Purpose: To develop a novel image biomarker quantitatively describing tumor biomechanical properties based on ventilation-induced tumor deformation (VITD) and evaluate its prognostic values.
Methods: Under IRB approval, 10-bin 4D-CT and PET scans for 7 patients with lung squamous cell carcinoma (SCC) were retrospectively solicited. Tumor surface models were created using binary segmentation for 10 respiratory phases and then regionally aligned using surface-based constrained harmonic registration. The resultant shape descriptors (detJ) across respiration phases were assembled as an intra-subject ventilation-induced tumor deformation model (VITD). Subsequently, tumor strain was approximated as the magnitude of the averaged shape changes (VITD_mag) between end-of-exhale (EOE) and end-of-inhale (EOI) phases in the relevant regions - areas showing significant vertex-wise correlations between VITD-detJ and the respective lung volume trajectory. The topographical distribution of ventilation load in the tumor-bearing lung was mapped by deforming EOE 4D-CT to EOI. Para-tumor stress (PTS) was then derived as an average of the registration-resultant displacements in the dilated para-tumor region. Finally, VITD_rigi was calculated as PTS over VITD_mag. Pearson’s correlation was determined between VITD derived metrics and PET-derived SUV_max.
Results: On 7 lung SCC tumors, VITD_mag, the extent of maximal shape changes in respiration, was negatively associated with SUV_max, indictive of elevated tumor metabolism (r= -0.88, p= 0.01). After controlling for PTS, an increase in the derived tumor rigidity (VITD_rigi) was found to be significantly associated with increased SUV_max, with r= 0.76 and p= 0.04.
Conclusion: We developed a novel noninvasive pipeline to characterize lung tumor rigidity based on VITD in 4D-CT. The derived VITD_rigi significantly correlated with tumor FDG uptake, indicating a correlation between image biomarker-derived tumor rigidity and metabolism. The observation is consistent with previously reported increasing rigidity in aggressive tumors. The proposed VITD model thus provides new biomechanical evidence in lung cancer prognosis and for future risk stratification.
Quantitative Imaging, FDG PET, CT