Artificial Intelligence–Based Modeling Can Predict Face Shape Based on Underlying Craniomaxillofacial Bone
Bibliographic record
Abstract
Reconstructing facial deformities is often challenging due to the complex 3-dimensional (3D) anatomy of the craniomaxillofacial skeleton and overlying soft tissue structures. Bilateral injuries cannot benefit from mirroring techniques and as such preinjury information (eg, 2D pictures or 3D imaging) may be utilized to determine or estimate the desired 3D face shape. When patient-specific information is not available, other options such as statistical shape models may be employed; however, these models require registration to a consistent orientation which may be challenging. Artificial intelligence (AI) has been used to identify facial features and generate highly realistic simulated faces. As such, it was hypothesized that AI can be used to predict 3D face shape by learning its relationship with the underlying bone surface anatomy in a subject-specific manner. An automated image processing and AI modeling workflow using a modified 3D UNet was generated to estimate 3D face shape using the underlying bone geometry and additional metadata (eg, body mass index and age) obtained from 5 publicly available computed tomography imaging datasets. Visually, the trained models provided a reasonable prediction of the contour and geometry of the facial tissues. The pipeline achieved a validation dice=0.89 when trained on the combined 5 datasets, with the highest dice=0.925 achieved with the single HNSCC dataset. Estimated predefect facial geometry may ultimately be used to aid preoperative craniomaxillofacial surgical planning, providing geometries for intraoperative templates, guides, navigation, molds, and forming tools. Automated face shape prediction may additionally be useful in forensic studies to aid in the identification of unknown skull remains.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".