Predicting the Efficacy of Spacer Placement for Pancreatic Cancer Using a Novel Patient-Specific Virtual Spacer
H Hooshangnejad 1*, K Ding2, (1) Johns Hopkins University, School of Medicine, Biomedical Engineering, Baltimore, MD, (2) Johns Hopkins University, School of Medicine, Radiation Oncology and Molecular Radiation Sciences, Baltimore, MD
Presentations
TH-F-TRACK 6-2 (Thursday, 7/29/2021) 4:30 PM - 5:30 PM [Eastern Time (GMT-4)]
Purpose: Stereotactic body radiation therapy (SBRT) is an effective way to achieve local-regional control for pancreatic cancer but is limited by proximal organs’ at risk (OARs). Our previous clinical trial demonstrated that endoscopic ultrasound-guided injection of hydrogel spacer increases separation between target and duodenum. Yet, spacer’s clinical benefits highly depend on patient anatomy. We developed a virtual spacer for pancreatic hydrogel spacer to generate augmented data to design a decision support system (DSS) and predict patient-specific dosimetric benefits.
Methods: A finite element method oriented spacer simulation algorithm (FEMOSSA) is developed and defined using the material properties, boundary conditions, surface interfaces, and interactions based on patient contour. Finite element analysis software is used to solve the differential equation system generated by FEMOSSA. Simulation accuracy was evaluated by pre-post injection CTs from four cases and quantified by dice similarity coefficient (DSC), overlap volume histogram (OVH), and radial nearest neighbor distance (RNND). Four virtual spacer scenarios were simulated for 15 SBRT patients: no injection, injection in pancreas interface with duodenal sections D1-D2, D1-D2-D3, and D1-D2-D3-D4. 60 corresponding SBRT plans (33Gy/5 fractions) are used for designing a Bayesian DSS.
Results: Average target and duodenal DSC increased, 0.86 to 0.89 and 0.44 to 0.63, respectively. Average mean and 5th percentile RNNDs differed by 0.5 and 2.1mm. OVH thresholds show average of less than 0.75mm difference. Results demonstrate a spacer-location-independent decrease in duodenal V20Gy and a highly spacer-location-dependent change in V33Gy. L1cc and L20cc (tumor expansion overlapping 1cc and 20cc duodenal volume) are good predictors of V20Gy and V33Gy. L1cc>14mm results in V33Gy=0. Baysian DSS predicts spacer injection related improvement in BED with root mean squared error of 3.6Gys.
Conclusion: A novel patient-specific virtual spacer is developed and validated for pancreas SBRT, and used to design DSS that indicates hydrogel efficacy, optimum location, and expected improvement.
Funding Support, Disclosures, and Conflict of Interest: Research Supported by National Cancer Institute R37CA229417 and Augmenix
Handouts
Keywords
Simulation, Finite Element Analysis, Treatment Planning
Taxonomy
IM/TH- Image-guided Surgery: Virtual Reality and Simulation
Contact Email