SU-E-J-134: Motion Modeling of Non-Small Cell Lung Nodules Based on Respiratory Mechanics.

PURPOSE: To quantify the movements of non-small cell lung nodules using 4D cone-beam computed tomography (4D-CBCT) that is automatically registered with planning CT, and to develop a mathematical model to predict the motion trajectory. Modeling the tumor motion may reduce the PTV and ultimately increase the therapeutic ratio. METHODS: Absolute coordinates of the lung nodules in 15 patients were quantified for each phase of 4D-CBCT scans using auto-registration methods. Assuming respiration follows an elliptical pattern spatially in the lung, these coordinates were fitted to trigonometric functions in each x-y-z direction. Adjusting for phase dependence, the motion could be compared quantitatively for inter-fractional and intra-patient variations to determine if this model is universally applicable and has predictive value. RESULTS: Examination of over 36 sets of 4D-CBCT data shows acceptable agreement (< 2mm) with the elliptical model for both individual scans and over the course of treatment. Some inter-fractional variations in amplitude and cycling periods indicate the need to remodel as patients' conditions change. The intra-patient variations are significant and strongly dependent on the patient lung volume and tumor location, thus individual modeling of tumor motion is expected. CONCLUSIONS: The model indicates good agreement and clinical relevance with non-small cell lung nodule motion, and it appears to be potentially relevant over the course of treatment. Most re-acquired 4D-CBCT images inter-fractionally were within the baseline spatial resolution of the auto- registration technique. However, if remodeling is necessary inter-fractionally, this model still has the potential for significant motion margin reduction over the course of treatment.

SU-E-J-116: Investigation of Accuracy and Precision of ITV Delineation by PET SUV Threshold.

PURPOSE: To determine PET SUV values in Pinnacle TPS and to assess the accuracy and precision of the ITV according to SUV thresholds. The goal is to minimize errors in target definition by using SUV showing enhanced metabolic activity of the tumor and fast CT imaging with less motion artifacts. METHODS: Mean PET values in individual patients were obtained from statistics of body contours for whole body PET-CT scans. The SUVs were calculated by normalizing the PET values in any voxel by the mean body PET value. These clinically acquired SUVs were plotted against values reported at the time of the scan for verification. GTVs were contoured on petCT scans and the ITV's were contoured using three published methods of SUV thresholds at 2.5, at ratio of 40 percent of the maximum, and 30 percent of the maximum added to 60 percent of the mean body. GTV volumes were plotted against the 3 sets of ITVs to investigate their relationships. RESULTS: Examination of 11 patients showed a strong linear relationship between clinically determined SUVs and those reported; indicating the validity of our SUV definition. Plots of GTV volumes versus ITV volumes for each of the 3 thresholds revealed a less clear relationship. The effectiveness of each method to generate a reasonable ITV was highly patient dependent; in general the 2.5 threshold gave the best results, while the 40 percent maximum produced the worst. CONCLUSIONS: While more data is required to make a definitive statement about the value of PET SUV threshold defined ITVs, the findings do seem to reveal a pattern between GTV and ITV size. If an appropriate SUV threshold is chosen, the GTV-ITV volume relationship is nearly linear, which suggests extending the GTV in volume rather than the margin distance as is common in ITV delineation.