Related Experiment Video
Updated: Sep 17, 2025
14:15
Multicolor 3D Printing of Complex Intracranial Tumors in Neurosurgery
Published on: January 11, 2020
7.2K
Automated Sella-Turcica Annotation and Mesh Alignment of 3D Stereophotographs for Craniosynostosis Patients Using a
Freek Bielevelt1,2, Najiba Chargi2, Joelle van Aalst1
1Radboudumc 3D Lab, Radboud University Medical Center.
The Journal of Craniofacial Surgery
|June 27, 2025
Summary
A new method using Principal Component Analysis (PCA) and a Feedforward Neural Network (FFNN) accurately predicts Sella turcica coordinates from 3D cranial models. This noninvasive approach improves craniosynostosis assessment compared to traditional methods.
Area of Science:
- Medical Imaging
- Computational Biology
- Pediatric Neurosurgery
Background:
- Craniosynostosis requires early diagnosis and treatment for neurological and developmental complications.
- Traditional CT scans for cranial assessment involve ionizing radiation exposure.
- 3D stereophotogrammetry offers a noninvasive alternative but faces challenges in aligning 3D models to standard reference frames like the Sella-turcica-Nasion (S-N) frame.
Purpose of the Study:
- To develop and validate a novel method for predicting the Sella turcica (ST) coordinate from 3D cranial surface models.
- To compare the accuracy of the proposed method against the conventional Computed Cranial Focal Point (CCFP) method.
- To assess the performance of the new method in aligning 3D cranial models within the S-N reference frame.
Main Methods:
- Utilized Principal Component Analysis (PCA) combined with a Feedforward Neural Network (FFNN) to predict ST coordinates.
- Trained and tested the PCA-FFNN model on a dataset of 153 CT scans, including 68 craniosynostosis subjects.
- Compared the PCA-FFNN method's accuracy against the CCFP method, particularly for asymmetric cranial deformations.
Main Results:
- The PCA-FFNN approach demonstrated significantly lower deviations in ST coordinate predictions (3.61 mm) compared to CCFP (8.38 mm, P<0.001).
- Improved accuracy was observed in ST coordinate predictions along the y-axes and z-axes using the PCA-FFNN method.
- Mesh realignment within the S-N reference frame showed enhanced accuracy with the PCA-FFNN method, indicated by reduced mean deviations and dispersion.
Conclusions:
- The PCA-FFNN approach offers a more reliable and noninvasive solution for cranial assessment in craniosynostosis.
- This method has the potential to improve the accuracy of craniosynostosis follow-up.
- Enhanced clinical outcomes can be achieved through more precise and nonionizing cranial assessment techniques.
Keywords:
3D stereophotogrammetrycraniosynostosisfeedforward neural networkprincipal component analysissella turcica
