Midface Hypoplasia and Cranial Base Morphology in Syndromic Craniosynostosis: A Comparative Analysis Study Using a Predictive Regression Model

My research focuses on how age-adjusted midface and cranial-based measurements improve surgical planning for syndromic craniosynostosis. Recent development include morphological modeling and predictive analytics to enable more precise, personalized craniofacial surgical planning. To begin, open the digital imaging archive.

Retrieve retrospective cranial CT scan datasets of pediatric patients diagnosed with syndromic craniosynostosis. Select subjects who meet the inclusion criteria, confirmed syndromic craniosynostosis diagnosis by craniofacial surgeons, complete medical records, and are below 12 years. Similarly, select control subjects with complete cranial and facial CT scans and no history of craniofacial anomalies.

After receiving non-contrast CT skull datasets in DICOM format, ensure imaging protocols meet departmental standards. Archive all verified dataset securely, and prepare them for import into a three-dimensional medical image processing software for reconstruction and analysis. Select 30 subjects and assign them equally into three cohorts, non-operated syndromic craniosynostosis, operated syndromic craniosynostosis, and normal control with 10 subjects in each group.

Display all imported datasets simultaneously in axial, sagittal, and coronal planes, along with a three-dimensional preview. To begin three-dimensional volumetric reconstruction, click New Mask to generate the initial reconstruction. Check the boxes for Fill holes and Keep largest, then press OK.Now, resolve segmentation issues caused by inappropriate Hounsfield unit ranges by applying the standard bone threshold preset.

Refine the segmentation mask further using the Edit Masks function to remove artifacts and ensure structural continuity. Navigate to the Analyze tab and click on Measure and Analyze. Mark the landmarks, the sella S at the midpoint of the sella turcica, the nasion N at the frontonasal suture junction, the basion BA at the anterior midpoint of the occipital bone at the spheno-occipital synchondrosis, and right and left zygomaticomaxillary suture ZMR, ZML at the infraorbital rim.

To confirm each landmark, view it from multiple angles and compare it with the axial, sagittal, and coronal planes for accuracy. Store all marked landmarks automatically in the Measurement section of the project tree. Export the coordinates and labels of each landmark using the Export measurements function to generate datasets in CSV or XLS format.

Select any two anatomical landmarks manually on the 3D reconstruction, allowing the software to automatically calculate and display the Euclidean distance between them in the measurements results window. Next, allow the software to automatically update the measurement list within the project tree upon completion. Export the complete dataset using the Export measurements function and save it in XLS format.

Import the exported linear measurement dataset into a spreadsheet program for analysis. Apply the Hariri-Ros-Nor regression formula to calculate predicted values for total cranial base length and maxillary width. Compare the predicted values with the measured values for each subject to evaluate consistency and morphological deviation.

Calculate the standard deviation of the differences between measured and predicted values to quantify variability in the dataset. Then prepare the finalized dataset for downstream statistical analysis. After compiling all measured and predicted values into a single dataset, import them into statistical analysis software.

Calculate descriptive statistics to summarize cranial base and midfacial measurement across all subject groups. Use the non-parametric Mann-Whitney U test to assess differences in cranial measurements between groups. Conduct a Pearson correlation analysis to determine the relationship between age and cranial measurements within each group.

A strong positive correlation was found between NBa and ZMR-ZML in the normal group, as shown in the scatter matrix, supported by a Pearson correlation of 0.992. Measured NBa values were consistently higher than predicted values across all groups with the greatest deviation observed in the operated syndromic craniosynostosis group. Measured ZMR-ZML values were significantly lower than predicted values in all groups, demonstrating considerable variability.

In the normal group, Pearson correlation analysis revealed strong associations between age in both NBa and ZMR-ZML. In the SC non-operated group, a strong correlation was observed only for ZMR-ZML, while the SC operated group showed weak correlations across all variables. Greater variability and wider interquartile ranges were observed in both NBa and ZMR-ZML values compared to the normal group in the SC operated cohort.

The ratio distributions for the normal control group were clustered around central values, while the operated and non-operated SC groups showed broader distributions with extreme values. The SC operated group displayed a lower median ratio and reduced spread compared to the SC non-operated group. We have established that the age-adjustment improve radiation accuracy of craniofacial morphology.

And this is important for customization or individualization in surgical planning for the patient. So we are addressing gaps in predictive accuracy by integrating each of the patient and his cranial-based measurements as a critical factor in our predictive model. Our protocol offers improved individualized prediction by adjusting for age, unlike static cranial-based models.