Development of An Integrative Digital and Physical Phantom Bladder for Optimized Prostate Radiation Treatment Planning
S Badih1*, T Johnson1, Y Khokhar1, E Salter1, J Chen1, A Deb1, D Ruan2, (1) Department of Bioengineering, University of California, Los Angeles, CA,(2) UCLA School of Medicine, Los Angeles, CA
Presentations
PO-GePV-M-70 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: It is important to understand the dosimetric impact of bladder filling status to optimize the balance between organ sparing and patient comfort for prostate cancer radiotherapy treatment. This study focuses on developing physical and numerical models to precisely characterize the relation between filling volume and bladder morphology.
Methods: We have engineered an integrated digital and physical imaging phantom that can be analyzed in vitro to identify radiation absorption patterns. A digital model was constructed by employing a deep learning algorithm for semantic segmentation of the bladder from pelvic CT scans. Statistical shape analysis, specifically PCA, was performed on bladder segmentations to explain the intrinsic morphology variations. In parallel, a distensible PDMS-based physical phantom bladder with reasonable biomechanical properties was fabricated using spin coating and sandwich molding and was simulated in a human torso context. 3D bladder meshes extracted from CT scans through semi-automatic and manual contouring tools were used as a blueprint for mold fabrication.
Results: Semantic segmentation yielded accurate predictions as verified by the network’s loss function that combines cross-entropy and a modified Sørensen–Dice coefficient, along with qualitative overlay maps. 3D bladder shape was analyzed and quantified in a PCA model utilizing 40 modes of variation to capture 99% of the variations determined by the R2X value. An optimized spin coating/sandwich molding fabrication process was established to achieve human mimicking distensibility and biomechanical properties.
Conclusion: This work provides a methodology for constructing a physical-numerical imaging phantom bladder that accurately recapitulates the material, structural and morphological properties of the bladder. This development supports the corroboration of calculated dose distribution based on digital models against physical measurement using thermoluminescent dosimeters with the physical model. The integrative nature of our approach provides an optimal pathway for the realistic assessment of radiation risks that can be translated directly to the clinic.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email