Automatic localization of craniofacial landmarks of cephalograms using artificial neural networks.
Bibliographic record
Abstract
In modern orthodontic practice, a great reliance is placed on objective and systematic methods of characterizing craniofacial forms, using measurements based on a set of agreed upon points known as craniofacial landmarks. Lateral skull x-ray images are usually used in cephalometric analysis to provide quantitative measurements of the head. Accurate location of those landmarks forms the basis for what is known as cephalometric evaluation. Distance and angles among these landmarks are compared with normative values to diagnose a patient's deviations from ideal form, evaluate the craniofacial growth and measure the effect of the treatment. Because of the large variability in the morphology of the human head, large variations of special coordinates of landmarks are observed and must be reduced. To reduce this variation, adaptive localization based on the, size, rotation and shifts of the skull is used. The adaptive system requires a training set that will account for all the variations in the cephalograms. A good training set is difficult to obtain due to unavailability of fixed workbenches of locations of landmarks that cephalometric measurements of x-rays can be compared with. To create a reliable training set, images are grouped into several clusters and one prototype representing that cluster is used in the training set. The work in this thesis reports two novel algorithms for locating craniofacial landmarks on digitized skull x-rays. The first algorithm is based on the use of neuro-fuzzy networks to minimize the search windows for each landmark. Parametric template matching is then used to pin point the exact location of the landmark inside a search window. The second algorithm uses a Multi-Layer Perceptron as a function approximator to predict the location of the landmark based on learned knowledge obtained from a training set. A new method for extracting a features vector from each image is also reported. This feature vector is used to represent images and also used for clustering images to obtain a reliable training set using K-means after providing it with initial estimates of centers of the groups. To reduce the dimension of the feature vector, we provide an efficient pruning technique for reduction of features based on sensitivity analysis. It is shown that this reduction will minimize the number of rules required for the fuzzy system while the clustering characteristics are preserved. Algorithms are simulated using C++ code. Results obtained using the two algorithms are compared with previous works. It is shown that the proposed algorithms outperform other methods found in the open literature.Dept. of Electrical and Computer Engineering. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2003 .E43. Source: Dissertation Abstracts International, Volume: 64-10, Section: B, page: 5116. Advisers: M. A. Sid-Ahmed; M. Ahmadi. Thesis (Ph.D.)--University of Windsor (Canada), 2003.
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 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| 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".