Clinical Implementation and Evaluation of an Artificial Intelligence-Driven One-Click Automatic Planning System for Functional Lung Avoidance Radiotherapy
Synopsis
This study implemented AP-FLART, which integrates dosimetric score-based beam angle selection, multi-modality-guided dose prediction, and function-guided dose mimicking, within RayStation and evaluated it on a test dataset of 33 lung cancer patients who underwent SPECT ventilation or perfusion imaging and lung radiotherapy, finding that automatic FLART plans significantly reduced high-function lung mean dose by 15.1% versus manual conventional radiotherapy plans, lowered the probability of grade >=2 radiation pneumonitis by 6.25 percentage points (27%) among FLART-benefiting patients, achieved clinical benefits similar to manual FLART plans, were clinically acceptable without modification in 87.9% of cases, and cut planning time from 2-3 hours to approximately 8 minutes.
Interpretation
AP-FLART was enhanced and implemented within the clinical treatment planning system RayStation, forming a one-click automatic planning workflow for functional lung avoidance radiotherapy. FLART planning previously depended on manual effort; this work integrates dosimetric score-based beam angle selection, multi-modality-guided dose prediction, and function-guided dose mimicking into a clinical planning system and puts it into practice. Based on a system implementation description and a test dataset of 33 lung cancer patients, representing engineering implementation plus evaluation on a clinical dataset.
Compared with manual conventional radiotherapy plans, automatic FLART plans significantly reduced high-function lung mean dose by 15.1% and, among FLART-benefiting patients, reduced the probability of grade >=2 radiation pneumonitis by 6.25 percentage points (27%) while maintaining comparable probabilities for other side effects. Extends the dosimetric advantage of functional lung avoidance from manual FLART to automatically generated plans and provides quantitative estimates at the normal tissue complication probability level. 33 patients who underwent SPECT ventilation or perfusion imaging and lung radiotherapy, with dosimetric metrics and normal tissue complication probabilities compared against manual ConvRT and manual FLART plans created by an experienced planner.
Blinded review by three clinicians indicated that 87.9% of automatic FLART plans were clinically acceptable without modification, and 68.7% were rated as comparable (38.4%) or superior (30.3%) to manual FLART plans. Provides blinded-review evidence on clinical acceptability of automatic plans, beyond dosimetric metric comparisons alone. Blinded review by three clinicians, with automatic and manual FLART plans for the 33-patient dataset.
AP-FLART reduced FLART planning time from 2-3 hours for manual planning to approximately 8 minutes. Substantially compresses planning time while maintaining estimated clinical benefits similar to manual FLART, pointing toward reduced workload and broader clinical adoption of FLART. Based on a planning-time comparison, representing a workflow efficiency metric.
Perspective
The results apply to lung cancer patients who underwent SPECT ventilation or perfusion imaging and lung radiotherapy, with the system implemented in the RayStation clinical treatment planning system; they indicate that one-click automatic FLART planning can achieve dosimetry, complication probability estimates, blinded clinical acceptability, and planning efficiency comparable to manual FLART, offering an implementable path for radiotherapy departments seeking to reduce FLART planning workload.
This is a summary-level reading without figures or supplementary materials, so the specific definitions of dosimetric metrics, the normal tissue complication probability model, the blinded review scale, and the criteria for identifying FLART-benefiting patients cannot be verified; these details affect understanding of the applicable scope and remain open questions to confirm by reading the full text.