Table 1.
Distribution of age, left and right ear of all CT data.
Table 2.
Distribution of age, left and right ear of CT data for difference of each set of models.
Fig 1.
Creation of a homologous ear model.
(A) Template model using the 3D ear image of a 40s-man, 5,507 corresponding points, 16,521 polygons, and 18 landmark points. (B) Another 3D ear image with the same 18 landmark points. (C) A superimposition image of A and B using 18 landmark points. The vertices of the template model were adjusted to those of the 3D image. (D) Completed homologous ear model.
Fig 2.
Separation of the upper auricle and lobule from the homologous model.
(A) The average image of a homologous model. (B) The upper auricle homologous model after removing the lobule area surrounded by the cauda helicis, the lower part of incisura intertragica, and otobasion inferius. (C) A homologous model for the lobule only.
Fig 3.
Visualization of morphological changes from minus three times to plus three times of the SD value of PC1 detected in PCA of 414 upper auricle models.
SD, standard deviation; PC, principal component; PCA, principal component analysis. (A) Side images. (B) Rotated 40 degrees backward for A. (C) Rotated 40 degrees forward for A. Morphological changes of protrusion of antihelix (arrow a) and curvature of helix (arrow b) are characteristic.
Table 3.
Result of PCA of 414 upper auricle models and the correlation between distributions of PC score and age.
Table 4.
Result of PCA of 414 lobule models and the correlation between distributions of PC score and age.
Fig 4.
Scree plot curve for five sets of models for the upper auricle.
PC, principal component; EV, eigen values. The EVs rapidly decrease until PC5 for the sets with 123, 242, and 300 models (arrow a) and PC6 for the sets with 363 and 414 models (arrow b).
Fig 5.
Slope of the DBV of two images created by minus three times and plus three times of the SD value by each PC for five sets of models for upper auricle models.
PC, principal component; DBV, distance between vertices; SD, standard deviation. The DBV of the sets with 300, 363, and 414 models dropped significantly until PC6 (arrow), and the gradient of the plots of 363 and 414 models was well approximated.
Fig 6.
Scree plot curve of five different sets of models for the lobule.
PC, principal component, EV, eigen values. The EVs decreased sharply for all sets of models until PC6 (arrow).
Fig 7.
Slope of the DBV of two images created by minus three times and plus three times the SD value of each PC of five different sets of models for the lobule.
PC, principal component; DBV, distance between vertices; SD, standard deviation. The DBV of all sets of models dropped significantly until PC6 (arrow).
Fig 8.
Visualization of shape change upper auricle models by superimposition between the average auricle model and those created by three times the SD value for each PC.
PC, principal component; SD, standard deviation. The shape changes (red regions) in 363 and 414 models are similar until PC5 (indicated by the red dotted line).
Fig 9.
Visualization of shape change lobule models by superimposition between the average lobule model and those created by three times of the SD value.
PC, principal component; SD, standard deviation. Shape changes (red regions) of 300, 363, and 414 models are similar until PC6 (indicated by the red dotted line).
Fig 10.
Biplot of PC1 and PC2 used for synthesizing the CIs.
PC, principal component; CI, composite image. Eight upper auricle models are synthesized using single PC values (-100 or +100) of four (2, 4, 5, 7) and two different PC values of four (1, 3, 6, 8) to synthesize the CIs.
Fig 11.
Biplots of PC1 and PC3 (left), PC1 and PC4 (center), PC1 and PC5 (left) used to synthesize the CIs.
PC, principal component; CI, composite image. Eighteen upper auricle models are synthesized using single PC values of six (10, 13, 16, 19, 22, 25) and two different PC values of twelve (9, 11, 12, 14, 15, 17, 18, 20, 21, 23, 24, 26).
Fig 12.
Biplots of PC2 and PC3 (upper left), PC2 and PC4 (upper center), PC2 and PC5 (upper right), PC3 and PC4 (lower left), PC3 and PC5 (lower center), PC4 and PC5 (lower right) used to synthesize the CIs.
PC, principal component; CI, composite image. Twenty-four upper auricle models are synthesized using two different PC values of twenty-four (27–50).
Fig 13.
Scree plot curve of 50 CIs created by each number of models.
PC, principal comportment; EV, Eigen values; CI, composite image; CCI, Classification Criterion Image. EVs strongly decrease at PC6. Since PC6 had a small EV, PC values up to PC5 were used to devise the final CCIs.
Fig 14.
Ten CCIs for each of the three sets of models of the upper auricle devised from single PC values (±100) from PC1–5 detected by PCA of 50 CIs.
PC, principal comportment; PCA, principal component analysis; CI, composite image; CCI, Classification Criterion Image. Each criterion image of 363 and 414 models is similar to each other. However, CIs from 300 and 363 models showed deviations in their shapes.
Fig 15.
Biplot of PC1 and PC2 for lobule.
PC, principal component; CI, composite image. Eight lobule models are synthesized using single PC values (–60 and +60) of four (2, 4, 5, 7) and double PC values of four (1, 3, 6, 8) to create the CIs.
Fig 16.
CCIs made from PCA of 414 ear homologous modeling from the upper auricle and lobule models.
PC, principal component; PCA, principal component analysis; CCI, classification criterion image. Overhanging of the auricle (black line) to the head (red line) is relatively strong in the upper auricles of the upper images of CCI Ⅱ, Ⅳ, and Ⅶ.
Fig 17.
Results of image determination for 414 upper auricle images by ten CCIs.
CCI, classification criterion image.
Fig 18.
Results of image determination for 414 lobule images by twelve CCIs.
CCI, classification criterion image.
Table 5.
Details of the results of the CCIs.