Click here to


Are you sure ?

Yes, do it No, cancel

Are Sub-Region Radiomic Features From Pre-Treatment FDG-PET-CT Biomarkers of Recurrence for Cervical Cancer?

I Vergalasova*, K Nie, Y Li, T Cui, B Liu, M Sayan, M Reyhan, N Yue, L Hathout, Rutgers Cancer Institute of New Jersey, New Brunswick, NJ


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: To assess the potential prognostic value of sub-region radiomic features of pre-treatment FDG-PET-CT for locally advanced cervical carcinoma patients treated with definitive concurrent chemoradiation (CCRT).

Methods: Gross tumor volumes (GTVs) were delineated on pre-treatment FDG-PET-CT by a single physician for patients undergoing CCRT. Additional sub-regions inside the GTV were auto-segmented based on various PET metabolic uptake thresholds: GTV_Edge, GTV30%, GTV50%, and GTV70%. GTV_Edge was generated with a signal-to-background ratio algorithm, whereas the latter three volumes were generated as thresholds of the maximum standardized uptake value (SUVmax) within a region-of-interest. Radiomics features of the PET images were then extracted from the whole GTV as well as the listed sub-regions clustered inside the tumor region. Association of radiomic features with recurrence was evaluated using independent sample t-tests with significance level p<0.05.

Results: A total of 27 patients, of whom 11 experienced recurrence, were included in this retrospective IRB-approved study. Median follow-up time was 19.9 months. A total of 309 features were extracted for analysis per region. No significant differences were found between patients that recurred compared to those who did not, for all extracted features from the physician contours. However, 5 features (Histogram of Gradient Orientations: 5 percentile area, 10 percentile area, and 0.025 quantile; Intensity Direction: local range maximum; Neighbor Intensity Difference: texture strength) from GTV_Edge demonstrated significant differences between groups. The only other sub-region that demonstrated significance was GTV30% with 3 features (Histogram of Gradient Orientations: 10 percentile and 0.025 quantile; as well as Intensity Direction: local entropy minimum).

Conclusion: The proposed sub-region radiomics analysis has the potential to serve as a novel biomarker in predicting recurrence for locally advanced cervical cancer and thus further investigation with a larger sample is warranted.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email