Pitt Logo LinkContact Us

DIGITAL THREE-DIMENSIONAL PHOTOGRAMMETRY: ACCURACY AND PRECISION OF FACIAL MEASUREMENTS OBTAINED FROM TWO COMMERCIALLY-AVAILABLE IMAGING SYSTEMS

Gornick, Matthew Charles (2011) DIGITAL THREE-DIMENSIONAL PHOTOGRAMMETRY: ACCURACY AND PRECISION OF FACIAL MEASUREMENTS OBTAINED FROM TWO COMMERCIALLY-AVAILABLE IMAGING SYSTEMS. Master's Thesis, University of Pittsburgh.

[img]
Preview
PDF - Primary Text
Download (4Mb) | Preview

    Abstract

    Traditionally, direct anthropometry, two-dimensional (2D) photogrammetry and cephalometry have served as primary methods to quantify craniofacial characteristics. Stereophotogrammetry, a more recent method, is able to capture a three-dimensional (3D) image of a subject's facial surface almost instantaneously. This image can then later be measured in a variety ways, allowing the calculation of linear distances and the quantification of angles, surface areas and volumes. Several 3D stereophotogrammetric systems are commercially available and although some systems have been independently validated, little is known about how measurement data generated by different systems compare. The objective of this study is to evaluate the accuracy and precision of craniofacial measurements obtained using different 3D stereophotogrammetry systems (3dMDface and Vectra 3D) by comparing their values to each other and to measurements obtained using a Microscribe mechanical digitizer. The study sample consisted of 18 mannequin heads, pre-labeled with 28 anthropometric landmarks. All possible inter-landmark distances (n = 378) were calculated and several error magnitude statistics were used to compare facial measurement techniques: mean absolute difference (MAD), relative error magnitude (REM), technical error of measurement (TEM) and intraclass correlation coefficient (ICC). Overall, measurements across all three facial measurement techniques were highly comparable. Over 99% of MAD values were less than 1 mm and over 99% of REM scores were deemed excellent or very good (REM < 4%). Similarly, 100% of TEM values were less than 1 mm and the average ICC across all 378 measures was above 0.99 for all possible method comparisons. Based on the constructed confidence intervals, none of the observed MAD, REM, TEM or ICC values for any of the 378 variables significantly exceeded our predefined error thresholds (p > 0.05). Thermal maps depicting 3D surface-to-surface comparisons also showed negligible differences, with an average Root Mean Squared value across all 18 3D models of 0.197 mm. Results indicate that measurements derived from the Vectra-3D and 3dMDface imaging systems are virtually identical. Furthermore, both systems demonstrated similarly high levels of accuracy when compared to the Microscribe digitizer. Both imaging systems produce facial measurements sufficiently similar to allow for their data to be combined or compared statistically.


    Share

    Citation/Export:
    Social Networking:

    Details

    Item Type: University of Pittsburgh ETD
    ETD Committee:
    ETD Committee TypeCommittee MemberEmail
    Committee ChairWeinberg, Sethsmwst46@pitt.edu
    Committee MemberMooney, Markmpm4@pitt.edu
    Committee MemberDoerfler, Richardrjdortho@comcast.net
    Title: DIGITAL THREE-DIMENSIONAL PHOTOGRAMMETRY: ACCURACY AND PRECISION OF FACIAL MEASUREMENTS OBTAINED FROM TWO COMMERCIALLY-AVAILABLE IMAGING SYSTEMS
    Status: Unpublished
    Abstract: Traditionally, direct anthropometry, two-dimensional (2D) photogrammetry and cephalometry have served as primary methods to quantify craniofacial characteristics. Stereophotogrammetry, a more recent method, is able to capture a three-dimensional (3D) image of a subject's facial surface almost instantaneously. This image can then later be measured in a variety ways, allowing the calculation of linear distances and the quantification of angles, surface areas and volumes. Several 3D stereophotogrammetric systems are commercially available and although some systems have been independently validated, little is known about how measurement data generated by different systems compare. The objective of this study is to evaluate the accuracy and precision of craniofacial measurements obtained using different 3D stereophotogrammetry systems (3dMDface and Vectra 3D) by comparing their values to each other and to measurements obtained using a Microscribe mechanical digitizer. The study sample consisted of 18 mannequin heads, pre-labeled with 28 anthropometric landmarks. All possible inter-landmark distances (n = 378) were calculated and several error magnitude statistics were used to compare facial measurement techniques: mean absolute difference (MAD), relative error magnitude (REM), technical error of measurement (TEM) and intraclass correlation coefficient (ICC). Overall, measurements across all three facial measurement techniques were highly comparable. Over 99% of MAD values were less than 1 mm and over 99% of REM scores were deemed excellent or very good (REM < 4%). Similarly, 100% of TEM values were less than 1 mm and the average ICC across all 378 measures was above 0.99 for all possible method comparisons. Based on the constructed confidence intervals, none of the observed MAD, REM, TEM or ICC values for any of the 378 variables significantly exceeded our predefined error thresholds (p > 0.05). Thermal maps depicting 3D surface-to-surface comparisons also showed negligible differences, with an average Root Mean Squared value across all 18 3D models of 0.197 mm. Results indicate that measurements derived from the Vectra-3D and 3dMDface imaging systems are virtually identical. Furthermore, both systems demonstrated similarly high levels of accuracy when compared to the Microscribe digitizer. Both imaging systems produce facial measurements sufficiently similar to allow for their data to be combined or compared statistically.
    Date: 22 June 2011
    Date Type: Completion
    Defense Date: 29 April 2011
    Approval Date: 22 June 2011
    Submission Date: 20 June 2011
    Access Restriction: No restriction; Release the ETD for access worldwide immediately.
    Patent pending: No
    Institution: University of Pittsburgh
    Thesis Type: Master's Thesis
    Refereed: Yes
    Degree: MDS - Master of Dental Science
    URN: etd-06202011-095206
    Uncontrolled Keywords: morphometrics; validation
    Schools and Programs: School of Dental Medicine > Dental Science
    Date Deposited: 10 Nov 2011 14:48
    Last Modified: 19 Jun 2012 11:02
    Other ID: http://etd.library.pitt.edu/ETD/available/etd-06202011-095206/, etd-06202011-095206

    Actions (login required)

    View Item

    Document Downloads