Automatic modeling of pectus excavatum corrective prosthesis using artificial neural networks
Pedro L Rodrigues1, Nuno F Rodrigues2, A C M Pinho3
1ICVS/3B's-PT Government Associate Laboratory, Braga/Guimarães, Portugal; Algoritmi Center, School of Engineering, University of Minho, Guimarães, Portugal; DIGARC-Polytechnic Institute of Cávado and Ave, Barcelos, Portugal.
Medical Engineering & Physics
|July 30, 2014
Summary
This study introduces a radiation-free method using 3D laser scanning to replace Computed Tomography (CT) for pectus excavatum prosthesis modeling. The novel approach accurately determines rib positioning, offering a safer alternative for chest wall reconstruction.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Thoracic Surgery
Background:
- Pectus excavatum is a common chest wall deformity requiring surgical correction.
- Current pre-operative diagnosis relies on Computed Tomography (CT), involving radiation exposure.
- Thoracic prosthesis modeling for anterior chest wall remodeling necessitates accurate anatomical data.
Purpose of the Study:
- To develop and validate a radiation-free methodology for thoracic prosthesis modeling in pectus excavatum.
- To replace Computed Tomography (CT) with a 3D laser scanner for improved patient safety.
- To achieve accurate determination of rib positioning and prosthesis placement regions using surface data.
Main Methods:
- A novel methodology utilizing a 3D laser scanner to capture skin surface points.
- Development of an artificial neural network (ANN) set for data analysis and prediction.
- Training ANN models with data from 165 male patients, incorporating soft tissue thickness (STT) measurements.
Main Results:
- The 3D laser scanning method estimated rib position with an average error of 5.0 ± 3.6mm.
- Incorporating manual initial STT values improved ANN performance, reducing the average error to 2.82 ± 0.76 mm.
- The achieved error is significantly lower than current manual prosthesis modeling (approx. 11 mm).
Conclusions:
- A 3D laser scanner can effectively replace CT for radiation-free pectus excavatum prosthesis modeling.
- The ANN-based approach provides accurate estimation of rib positioning for prosthesis personalization.
- This radiation-free procedure offers a valuable advancement in personalized chest wall reconstruction.
