Artificial Intelligence-supported Reading Versus Standard Double Reading for the Interpretation of Magnetic Resonance Imaging in the Detection of Local Recurrence for Nasopharyngeal Carcinoma: a Randomised Controlled Multicenter Study

The aim of this randomized controlled study is to investigate whether the previously developed artificial intelligence model can triage post-radiotherapy magnetic resonance images of patients with nasopharyngeal carcinoma and assist radiologists in their interpretation.

Study Overview

• Status: Not yet recruiting
• Conditions: Artificial Intelligence Supported Image Reviewing
• Intervention / Treatment: Diagnostic test: AI

• Study Type: Observational
• Enrollment (Estimated): 10400

Contacts and Locations

• Study Contact: Name: Fang-Yun Xie; Phone Number: +8602087342926; Email: xiefy@sysucc.org.cn
• Study Contact Backup: Name: Pu-Yun OuYang; Phone Number: +8602087342926; Email: ouyangpy@sysucc.org.cn

Study Locations

• China
  • Guangdong
    • Guangzhou, Guangdong, China, 510060
      • Sun Yat-sen University Cancer Center
      • Contact: Fang-Yun Xie; Phone Number: +8602087342926; Email: xiefy@sysucc.org.cn
      • Contact: Pu-Yun OuYang; Phone Number: +8602087342926; Email: ouyangpy@sysucc.org.cn
      • Principal Investigator: Fang-Yun Xie

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Probability Sample

Study Population

This study enrolled patients with treatment naive nasopharyngeal carcinoma who have finished radiotherapy for 6 months or more and have no tumor residue in previous examinations.

Description

Inclusion Criteria:

• Patients with treatment naive nasopharyngeal carcinoma who had finished radiotherapy for 6 months or more
• The previous magnetic resonance imaging examination had showed complete remission in the primary site
• Images are acquired using a 3T magnetic resonance imaging device, including unenhanced T1-weighted and T2-weighted sequences and contrast-enhanced T1-weighted sequences

Exclusion Criteria:

• Patients are enrolled in this study for a specific magnetic resonance imaging scan and not for subsequent follow-up magnetic resonance imaging scans.

Study Plan

How is the study designed?

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
AI-supported reading; The AI model predicts the incidence of local recurrence. If the incidence is below 60%, one radiologist will interpret the MR images. If the incidence is above 60%, two radiologists will interpret the MR images. The radiologists will be provided with the predictive incidence and contours in their interpretation if desired. If two radiologists provide contradictory interpretations, a third radiologist will participate in the discussion to reach a consensus.	Diagnostic test: AI; An artificial intelligence model predicts the risk and contours of local recurrence for MR images and triages them before radiologists interpret them.
Standard double reading; The MR images will be interpreted by two radiologists, and in cases of disagreement, a third radiologist will be consulted to reach a consensus.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Time Frame
sensitivity	through study completion, an average of 2 years

Secondary Outcome Measures

Outcome Measure	Time Frame
specificity	through study completion, an average of 2 years
positive predictive value	through study completion, an average of 2 years
negative predictive value	through study completion, an average of 2 years
total time of interpretation for all the MR images	through study completion, an average of 2 years
the rate of discussion with a third radiologist	through study completion, an average of 2 years
the detection rate of local recurrence in the AI-supported reading group	through study completion, an average of 2 years
the sensitivity in the subgroups of different rT-stage	through study completion, an average of 2 years
the incidence of cases whose recurrent risks and contours cannot be provided by the AI model	through study completion, an average of 2 years

Collaborators and Investigators

• Sponsor: Sun Yat-sen University
• Investigators: Principal Investigator: Fang-Yun Xie, Sun Yat-sen University

Publications and helpful links

General Publications

• OuYang PY, He Y, Guo JG, Liu JN, Wang ZL, Li A, Li J, Yang SS, Zhang X, Fan W, Wu YS, Liu ZQ, Zhang BY, Zhao YN, Gao MY, Zhang WJ, Xie CM, Xie FY. Artificial intelligence aided precise detection of local recurrence on MRI for nasopharyngeal carcinoma: a multicenter cohort study. EClinicalMedicine. 2023 Aug 30;63:102202. doi: 10.1016/j.eclinm.2023.102202. eCollection 2023 Sep.

Study record dates

Study Major Dates

• Study Start (Estimated): April 1, 2024
• Primary Completion (Estimated): April 1, 2026
• Study Completion (Estimated): April 1, 2026

Study Registration Dates

• First Submitted: April 4, 2024
• First Submitted That Met QC Criteria: April 9, 2024
• First Posted (Actual): April 10, 2024

Study Record Updates

• Last Update Posted (Actual): April 10, 2024
• Last Update Submitted That Met QC Criteria: April 9, 2024
• Last Verified: April 1, 2024

More Information

Terms related to this study

• Keywords: magnetic resonance imaging; artificial intelligence; image reviewing
• Additional Relevant MeSH Terms: Pathologic Processes; Neoplasms by Histologic Type; Neoplasms; Neoplasms by Site; Carcinoma; Neoplasms, Glandular and Epithelial; Disease Attributes; Pharyngeal Neoplasms; Otorhinolaryngologic Neoplasms; Head and Neck Neoplasms; Nasopharyngeal Diseases; Pharyngeal Diseases; Stomatognathic Diseases; Otorhinolaryngologic Diseases; Nasopharyngeal Neoplasms; Nasopharyngeal Carcinoma; Recurrence
• Other Study ID Numbers: B2024-039-01

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No