Clinical Validation of AI-Assisted Radiotherapy Contouring Software for Thoracic Organs at Risk

Prospective, Multicenter, Randomized Evaluation of the Performance and Clinical Applicability of AI-Assisted Radiotherapy Contouring Software for Thoracic Organs at Risk

The goal of this clinical trial is to evaluate performance and clinical applicability of AI-assisted radiotherapy contouring software (iCurveE) for thoracic organs at risk. The main question it aims to answer is:

• Does AI-assisted contouring (AI contouring with manual modification) offer greater accuracy and time efficiency compared to manual contouring? After screening, the qualified participants' thoracic CT images will be anonymized and segmented using three methods: manual, AI (AI-only), and AI-assisted contouring. The researchers will compare the results generated by the three different contouring methods with the ground truth established by expert consensus, in order to evaluate both accuracy and time-related parameters

Study Overview

• Status: Completed
• Conditions: Breast Cancer; Esophageal Cancer; Lung Cancer
• Intervention / Treatment: Device: contouring of thoracic organs at risk

• Study Type: Observational
• Enrollment (Actual): 500

Contacts and Locations

Study Locations

• China
  • Tianjin Municipality
    • Tianjin, Tianjin Municipality, China, 300060
      • Tianjin Medical University Cancer Institute and Hospital, Tianjin Key Laboratory of Cancer Prevention and Therapy

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 18 years and older (Adult, Older Adult)
• Accepts Healthy Volunteers: No

• Sampling Method: Probability Sample

Study Population

This trial will enroll 500 patients with lung, esophageal, or breast cancer, who are scheduled to receive thoracic radiotherapy across five clinical cancer institutes.

Description

Inclusion Criteria:

1. ≥18 years old, no gender limit.
2. Patients diagnosed with breast cancer, lung cancer, or esophageal cancer, who are scheduled for chest CT scanning followed by thoracic radiotherapy.
3. CT slice thickness ≤5mm.
4. Patients understand the goal of the trial, are willing to attend the trial and sign the informed consent.

Exclusion Criteria:

1. Congenital malformations or abnormal anatomical structures resulting from non-tumor factors in the scan area.
2. Artifact, prosthesis or implantation causing images undistinguishable.
3. CT images not conforming to DICOM standards.
4. Investigators consider not suitable.

Study Plan

How is the study designed?

Number of groups / cohorts

3

Cohorts and Interventions

Group / Cohort
Independent manual contouring; Manual contouring refers to physicians using the brush tool on the contouring platform to segment thoracic organs at risk manually, without the use of auto-segmentation tools.
AI-assisted contouring; After generating the AI contouring results, investigators will import them into the contouring platform and perform manual modifications, producing the AI-assisted contouring.
AI contouring; AI contouring refers to the auto-segmentation results generated by the Res-SE Net model, with the model integrated into the auto-segmentation software (iCurveE).

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Contouring time (min)	Manual contouring time is recorded from the time the CT is loaded on the contouring platform to the completion of contouring. AI-assisted contouring time is defined as the sum of the auto-segmentation model runtime, the transfer to the contouring platform, and the subsequent manual modification.	Within 6 months after enrollment
volumetric DICE similarity coefficient, vDSC	vDSC= 2×（A∩B）/(A+B), where A refers to the volume of the ground truth, and B refers to the volume of the manual, AI, or AI-assisted contour.	Within 6 months after enrollment

