Adaptive 4D volume perfusion CT of lung cancer: effects of computerized motion correction and the range of volume coverage on measurement reproducibility.

OBJECTIVE

The purpose of this study was to determine whether measurement reproducibility can be improved using computerized motion correction and whole-tumor coverage in adaptive 4D perfusion CT of lung cancer.

SUBJECTS AND METHODS

Perfusion CT covering the entire z-axis of a mass was performed of 40 patients with lung cancer. Each perfusion CT study was performed in 93.5 seconds and included 17 repeated dynamic CT scans obtained using the Adaptive 4D Spiral mode. Tumor blood flow (BF), blood volume (BV), and permeability were measured in four different manners: in the entire tumor (whole-tumor coverage) without the use of motion correction; in the entire tumor with motion correction; in a small volume of interest (VOI) of tumor without motion correction; and in a small VOI with motion correction. Intra- and interobserver reproducibility were assessed through Bland-Altman analyses.

RESULTS

The 95% limits of intraobserver reproducibility for BF, BV, and permeability were as follows: -52.1% to 48.0%, -22.4% to 27.8%, and -33.2% to 38.5%, respectively, in the whole tumor without motion correction; -53.3% to 45.6%, -17.7% to 20.6%, and -31.5% to 37.0% in the whole tumor with motion correction; -107.8% to 97.4%, -98.3% to 93.7%, and -132.3% to 100.7% in a small VOI of tumor without motion correction; and -74.9% to 98.6%, -74.5% to 88.1%, and -109.8% to 114.1% in a small VOI with motion correction. The 95% limits of interobserver reproducibility for BF, BV, and permeability were as follows: -57.0% to 62.5%, -36.8% to 52.6%, and -47.7% to 66.0%, respectively, in the whole tumor without motion correction; -55.7% to 55.8%, -25.8% to 42.0%, and -35.3% to 46.7% in the whole tumor with motion correction; -146.6% to 165.1%, -117.1% to 137.7%, and -143.2% to 149.8% in a small VOI of tumor without motion correction; and -106.2% to 133.6%, -99.5% to 122.4%, and -108.6% to 170.0% in a small VOI of tumor with motion correction. Overall, the best reproducibility was obtained when measurements were obtained in the entire tumor (i.e., whole-tumor coverage) and when motion correction was used.

CONCLUSION

Measurement reproducibility of perfusion parameters improved when measurements in the entire tumor (i.e., whole-tumor coverage) were obtained and computerized motion correction was used. The best reproducibility in parameter values was obtained with motion correction and whole-tumor coverage.