Purpose: To report our initial clinical experience with automated IMRT planning for prostate radiotherapy plans using ECHO. ECHO is a home-grown independent optimization engine based on hierarchical constrained optimization, integrated with Eclipse® system through application program interface. ECHO does not use Eclipse optimization engine and the optimal fluence map generated by ECHO is imported into Eclipse for leaf sequencing and final dose calculation.
Methods: ECHO was released to clinical use at our institution starting from Dec.2019 for prostate post-brachy SBRT (25Gy in 5fractions) and moderate hypofractionation prostate radiotherapy (MODHYPO) (70.2Gy in 26fractions) without nodal involvement. Prior to clinical release, training was performed on randomly selected 10 SBRT and 8 MODHYPO cases to fine-tune the ECHO parameters, and validated retrospectively on 50 cases (21 SBRT, 29 MODHYPO). We performed a comparison study with the manually generated clinical plans for all dose metrics used for prostate SBRT and MODHYPO patients in our clinic. Over a 14 months period of ECHO clinical implementation, ECHO has generated 152 clinical plans (130 SBRT and 22 MODHYPO) using 9 coplanar beams IMRT.
Results: In pre-clinical validation, ECHO plans met all the clinical criteria and had high quality with less variability compared with manually generated clinical plans. After clinical implementation, the average time to produce an ECHO plan was 28 minutes (range 16-39). All ECHO plans were produced by one single run. They all met or were better than the institutional clinical criteria. All ECHO plans were delivered after passing intuitional quality assurance process.
Conclusion: ECHO required only few patients for training and produced high-quality plan within short time consistently for post-brachytherapy and moderate hypofractionation prostate radiotherapy. It streamlined the clinical planning workflow, improved the plan quality on average and consistency, and shortened planning time by one day from simulation to treatment in our clinic.
Funding Support, Disclosures, and Conflict of Interest: This work was partially supported by the MSK Cancer Center Support Grant/Core Grant (NIH P30 CA008748)