker method. CLINICAL RELEVANCE/APPLICATION Artifact free 4D-CT scans are generated automatically for radiation therapy planning