MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Combined Morphologic and Metabolic Pipeline for Treatment Response Evaluation - a Pilot Study On Pancreatic Adenocarcinoma Patients

Y Lao*, J David , A Torosian , W Yang , R Tuli , Cedars-Sinai Medical Center, Los Angeles, CA

Presentations

(Tuesday, 7/31/2018) 11:00 AM - 12:15 PM

Room: Davidson Ballroom B

Purpose: To better assess treatment response on pancreatic adenocarcinoma (PA) patients, we aim to perform regional comparisons of 3D tumor surfaces pre- and post- chemoradiotherapy (CRT), utilizing surface measurements containing both morphological and metabolic features.

Methods: Thirty-one PA patients with pre- and 6-8 week post-CRT 18F FDG-PET/CT scans were evaluated. Gross tumor volumes (GTV) were manually defined on respective CT images. As shown in Fig.1, on each of the GTV, 3D mesh was generated, followed by surface based registration. For each surface vertex, a multivariate vector was formed from two components: anatomic (deformation tensors resulted from surface registration), and metabolic (regional SUV obtained from radius to surface projections). Paired Mahabanobis distance (Mdist) between pre- and post-CRT tumor surfaces with multivariate vectors was calculated for each patient. The classification capacity of Mdist was evaluated using risk stratification analysis from two
methods: M1 sorted each of the metrics, and evenly divided subjects into low-risk group (LG) and high-risk group (HG); M2 used k-means clustering algorithm to automatically separate LG and HG based on multivariate data, consisting of the tested metric and age. Log-rank tests were performed between LG and HG stratified using both M1 and M2, respectively. As a comparison, same analyses were performed on global SUVmax and GTV volume.

Results: As shown in Fig.2, the fused Mdist outperformed traditional morphologic and metabolic measurements in patient risk stratification, either alone (M1, p=0.0003) or combined with age (M2, p=0.0005). As a contrast, log-rank p-values for patients stratified by global SUVmax were 0.0587 for M1 and 0.0064 for M2, while patients stratified by GTV volume fails to reach significant differences.

Conclusion: We developed a PET/CT-based pipeline combining morphologic and metabolic properties of the tumors for CRT response evaluation in PA patients. The multiparametric model outperforms SUVmax and GTV volume to predict overall survival.

Keywords

3D, Statistical Analysis, PET

Taxonomy

IM- PET : Quantitative imaging/analysis

Contact Email