Skip to content

Automatic differential diagnosis of fresh and old vertebral compression fractures on thoracolumbar CT based on convolutional neural network

Deep learning model for automated identification of fresh and old vertebral compression fractures on thoracolumbar CT images

Status
Active, not recruiting
Phases
Unknown
Study type
Observational
Source
ChiCTR
Registry ID
ChiCTR2400087318
Enrollment
Unknown
Registered
2024-07-25
Start date
2023-09-01
Completion date
Unknown
Last updated
2024-07-29

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

vertebral compression fractures

Interventions

Gold Standard:Thoracolumbar MRI findings were used as the gold standard by two spine surgeons with more than 10 years of experience. Fresh vertebral fractures are defined as back pain lasting less tha
Index test:1. Construct a deep learning model for automatic segmentation of fractured vertebrae on thoracolumbar CT, and calculate the segmentation accuracy of the model
2. Construct a classification model for the identification of fractured vertebrae after segmentation, and calculate the AUC, accuracy, sensitivity ,specificity,precision, F1 score of the model.

Sponsors

Sun Yat-sen Memorial Hospital, Sun Yat-sen University
Lead Sponsor

Eligibility

Sex/Gender
All
Age
18 Years to 100 Years

Inclusion criteria

Inclusion criteria: ? Patients with benign vertebral compression fractures; ? Clinical and imaging data (thoracolumbar CT) are complete

Exclusion criteria

Exclusion criteria: ? Suspected infection or tumor-related pathological fracture; ? Poor image quality or presence of foreign artifacts; ? Clinical and imaging data (thoracolumbar CT) were missing.

Design outcomes

Primary

MeasureTime frame
Area under ROC curve;Accuracy;Dice Similarity Coefficient;

Secondary

MeasureTime frame
Sensitivity;Specificity;Precision;F1 score;

Countries

China

Contacts

Public ContactDongsheng Huang

Sun Yat-sen Memorial Hospital, Sun Yat-sen University

hdongsh@mail.sysu.edu.cn+86 34071700

Outcome results

None listed

Source: ChiCTR (via WHO ICTRP) · Data processed: Feb 4, 2026