Click here to


Are you sure ?

Yes, do it No, cancel

An Investigation Into the Correlations Between Quantified Target Change in CBCT Images and Clinical Outcomes for Lung SBRT

J Allen*, R McGarry , W Luo , University of Kentucky, Lexington, KY


(Tuesday, 7/16/2019) 4:30 PM - 5:30 PM

Room: Exhibit Hall | Forum 3

Purpose: Response Evaluation Criteria in Solid Tumors (RECIST) does not currently include evaluation of radiographic signs described by contrast-to-noise (CNR) or density (HU) change. Aim of study is to determine if CNR and HU acquired from daily CBCT exams can be used for tumor evaluation for lung SBRT patients and to correlate the results with patient outcomes.

Methods: Pre-treatment CBCT images for lung SBRT patients were exported to the Varian Eclipse treatment planning system and analyzed using ImageJ. The target CNR was determined using a slice through the central portion of the tumor. Mean HU was determined using Eclipse and a spherical region of interest (ROI) within the tumor. The change in CNR and HU was compared with tumor response.

Results: Ten lung SBRT patients treated on TrueBeam were evaluated with two different fraction schemes, 5000cGy and 5500cGy in five fractions. CNR measurements showed an average decrease of 0.36 with a standard deviation of 0.47. Maximum decrease in CNR was 1.00 and the minimum was -0.13, representing an increase in CNR. The average decrease in mean HU between the first and fifth fractions was 110 with a standard deviation of 96. Maximum decrease in HU was 110 and the minimum was -33, representing an increase in the HU value. Patient follow-ups showed that the decrease/increase in tumor size corresponded to the decrease/increase in CNR. Similar pattern was observed for HU change.

Conclusion: The results have exhibited the target change described by CNR and HU between the first and last fraction was correlated with tumor response. This indicates that the use of radiographic signs described by CNR and HU may be a viable means of tumor evaluation. Future plans for this study involve more extensive data collection, and statistical analysis to reveal the correlation of results with patient treatment outcomes.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email