Date Approved

8-13-2026

Graduate Degree Type

Project

Degree Name

Medical Dosimetry (M.S.)

Degree Program

Allied Health Sciences

First Advisor

Kristen Vu

Second Advisor

Robert Hammond

Third Advisor

Ryan Schurr

Fourth Advisor

Zheng Shi

Academic Year

2025/2026

Abstract

Purpose: Automated treatment planning systems (ATPS) are critical for improving plan consistency and efficiency. This study evaluated the implementation and performance of the Pareto optimal projection search algorithm (POPSA), a proprietary rule-based algorithm (RBA) in prostate site. The RBA's performance was benchmarked against established knowledge-based planning (KBP) and deep learning (DL) dose prediction tools.

Methods:  A retrospective comparative study used a cohort of N=10 previously treated patients, consisting of 10 prostate with lymph nodes cases. Three plans were generated per patient: manual reference (RA) plan, KBP plan, and RBA script-only plan. Performance was quantified by target volume coverage, organ at risk (OAR) sparing, planning consistency (reduction in dosimetric variance), and effective planning working time (EPWT).

Results: The POPSA script achieved significant efficiency gains, reducing estimated manual planning time by approximately 70% across the cohort. Dosimetrically, POPSA demonstrated exceptional algorithmic stability, maintaining PTV7000 V100% coverage within a narrow band (94.98% to 95.10%).  It showed improved consistency (reduced dosimetric variance) compared to manual plans within each site. While KBP and RA were statistically superior at minimizing maximum planning target volume dose (P< 0.001), all methods safely met the clinical threshold of less than 110%. POPSA significantly outperformed both KBP and RA in low-to-intermediate rectal sparing (rectumV2500 and rectumV3500), aggressively driving doses to the lower limits of feasibility

Conclusions: The POPSA successfully standardizes prostate planning across diverse anatomies while drastically reducing resource intensity. By consistently meeting strict clinical constraints and exceeding the passing margins of treated plans in key OAR metrics, POPSA provides a reliable automated workflow that allows dosimetrists to transition into critical plan evaluators without compromising the standard of care.

Keywords: POPSA, Automated planning, Rule-Based Algorithm, Eclipse Scripting API, Dose prediction, Knowledge-based planning, Planning Time

Download

Share

COinS