KEY POINTS
- The ECHO system uses constrained hierarchical optimization: mandatory organ-at-risk limits are enforced first, target coverage is then maximized, and normal-tissue dose is reduced without materially compromising the achieved coverage. Seven lung radiotherapy fractionation schedules were implemented.
- For each schedule, parameters were tuned using 5–7 previously treated patients and validated in a further 20–25 patients. Automated plans underwent institutional patient-specific quality assurance before clinical release and continued to pass routine clinical QA.
- From May 2021 through 2025, 431 patients received ECHO-planned conventional or moderately hypofractionated IMRT. A further 93 lung SBRT patients were planned with ECHO VMAT from April 2023, producing a total clinical cohort of 524 patients.
- Mean planning target volume was 391 cm³ for IMRT, ranging from 12 to 3,441 cm³, and 17 cm³ for VMAT SBRT, ranging from approximately 3 to 56 cm³. The system supported conventional, hypofractionated, palliative, retreatment, and three- to four-fraction SBRT schedules.
- ECHO IMRT produced more consistent target coverage, a lower maximum target dose, and a higher minimum dose than manual optimization. VMAT target coverage was similar overall but showed less interpatient variability; organ-at-risk differences were generally small and not considered clinically meaningful.
- Clinically acceptable plans were achieved on the first run in 68% of IMRT cases and 85% of VMAT cases. Approximately 70% of reruns followed physician- or planner-initiated changes to contours, beam arrangements, or clinical criteria rather than optimization failure.
- Average background calculation time was 33 minutes for IMRT and 56 minutes for VMAT, but hands-on preparation and submission required only about 10 minutes, versus 105 and 102 minutes for manual IMRT and VMAT planning. Estimated active planner time fell by approximately 1.5 hours per case.
CLINICAL TAKEAWAY
ECHO demonstrates that automated lung planning can be deployed at scale across markedly different anatomies and fractionation schedules while improving consistency and reducing repetitive planner work. Implementation still requires local software integration, protocol-specific commissioning, clinical review, and patient-specific QA.