Limits...
New statistical method to analyze three-dimensional landmark configurations obtained with cone-beam CT: basic features and clinical application for rapid maxillary expansion.

Gamble J, Lagravère MO, Major PW, Heo G - Korean J Radiol (2012)

Bottom Line: A method using tools from persistent homology and dimensionality reduction is presented and used to identify variability between the subjects.The analysis was in agreement with previous results using conventional methods, which found significant differences between treatment groups and the control, but no distinction between the types of treatment.This method of analysis allows entire configurations to be considered as a whole, and does not require specific inter-landmark distances or angles to be selected.

View Article: PubMed Central - PubMed

Affiliation: Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Canada.

ABSTRACT

Objective: To describe a statistical method of three-dimensional landmark configuration data and apply it to an orthodontic data set comparing two types of rapid maxillary expansion (RME) treatments.

Materials and methods: Landmark configurations obtained from cone beam CT scans were used to represent patients in two types (please describe what were two types) of RME groups and a control group over four time points. A method using tools from persistent homology and dimensionality reduction is presented and used to identify variability between the subjects.

Results: The analysis was in agreement with previous results using conventional methods, which found significant differences between treatment groups and the control, but no distinction between the types of treatment. Additionally, it was found that second molar eruption varied considerably between the subjects, and this has not been evaluated in previous analyses.

Conclusion: This method of analysis allows entire configurations to be considered as a whole, and does not require specific inter-landmark distances or angles to be selected. Sources of variability present themselves, without having to be individually sought after. This method is suggested as an additional tool for the analysis of landmark configuration data.

Show MeSH

Related in: MedlinePlus

Correlations > 0.55 between inter-landmark distances and first β0 coordinate, colour-coded by strength. This coordinate is most associated with overall size, particularly vertically (and diagonally).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3303895&req=5

Figure 4: Correlations > 0.55 between inter-landmark distances and first β0 coordinate, colour-coded by strength. This coordinate is most associated with overall size, particularly vertically (and diagonally).

Mentions: The first β0 coordinate represents the size and number of zero-dimensional topological features (i.e. the size and number of separate components). Since all the configurations consist of one component (the craniofacial complex), this coordinate is mainly associated with size, and with inter-landmark distances that are far apart. Figure 4 shows the inter-landmark distances that are most strongly correlated (r > 0.55) with the first β0 coordinate run diagonally and vertically between the infra orbital and menton landmarks, and between the menton and maxillary landmarks. It seems that this coordinate is an overall size measure, but is particularly related to vertical length. This is confirmed in Figure 5 which shows the outline of the mean shape obtained from the landmark configurations with the 10% lowest, middle, and highest first β0 coordinate values. It can be seen that the largest coordinate values correspond to relatively higher infra orbital landmarks, and lower menton landmarks, indicating a vertical elongation compared to the other subjects. To see if there are any significant differences between the treatment groups or over time with respect to this coordinate, a repeated measures ANOVA was performed using the factors time and group, as well as the time*group interaction. Overall, this coordinate shows significant changes over time (p = 0.001 approx.), and how the change occurs is not the same for all groups (interaction significant, p = 0.024). In this coordinate, differences were most pronounced at Time 2, when the bone-anchored group was significantly different from the control group (p = 0.033), and the difference between the control and tooth-anchored groups was weakly significant (p = 0.077). No significant differences between groups were seen at time points 3 and 4, although looking at the profile plot (Fig. 6), the mean values for the two treatment groups seem to be higher than for the control group.


New statistical method to analyze three-dimensional landmark configurations obtained with cone-beam CT: basic features and clinical application for rapid maxillary expansion.

Gamble J, Lagravère MO, Major PW, Heo G - Korean J Radiol (2012)

Correlations > 0.55 between inter-landmark distances and first β0 coordinate, colour-coded by strength. This coordinate is most associated with overall size, particularly vertically (and diagonally).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3303895&req=5

Figure 4: Correlations > 0.55 between inter-landmark distances and first β0 coordinate, colour-coded by strength. This coordinate is most associated with overall size, particularly vertically (and diagonally).
Mentions: The first β0 coordinate represents the size and number of zero-dimensional topological features (i.e. the size and number of separate components). Since all the configurations consist of one component (the craniofacial complex), this coordinate is mainly associated with size, and with inter-landmark distances that are far apart. Figure 4 shows the inter-landmark distances that are most strongly correlated (r > 0.55) with the first β0 coordinate run diagonally and vertically between the infra orbital and menton landmarks, and between the menton and maxillary landmarks. It seems that this coordinate is an overall size measure, but is particularly related to vertical length. This is confirmed in Figure 5 which shows the outline of the mean shape obtained from the landmark configurations with the 10% lowest, middle, and highest first β0 coordinate values. It can be seen that the largest coordinate values correspond to relatively higher infra orbital landmarks, and lower menton landmarks, indicating a vertical elongation compared to the other subjects. To see if there are any significant differences between the treatment groups or over time with respect to this coordinate, a repeated measures ANOVA was performed using the factors time and group, as well as the time*group interaction. Overall, this coordinate shows significant changes over time (p = 0.001 approx.), and how the change occurs is not the same for all groups (interaction significant, p = 0.024). In this coordinate, differences were most pronounced at Time 2, when the bone-anchored group was significantly different from the control group (p = 0.033), and the difference between the control and tooth-anchored groups was weakly significant (p = 0.077). No significant differences between groups were seen at time points 3 and 4, although looking at the profile plot (Fig. 6), the mean values for the two treatment groups seem to be higher than for the control group.

Bottom Line: A method using tools from persistent homology and dimensionality reduction is presented and used to identify variability between the subjects.The analysis was in agreement with previous results using conventional methods, which found significant differences between treatment groups and the control, but no distinction between the types of treatment.This method of analysis allows entire configurations to be considered as a whole, and does not require specific inter-landmark distances or angles to be selected.

View Article: PubMed Central - PubMed

Affiliation: Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Canada.

ABSTRACT

Objective: To describe a statistical method of three-dimensional landmark configuration data and apply it to an orthodontic data set comparing two types of rapid maxillary expansion (RME) treatments.

Materials and methods: Landmark configurations obtained from cone beam CT scans were used to represent patients in two types (please describe what were two types) of RME groups and a control group over four time points. A method using tools from persistent homology and dimensionality reduction is presented and used to identify variability between the subjects.

Results: The analysis was in agreement with previous results using conventional methods, which found significant differences between treatment groups and the control, but no distinction between the types of treatment. Additionally, it was found that second molar eruption varied considerably between the subjects, and this has not been evaluated in previous analyses.

Conclusion: This method of analysis allows entire configurations to be considered as a whole, and does not require specific inter-landmark distances or angles to be selected. Sources of variability present themselves, without having to be individually sought after. This method is suggested as an additional tool for the analysis of landmark configuration data.

Show MeSH
Related in: MedlinePlus