Using neural network for restoring the lost surface of skull bones

Objective: To assess the sensitivity, specificity and accuracy of a digital algorithm based on convolutional neural networks used for restoring the lost surface of the skull bones. Materials and methods. The neural network was trained over 6,000 epochs on 78,000 variants of skull models with artif...

Full description

Bibliographic Details
Main Author: Mishinov, Sergey V.
Format: Article
Language:English
Published: Saratov State Medical University 2023-03-01
Series:Saratov Medical Journal
Subjects:
Online Access:https://sarmj.org/sites/default/files/pdf/2023/Saratov_Med_J-2023-0102.pdf
_version_ 1827298911915606016
author Mishinov, Sergey V.
author_facet Mishinov, Sergey V.
author_sort Mishinov, Sergey V.
collection DOAJ
description Objective: To assess the sensitivity, specificity and accuracy of a digital algorithm based on convolutional neural networks used for restoring the lost surface of the skull bones. Materials and methods. The neural network was trained over 6,000 epochs on 78,000 variants of skull models with artificially generated skull injuries. The key parameters of the algorithm were assessed on 222 series of multislice computed tomography (MSCT) of patients with defects of the skull bones, presented in DICOM format. Results. For the group as a whole, the sensitivity, specificity, and accuracy rates were 95.3%, 85.5%, and 79.4%, respectively. Multiple experiments were conducted with a step-by-step elimination of 3D models in order to find the underlying cause of unsatisfactory outcomes of the skull lost surface restoration. Incorrect identification of the defect zone most often occurred in the area of the facial skeleton. After excluding series with the presence of artifacts, the mean increase in metrics was 2.6%. Conclusion. The accuracy of identifying the reference points (specificity) on a 3D model of the skull by the algorithm had the greatest impact on the ultimate accuracy of repairing the lost surface. The maximum accuracy of the algorithm allowing the use of the resulting surfaces without additional processing in a 3D modeling environment was achieved in series without the presence of artifacts in computed tomography (83.5%), as well as with defects that did not extend to the base of the skull (79.5%).
first_indexed 2024-04-24T15:24:08Z
format Article
id doaj.art-0d93b1e1876f4bb7aeb5e58e2487b102
institution Directory Open Access Journal
issn 2712-8253
language English
last_indexed 2024-04-24T15:24:08Z
publishDate 2023-03-01
publisher Saratov State Medical University
record_format Article
series Saratov Medical Journal
spelling doaj.art-0d93b1e1876f4bb7aeb5e58e2487b1022024-04-02T06:52:28ZengSaratov State Medical UniversitySaratov Medical Journal2712-82532023-03-01411510.15275/sarmj.2023.0102Using neural network for restoring the lost surface of skull bonesMishinov, Sergey V. 0Ya.L. Tsivyan Research Institute of Traumatology and Orthopedics of Novosibirsk, Novosibirsk, RussiaObjective: To assess the sensitivity, specificity and accuracy of a digital algorithm based on convolutional neural networks used for restoring the lost surface of the skull bones. Materials and methods. The neural network was trained over 6,000 epochs on 78,000 variants of skull models with artificially generated skull injuries. The key parameters of the algorithm were assessed on 222 series of multislice computed tomography (MSCT) of patients with defects of the skull bones, presented in DICOM format. Results. For the group as a whole, the sensitivity, specificity, and accuracy rates were 95.3%, 85.5%, and 79.4%, respectively. Multiple experiments were conducted with a step-by-step elimination of 3D models in order to find the underlying cause of unsatisfactory outcomes of the skull lost surface restoration. Incorrect identification of the defect zone most often occurred in the area of the facial skeleton. After excluding series with the presence of artifacts, the mean increase in metrics was 2.6%. Conclusion. The accuracy of identifying the reference points (specificity) on a 3D model of the skull by the algorithm had the greatest impact on the ultimate accuracy of repairing the lost surface. The maximum accuracy of the algorithm allowing the use of the resulting surfaces without additional processing in a 3D modeling environment was achieved in series without the presence of artifacts in computed tomography (83.5%), as well as with defects that did not extend to the base of the skull (79.5%).https://sarmj.org/sites/default/files/pdf/2023/Saratov_Med_J-2023-0102.pdfneurosurgerycranioplasty3d modeling
spellingShingle Mishinov, Sergey V.
Using neural network for restoring the lost surface of skull bones
Saratov Medical Journal
neurosurgery
cranioplasty
3d modeling
title Using neural network for restoring the lost surface of skull bones
title_full Using neural network for restoring the lost surface of skull bones
title_fullStr Using neural network for restoring the lost surface of skull bones
title_full_unstemmed Using neural network for restoring the lost surface of skull bones
title_short Using neural network for restoring the lost surface of skull bones
title_sort using neural network for restoring the lost surface of skull bones
topic neurosurgery
cranioplasty
3d modeling
url https://sarmj.org/sites/default/files/pdf/2023/Saratov_Med_J-2023-0102.pdf
work_keys_str_mv AT mishinovsergeyv usingneuralnetworkforrestoringthelostsurfaceofskullbones