The comparison of radiographic dental characteristics, mid-palatal suture morphology and development of artificial neural network model in children with unilateral cleft lip and palate with non-cleft individuals
Cleft lip and palate (CLP) refers to a condition characterized by the lack of union between the upper lip and the roof of the mouth. This congenital defect may present with significant alterations in their shape and extent. Among different types of CLP, unilateral cleft lip and palate (UCLP) was the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2024
|
| Subjects: | |
| Online Access: | http://eprints.usm.my/60862/1/MOHAMED%20ZAHOOR%20UL%20HUQH-FINAL%20THESIS%20P-SGD000620%28R%29-E.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Cleft lip and palate (CLP) refers to a condition characterized by the lack of union between the upper lip and the roof of the mouth. This congenital defect may present with significant alterations in their shape and extent. Among different types of CLP, unilateral cleft lip and palate (UCLP) was the most frequently observed. In Malaysia, the overall prevalence rate of CLP is estimated to be one in every 611 live births. CLP impacts craniofacial development in three planes: sagittal, vertical, and transverse planes. There are several techniques for assessing the craniofacial system, maxillary morphometry, dental arch relationships, and distinct features of CLP such as cephalometric radiographs, cone-beam computed tomography (CBCT), and maxillofacial morphologic parameters. In the first phase of the study, 100 subjects (50 cases and 50 controls) aged between 8-16 years were recruited; data were obtained from patients’ records of Hospital Universiti Sains Malaysia (Hospital USM) and Hospital Raja Perempuan Zainab II (HRPZ II). An artificial intelligence (AI) enabled WebCeph software was used to compare the 14 dental characteristics (DC) between UCLP and NC individuals. A dataset was created based on socio-demographic factors such as age, gender, cleft type, and category of malocclusion associated with the cleft. A model was developed by incorporating a novel method that combines a bootstrap, the multi-layer feed-forward neural network (MLFFNN), and ordinal logistic regression (OLR) utilizing R-syntax. The main goal was to predict the maxillary arch growth and evaluate the maxillary arch dimensions in children with UCLP and NC. In the second phase of the study, the mid-palatal suture (MPS) morphology via its bone densities was compared between UCLP and NC individuals. The CBCT images of the CLP patients which were obtained for diagnostic purposes and also CBCT images of the non-cleft (NC) individuals of the same age group were collected from patients who attended specialist clinics before receiving orthodontic treatment. In this phase, the advanced strategy was implemented in three sections, including the creation of R-syntax for the biometry hybrid approach consisting of data bootstrap, MLFFNN, and binary logistic regression (BLR). The purpose of developing a BLR model was to predict the technique of rapid maxillary expansion (RME). The findings revealed significant variations among the 10 DC in the UCLP group compared to the NC group. The hybrid OLR model developed with R-syntax has shown exceptional modelling with a greater accuracy of 97.53% in predicting the maxillary arch growth. Among the MPS maturation stages assessed from the study population, Stage E (37%) was the most prevalent, followed by Stage D (27%) and Stage C (20%), while Stage B and Stage A were equally prominent (8%). The proposed hybrid method with BLR using R-syntax has demonstrated excellent performance of the model with a higher accuracy of 99.98%. The outcome of the study suggests a strong correlation between sex, age, and cleft occurrence. It has been concluded that patients with CLP and cleft palate (CP) have smaller maxillary dimensions in sagittal and transverse planes compared to healthy individuals. The MPS fusion was found to be highest in children aged between 14-16 years. A greater percentage of fusion D (27%) and E (37%) of MPS stages were observed in female children. This information is vital for clinicians in making precise clinical decisions, especially in managing children with UCLP. |
|---|