NCT05869058

DCNN Developed for Detection and Assessing the Perfusion of PTG

Development and Improvement of a Deep Convolutional Neural Network for Detection and Assessing the Perfusion of Parathyroid Gland During Endoscopic Thyroidectomy

• Status: Recruiting
• Phase: —
• Study type: Observational
• Enrollment: 300 (estimated)
• Sponsor: Sun Yat-Sen Memorial Hospital of Sun Yat-Sen University · Academic / Other
• Sex: All
• Age: 18 Years – 70 Years
• Healthy volunteers: Not accepted

Summary

Since the anatomical location and appearance of the parathyroid gland (PTG) vary, detection of the PTG and preserving the blood supply are among the difficulties encountered during a thyroidectomy procedure. We are planning to train a deep convolutional neural network based on a larger sample of endoscopic images to develop a model to assist surgeons in detection of PTG during endoscopic thyroidectomy. Furthermore, we would like to train a DCNN to predict blood perfusion based on endoscopic images comparing to indocyanine green fluorescence angiography as reference standard, and assess the performance of DCNN in predicting postoperative hypoparathyroidism.

Detailed description

Since the anatomical location and appearance of the parathyroid gland (PTG) vary, detection of the PTG and preserving the blood supply are among the difficulties encountered during a thyroidectomy procedure. Resection of the PTG by mistake or interruption of the blood supply may lead to transient or permanent hypoparathyroidism, which would require short-term or lifelong calcium and/or vitamin D supplement. We are planning to train a deep convolutional neural network based on a larger sample of endoscopic images to develop a model to assist surgeons in detection of PTG during endoscopic thyroidectomy. Although several researchers indicated that indocyanine green fluorescence angiography could be used to assess the perfusion of the PTG intraoperatively, it may cause allergic reaction and need repetitive injection. Therefore, we would like to train a DCNN to predict blood perfusion based on endoscopic images comparing to indocyanine green fluorescence angiography as reference standard, and assess the performance of DCNN in predicting postoperative hypoparathyroidism. This research may lead to the development of endoscopic modules in PTG detection and PTG perfusion prediction to reduce postoperative hypoparathyroidism.

Conditions

• Thyroidectomy

Interventions

Type	Name	Description
DIAGNOSTIC_TEST	a deep convolutional neural network	a deep convolutional neural network developed for detection and assessing the perfusion of parathyroid gland during endoscopic thyroidectomy

Timeline

• Start date: 2023-06-13
• Primary completion: 2026-04-01
• Completion: 2026-10-01
• First posted: 2023-05-22
• Last updated: 2025-12-03

Locations

1 site across 1 country: China