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Rate of time efficiency improvement	Rate of efficiency time improvement= (manual contouring duration - AI-assisted contouring duration)/ manual contouring duration*100%	Within 6 months after enrollment
95th percentile Hausdorff Distance, HD95	HD95(A, B) = max (h95(A, B), h95(B, A)), where h95(A, B) is the 95th percentile of the shortest distances from all points on surface A to surface B, and vice-versa for h95(B, A). A represents the ground truth and B represents the manual, AI or AI-assisted delineation	Within 6 months after enrollment
Surface DICE similarity coefficient, sDSC	sDSC = (|S(A) ∩ S(B)τ| + |S(B) ∩ S(A)τ|) / (|S(A)| + |S(B)|), where S(A) and S(B) are the sets of points on the surfaces of A and B, S(B)τ represents the points on surface B that are within the tolerance τ of surface A, and S(A)τ represents the points on surface A that are within the tolerance τ of surface B. A represents the ground truth and B represents the manual, AI or AI-assisted delineation	Within 6 months after enrollment
Volumetric revision index, VRI	VRI = [(A- A∩B) + (B- A∩B)] /A, where A refers to the volume of the ground truth, and B refers to the volume of the manual, AI, or AI-assisted contour.	Within 6 months after enrollment
Recall, Rec	Rec = | A∩B| / A, where A refers to the volume of the ground truth, and B refers to the volume of the manual, AI, or AI-assisted contour.	Within 6 months after enrollment
Precision, Pre	Pre= |A∩B| / B, where A refers to the volume of the ground truth, and B refers to the volume of manual, AI, or AI-assisted contour.	Within 6 months after enrollment
Relative volume difference, RVD	RVD = |A-B| /A, where A refers to the volume of the ground truth, and B refers to the volume of the manual, AI, or AI-assisted contour.	Within 6 months after enrollment
Investigators satisfaction score for AI contouring	Evaluated on a 1-5 Likert scale: 1 - strongly dissatisfied, 2 - dissatisfied, 3 - neutral, 4 - satisfied, 5 - strongly satisfied.	Within 6 months after enrollment

Other Outcome Measures

Outcome Measure	Measure Description	Time Frame
Number of adverse events, AEs	Participant Adverse events during CT scanning	Within 1 day after CT scanning
Number of device defects during AI-assisted contouring	Number of failures in generating, transferring, or saving auto-segmentation results	Within 6 months after enrollment

Collaborators and Investigators

• Sponsor: Tianjin Medical University Cancer Institute and Hospital
• Collaborators: Shanxi Province Cancer Hospital; Fifth Affiliated Hospital, Sun Yat-Sen University; Guangzhou Perception Vision Medical Technology Co. Ltd; People's Hospital of Guangxi Zhuang Autonomous Region; Union Hospital, Tongji Medical College, Huazhong University of Science and Technology
• Investigators: Principal Investigator: Zhiyong Yuan, Ph.D., Tianjin Medical University Cancer Institute and Hospital

Study record dates

Study Major Dates

• Study Start (Actual): September 30, 2022
• Primary Completion (Actual): July 27, 2023
• Study Completion (Actual): March 6, 2024

Study Registration Dates

• First Submitted: March 2, 2023
• First Submitted That Met QC Criteria: March 27, 2023
• First Posted (Actual): March 28, 2023

Study Record Updates

• Last Update Posted (Actual): February 12, 2026
• Last Update Submitted That Met QC Criteria: February 9, 2026
• Last Verified: February 1, 2026

More Information

Terms related to this study

• Keywords: Radiotherapy; Artificial Intelligence; contouring; thoracic organs at risk
• Additional Relevant MeSH Terms: Neoplasms by Site; Neoplasms; Respiratory Tract Diseases; Gastrointestinal Neoplasms; Digestive System Neoplasms; Digestive System Diseases; Gastrointestinal Diseases; Lung Diseases; Head and Neck Neoplasms; Respiratory Tract Neoplasms; Thoracic Neoplasms; Esophageal Diseases; Skin Diseases; Breast Diseases; Skin and Connective Tissue Diseases; Lung Neoplasms; Esophageal Neoplasms; Breast Neoplasms
• Other Study ID Numbers: PVMEDCT202201

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: YES
• IPD Plan Description: The protocol of this study are available from the corresponding author upon reasonable request after the manuscript publication.
• IPD Sharing Time Frame: Beginning 1 year after publication with no end date.
• IPD Sharing Access Criteria: Requests must include a detailed protocol, analysis plan, and data exchange with institutional approvals in place before data transfer of any information.
• IPD Sharing Supporting Information Type: STUDY_PROTOCOL

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No
• product manufactured in and exported from the U.S.: No