Dosimetric comparison of dose accumulation between rigid registration and deformation registration in intensity-modulated radiation therapy for large volume non-small cell lung cancer

BACKGROUND: To evaluate the cumulative dose to the target volumes and organs at risk (OARs) after replanning during intensity-modulated radiation therapy (IMRT) for large volume non-small cell lung cancer (NSCLC) based on rigid registration and deformation registration technologies. METHODS: Thirty patients with large volume NSCLC who were treated with IMRT were selected, and two four-dimensional computed tomography (4DCT) scans were acquired before radiotherapy and after 20 fractions of radiotherapy. The initial treatment plan (Plan(1)) based on the average density projection CT (CT(1-avg)) of the first 4DCT images and the second treatment plan (Plan(2)) based on CT(2-avg) of the second 4DCT images were calculated. Then, the dose distributions of Plan(2) and Plan(1) were accumulated based on rigid and deformation registration technologies to obtain Plan(rig) and Plan(def), respectively. Finally, the volume changes of the gross tumor volume (GTV) and OARs between the two CT scans, and the dose-volume parameters among Plan(1), Plan(2), Plan(rig) and Plan(def) were compared. RESULTS: Compared with those on the first CT, the mean GTV and heart volume on the second CT decreased by 44.2% and 5.5%, respectively, while the mean volumes of the ipsilateral lung, contralateral lung and total lung increased by 5.2%, 6.2% and 5.8%, respectively. The differences in the above volume parameters between the two CT scans were statistically significant (P<0.05). Compared with those in Plan(1), the D(95), D(98) and V(100%) values of the IGTV (GTV fusion of 10 CT phases) and planning target volume (PTV) in Plan(2) did not change significantly (P>0.05), and those of Plan(rig) and Plan(def) decreased slightly (P<0.05). The dose-volume parameters of the spinal cord, heart, ipsilateral lung and total lung in Plan(2), Plan(rig) and Plan(def) were significantly lower than those in Plan(1) (P<0.05). Among these parameters, V(30) and the mean dose to the heart in Plan(2), Plan(rig) and Plan(def) decreased by 27.3%, 16.5%, and 15.3% and 15.2%, 6.6%, and 5.6% compared to those in Plan(1), respectively; V(20) and the mean dose to the total lung in Plan(2), Plan(rig) and Plan(def) decreased by 15.6%, 4.5%, and 3.7% and 15.7%, 6.2%, and 5.1% compared to those in Plan(1), respectively. Some dose-volume parameters (including D(95) and D(98) to the target volume, V(40) of the heart, V(20) and the mean dose to the ipsilateral lung and the total lung) of Plan(def) were slightly higher than those in Plan(rig) (P<0.05). The Dice similarity coefficients (DSCs) of the OARs after deformation registration were significantly higher than those after rigid registration (P<0.05). CONCLUSIONS: The dose-volume parameters of OARs in Plan(2) were noticeably different from those in Plan(1), so all of these parameters have large deviations in evaluating the actual dose to the OARs. And, the dose-volume parameters obtained by deformation registration can better predict the actual dose than those obtained by rigid registration.