Related Experiment Video
Updated: Sep 23, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Cranio-maxillofacial post-operative face prediction by deep spatial multiband VGG-NET CNN
Rizwan Ali1, Rui Lei1, Haifei Shi2
1Department of Plastic Surgery, The First Affiliated Hospital, School of Medicine, Zhejiang University No. 79 Qingchun Road, Hangzhou 310003, Zhejiang, China.
American Journal of Translational Research
|May 13, 2022
Summary
This study introduces a new deep learning model for accurate post-operative facial prediction in reconstructive surgery. The advanced VGG NET CNN significantly improves prediction accuracy, aiding surgical planning and patient outcomes.
Area of Science:
- Medical Imaging and Computer-Aided Surgery
- Artificial Intelligence in Healthcare
- Plastic and Reconstructive Surgery
Background:
- Current computational techniques in plastic surgery lack precision and are time-consuming, limiting their clinical adoption.
- Existing computer-aided surgical preparation systems are complex and require extensive manual input, hindering their use in clinical decision-making and patient communication.
- Precise pre-operative evaluation using 3D models is crucial for optimal aesthetic and reconstructive outcomes.
Purpose of the Study:
- To develop a novel deep learning architecture for enhanced diagnostics, risk stratification, and post-operative facial prediction in reconstructive surgery.
- To improve the accuracy and efficiency of computational methods in plastic and reconstructive surgery.
- To provide a tool for better pre-operative planning and post-operative outcome simulation.
Main Methods:
- Preprocessing involved weighted adaptive median filtering and Laplacian partial differential equation-based histogram equalization.
- A Smart Restorative Frustum model was used for 3D conversion of target areas for visualization.
- A deep spatial Multiband VGG NET Convolutional Neural Network (CNN) was employed for post-operative face prediction, trained on a large dataset of CT scans and clinical records.
Main Results:
- The proposed deep spatial Multiband VGG NET CNN achieved high post-operative face prediction accuracy.
- Performance metrics, including Jaccard and Dice scores, demonstrated superior accuracy compared to traditional methods.
- The model achieved 93.7% prediction accuracy, 99.9% sensitivity, and 99.8% specificity, outperforming existing approaches.
Conclusions:
- The developed deep learning method offers superior performance for post-operative face prediction in reconstructive surgery.
- The approach provides a generalized framework applicable to similar datasets, potentially advancing computational techniques in the field.
- This technology can enhance clinical decision-making, improve patient communication, and optimize surgical planning for better aesthetic and reconstructive results.
