3207 - Plan-of-the-Day Adaptive Bladder Radiotherapy: A Statistical Shape Model Approach for Optimized Target Coverage and Normal Tissue Sparing

Purpose/Objective(s):

To design and validate an automated, patient-specific statistical shape model (SSM) for Plan-of-the-Day (PoD) adaptive radiotherapy in muscle-invasive bladder cancer. This approach characterizes non-linear, anisotropic bladder deformation to generate a PTV library that optimizes target coverage while significantly reducing irradiated normal tissue volume compared to rigid geometric margins.

Materials/Methods:

Nine patients (264 daily CBCTs) were analyzed. An SSM was trained for each patient using the planning CT and the first five daily CBCTs. Bladder surface variability was quantified using Signed Distance Functions (SDF) to calculate the Mean (M) and Standard Deviation (SD) of boundary positions at every voxel. This captures local anatomical stability, learning that the inferior/posterior wall of the bladder is relatively stable compared to the filling-dependent motion of the anterior/superior wall. PTVs were generated using the formula: M - (Z * SD) - m <= 0, where a confidence coefficient (Z) and safety margin (m) were trained to create three PTVs (PTV_Small, PTV_Medium, PTV_Large).

The primary metric was the "pass rate," defined as the percentage of treatment days achieving > 98% daily bladder volume enclosed by the selected PTV. This model provides topological adaptation, accounting for bladder expansion into concavities in the superior direction. Benchmarks included an ITV+5mm (union of the simulation CT bladder V_sim, and the first five daily bladders) and a Conventional PTV (1.5cm expansion superior/anterior, 1.0cm elsewhere). Validation used fractions from day 6 onwards (Fx 6+).

Results:

Optimization yielded parameter sets [Z, m] of [1.2, 1mm] for PTV_Small, [1.6, 2mm] for PTV_Medium, and [2.8, 2mm] for PTV_Large. In validation, the PTV_Large achieved a 95.8% pass rate (average volume 3.03 ± 1.17x V_sim), outperforming Conventional PTV (84.5%, 2.94 ± 0.47x V_sim) and ITV+5mm (89.0%, 2.53 ± 0.81x V_sim). The PTV_Small achieved substantial sparing (1.73 ± 0.60x V_sim) with a 67.8% pass rate. When excluding outliers where extreme filling would clinically require re-voiding (daily volume > 2.0x V_sim), the PTV_Small pass rate rose to 74.4%. Overall, this strategy allowed 74% of compliant fractions to be treated with the PTV_Small, reducing irradiated volume by 32% versus the ITV approach and 41% versus conventional margins.

Conclusion:

This SSM-based PoD strategy offers rapid offline adaptive re-planning. By shifting computation offline, the daily "Image -> Select -> Treat" workflow ensures temporal robustness against intra-fraction bladder filling. The resulting library provides a reliable safety net while enabling significant normal tissue sparing on the majority of treatment days, effectively addressing the anisotropic and topological challenges of daily bladder radiotherapy.