Virtual Library
Start Your Search
A. Tibdewal
Author of
-
+
P2.14 - Radiotherapy (ID 715)
- Event: WCLC 2017
- Type: Poster Session with Presenters Present
- Track: Radiotherapy
- Presentations: 1
- Moderators:
- Coordinates: 10/17/2017, 09:30 - 16:00, Exhibit Hall (Hall B + C)
-
+
P2.14-005 - Determination of Optimal Cut off SUV Threshold for Auto-Contouring of GTV Using PETCT for Early Stage NSCLC (ID 8812)
09:30 - 09:30 | Author(s): A. Tibdewal
- Abstract
Background:
[18]F-Fluorodeoxyglucose positron emission tomography-computed tomography ([18]F-FDG-PET-CT) has the potential to increase the precision in contouring of gross tumour volume (GTV). However, detection of tumour edge in the halo around the tumour has been a problem. A surgical histopathological examination is the current gold standard for tumour size estimation in NSCLC. The aim of this study was to determine an optimal cut-off of standardized uptake value (SUV) on FDG-PET-CT images that correlates best with tumour size on surgical histopathology examination.
Method:
From January 2013- July 2014, 25 consecutive patients with diagnosed early NSCLC (pT1-pT3,N0M0) who underwent surgical resection (either a lobectomy or pneumonectomy) were accrued. GTVs were delineated on a preoperative FDG-PET-CT scan (acquired within 8 weeks before surgery) by automatic delineation using % threshold SUV at 20%, 30%, 40%, 50% of maximal uptake and threshold as absolute SUV 2, 2.5, 3, 3.5 and 4. The maximum tumour size was recorded from the surgical histopathology reports. First order linear regression was used to obtain values of optimal cut off SUV for each patient at which maximum size of GTV on FDG-PET-CT matched with maximum tumor size on histopathology. Different SUV thresholds of GTV delineation were compared with histopathology with respect to the estimation of maximum tumour size using Bland-Altman plots. The above methodology was carried out in 25 patients in test set. 12 additional patients were used to validate the results of the test set.
Result:
On analysis of 25 patients in the test set using first order linear approximation, the mean optimal cut-off values for GTV delineation on FDG-PET-CT images were 35.6 % ± 18.6 for % threshold SUV and a 4.35 ± 1.7 for absolute SUV. On analysis of 12 more patients in the validation set, the mean optimal cut-off values for GTV delineation on FDG-PET-CT images were 36.9 ± 16.9 % threshold SUV and a 4.1 ± 1.6 absolute SUV. After combined analysis of all 37 patients, the mean optimal cut-off values for GTV delineation on FDG-PET-CT images were 36 ± 17.9 % threshold SUV and a 4.27 ± 1.7 absolute SUV. On comparing various methods of delineation by Bland-Altman plots, auto-contouring with percentage threshold of 40% and absolute SUV 4 were in greater agreement with the histopathological tumour size.
Conclusion:
Auto-contouring of GTV in NSCLC with the help of optimal cut-off SUV in FDG-PET-CT will improve precision in delineation, reduce inter-observer variability and importantly will save time.
