CT-based Radiomics and Deep Learning Models to Predict Thyroid Cartilage Invasion and Patient Prognosis in Laryngeal Carcinoma

Laryngeal Carcinoma Clinical Trial

Official title:

CT-based Radiomics, Two-dimensional and Three-dimensional Deep Learning Models to Predict Thyroid Cartilage Invasion and Patient Prognosis in Laryngeal Carcinoma: a Multicenter Study

Clinical Trial Details — Status: Recruiting

Administrative data

• NCT number: NCT06463756
• Other study ID #: 2024-Chenx
• Status: Recruiting
• Start date: August 13, 2023
• Est. completion date: October 13, 2024

Study information

• Verified date: June 2024
• Source: First Affiliated Hospital of Chongqing Medical University
• Contact: n/a
• Is FDA regulated: No
• Study type: Observational

Clinical Trial Summary

This retrospective study was to develop and verify a CT-based radiomics model, 2D deep learning model and 3D deep learning model to preoperatively predict the thyroid cartilage invasion of laryngeal cancer patients, so as to provide more accurate diagnosis and treatment basis for clinicians. And the performance of the aforementioned models was compared with two radiologists. In addition, the researchers investigated the prediction of survival outcomes of patients by the above optimal models.

Description:

Laryngeal squamous cell carcinoma (LSCC), as one of the most common head and neck tumors, is the eighth leading cause of cancer-associated death worldwide. The treatment decisions has a profound impact on both tumor control and functional prognosis of LSCC patients. And these decisions are primarily based on tumor staging, with the invasion of the thyroid cartilage serving as a crucial determinant. Consequently, the presence of thyroid cartilage invasion indicates an advanced stage (T3 or T4) diagnosis for the LSCC patients. For patients without thyroid cartilage invasion, partial laryngectomy may be considered to preserve laryngeal function. However, for patients with advanced laryngeal carcinoma and thyroid cartilage invasion extending beyond the larynx, total laryngectomy is often necessary to completely remove the tumor and extend survival time. Therefore, accurate assessment of thyroid cartilage invasion is vital for treatment decision-making and prognosis evaluation for LSCC patients. Recently, artificial intelligence, in the form of machine learning and deep learning (DL), has been wildly applied in medical imaging. Radiomics, a prominent role in machine learning, can extract high-dimensional feature data and quantitatively provide valuable information for tumor staging and prognosis. In addition, DL can automatically capture and learn discriminative features through an end-to-end way for accurate prediction. Collectively, the aims of this multicenter study were to develop and validate a radiomics model, a 2D DL model, and a 3D DL model base on the venous-phase CT images, and compare the performance of these models to two radiologists in predicting thyroid cartilage invasion. Additionally, the researchers investigated the prognostic value of the optimal predicted model of thyroid cartilage invasion.

Recruitment information / eligibility

• Status: Recruiting
• Enrollment: 400
• Est. completion date: October 13, 2024
• Est. primary completion date: September 13, 2024
• Accepts healthy volunteers: No
• Gender: All
• Age group: 18 Years to 81 Years

Eligibility

Inclusion Criteria:

1. Availability of complete clinical data

2. Surgery-proven or biopsy-proven diagnosis of laryngeal squamous cell carcinoma

3. CT examination performed within 2 weeks before surgery

Exclusion Criteria:

1. Patients who received preoperative chemotherapy or radiation therapy

2. CT images with significant artifacts

3. Patients with tumor recurrence

4. Patients with a maximum tumor diameter of less than 5 mm

Related Conditions & MeSH terms

• Carcinoma
• Laryngeal Carcinoma
• Laryngeal Neoplasms
• Thyroid Diseases

Intervention

Other:
• AI
Radiomics extracts quantitative information from medical images to generate high-dimensional feature vectors for analysis. It aims to provide insights into disease processes and improve diagnosis. Deep learning utilizes neural networks with multiple layers to learn complex patterns from data. In medical imaging, it enables accurate and efficient analysis for disease detection and diagnosis.

Locations

Country	Name	City	State
China	The First Affiliated Hospital of Chongqing Medical University	Chongqing

Sponsors (2)

Lead Sponsor	Collaborator
First Affiliated Hospital of Chongqing Medical University	Nankai University

Country where clinical trial is conducted

China, 

Outcome

Type	Measure	Description	Time frame	Safety issue
Primary	Area under the curve, AUC	Area under the curve(AUC) is a metric widely used in machine learning and medical research to evaluate the performance of models in binary classification problems. It reflects the ability of a model to identify true positives (True Positives) while avoiding falsely classifying negative examples as positive (False Positives).	Through study completion, an average of 6 months
Secondary	Disease-Free-Survival, DFS	Disease-Free Survival (DFS) refers to the time from the start of randomization (usually the starting point of a clinical trial) to the recurrence of the disease or death of the patient due to disease progression. DFS is an important clinical and statistical indicator used to evaluate the long-term effects of cancer treatment.	The date of surgery and the occurrence of events such as disease progression, the date of the last follow-up, or death from any cause, and the follow-up time was at least 3 years