Development of Three-dimensional Deep Learning for Automatic Design of Skull Implants

Chang Gung Medical Foundation

Status

Enrolling

Conditions

Skull Defect

Treatments

Device: 3D deep learning neural network system

Study type

Observational

Funder types

Other

Identifiers

NCT05603949

202201082B0

Details and patient eligibility

About

This project aims to develop an effective deep learning system to generate numerical implant geometry based on 3D defective skull models from CT scans. This technique is beneficial for the design of implants to repair skull defects above the Frankfort horizontal plane.

Full description

Designing a personalized implant to restore the protective and aesthetic functions of the patient's skull is challenging. The skull defects may be caused by trauma, congenital malformation, infection, and iatrogenic treatments such as decompressive craniectomy, plastic surgery, and tumor resection. The project aims to develop a deep learning system with 3D shape reconstruction capabilities. The system will meet the requirement of designing high-resolution 3D implant numerical models efficiently.

A collection of skull images were used for training the deep learning system. Defective models in the datasets were created by numerically masking areas of intact 3D skull models. The final implant design should be verified by neurosurgeons using 3D printed models.

Enrollment

6 estimated patients

Sex

All

Ages

15 to 80 years old

Volunteers

No Healthy Volunteers

Inclusion criteria