Click here to

Session: Imaging for Treatment Assessment and Outcome Modeling [Return to Session]

Quantifying the Impact of Deformable Dose Accumulation Using Voxel-Level Dose-Response Analysis for Non-Small Cell Lung Cancer

Y He*, G Cazoulat, C Wu, L Almodovar-Abreu, E Mccollum, B Rigaud, P Balter, J Pollard-Larkin, D Rhee, L Court, Z Liao, R Mohan, K Brock, The University of Texas MD Anderson Cancer Center, Houston, TX


SU-F-206-8 (Sunday, 7/10/2022) 2:00 PM - 3:00 PM [Eastern Time (GMT-4)]

Room 206

Purpose: To quantify the impact of dose accumulation in correlating radiation damage represented by image density change for non-small cell lung cancer (NSCLC)

Methods: Planning data, weekly 4DCTs, and a 4-month follow-up PETCT were retrospectively acquired from 45 NSCLC IMRT patients. The planned dose (4DPlan) was obtained by deforming and summing the phase doses on the exhale phase (exCT). The accumulated dose (4DAccum) was computed by deforming and summing weekly 4D doses to the planning exCT. Normal ipsilateral lung was tri-partitioned equally along the superior-inferior direction. We performed Pearson correlation between lung volume change over the treatment (∆vol) and mean lung dose (MLD) difference between 4DAccum and 4DPlan. To quantify normal tissue response, we deformed PETCT to planning exCT to compute image density change (IDC). We performed Pearson correlations between IDC and MLD for 4DAccum and 4DPlan, and compared the voxel-level 4DAccum-IDC and 4DPlan-IDC relationships by adapting the Lyman normal tissue complication probability (NTCP) model using mean dose and IDC of voxels within dose intervals.

Results: Compared to 4DPlan, 4DAccum increased overall MLD by 4.1±6.5% (p<0.001). For the inferior lung, 4DAccum increased MLD compared to 4DPlan by 22.8±50.9% (p<0.01) (negatively correlated with ∆vol (r=-0.56, p<0.001)), and patients with negative vs. positive ∆vol experienced average IDC of 95.8HU vs. 32.6HU (p=0.05). For NTCP modelling, global minimum and maximum IDC were 3.7HU and 130.9HU, respectively, and the fit resulted in TD₅₀=42.8Gy, m=0.39 for 4DPlan (R²=0.99) and TD₅₀=45.0Gy, m=0.36 for 4DAccum (R²=1.00). The 2.2Gy difference in TD₅₀ translates to an estimated 5% reduction in pneumonitis risk using the 31.4Gy average MLD.

Conclusion: Differences between planned and accumulated dose resulted in statistically significant impact on voxel-based toxicity assessments. IDC demonstrated a strong dose dependence and the potential to provide early and quantitative indications of lung damage. Further work is required to determine clinical impact.

Funding Support, Disclosures, and Conflict of Interest: This work was supported by: NIH #1P01CA261669, Helen Black Image Guided Fund, a Co-Development and Collaboration Agreement between RaySearch Laboratories and the UT MD Anderson Cancer Center (MDACC), CPRIT-CURE Summer Undergraduate Research Program, Tumor Measurement Imitative at MDACC, and the Image Guided Cancer Therapy Research Program at MDACC.


Image-guided Therapy, Dose Response, CT


IM/TH- Image Registration: CT

Contact Email