Academia.eduAcademia.edu

Figure from:Determination of stature from skeletal and skull measurements by CT scan evaluation

See full PDFdownloadDownload figure
Single regression equations and multiple regression equations obtained from combined data of two and three bones.

Table 3 Single regression equations and multiple regression equations obtained from combined data of two and three bones.

Related Figures (11)

Fig. 1. CT scan of the right lower limb in the coronal plane: measurements of femoral diaphysis length (1) and width (2).
Fig. 1. CT scan of the right lower limb in the coronal plane: measurements of femoral diaphysis length (1) and width (2).
Fig. 2. (A) CT scan of the skull in the coronal plane: measurement of maximum cranial width (1) and distance between the mastoid processes (2); (B) CT scan of the skull in the sagittal plane: Measurement of maximum cranial length (1), length of cranial base (2), distance from the posterior extremity of the cranial base to the nasal bone (3) and frontal sinus height (4).
Fig. 2. (A) CT scan of the skull in the coronal plane: measurement of maximum cranial width (1) and distance between the mastoid processes (2); (B) CT scan of the skull in the sagittal plane: Measurement of maximum cranial length (1), length of cranial base (2), distance from the posterior extremity of the cranial base to the nasal bone (3) and frontal sinus height (4).
Descriptive statistical analysis. Considering all the subjects, the smallest difference between h, and h,, is obtained using femoral length: These differences are greater than 10 cm in less than 15% of all cases, and in about 60% of the cases the difference ranges from —4 cm to +4 cm. The equation based on Ba-N distance generates poorer results of stature estimation: The differences between the estimated statures and the measured values is greater than 10 cm in about 25% of all cases, and in about 55% of the cases the difference ranges from —4 cm to +4 cm; similar results are obtained using the equation based on the Ba-NB distance. Table 2 femoral diaphyseal lengths), Ba-N, and Ba-NB. From this point onward, all the statistical analyses are performed on these three lengths.
Descriptive statistical analysis. Considering all the subjects, the smallest difference between h, and h,, is obtained using femoral length: These differences are greater than 10 cm in less than 15% of all cases, and in about 60% of the cases the difference ranges from —4 cm to +4 cm. The equation based on Ba-N distance generates poorer results of stature estimation: The differences between the estimated statures and the measured values is greater than 10 cm in about 25% of all cases, and in about 55% of the cases the difference ranges from —4 cm to +4 cm; similar results are obtained using the equation based on the Ba-NB distance. Table 2 femoral diaphyseal lengths), Ba-N, and Ba-NB. From this point onward, all the statistical analyses are performed on these three lengths.
Fig. 3. Histograms representing data frequency distribution of Ba-NB, Ba-N, femur length and height.
Fig. 3. Histograms representing data frequency distribution of Ba-NB, Ba-N, femur length and height.
Fig. 4. Plot of height (h,) as a function of the lengths of three bones: (A) femur length; (B) Ba-N distance; (C) Ba-NB distance.
Fig. 4. Plot of height (h,) as a function of the lengths of three bones: (A) femur length; (B) Ba-N distance; (C) Ba-NB distance.
The values of E% were calculated in all three cases of multiple regression (Table 3). They show a very small difference compared to the best result obtained by performing single linear regression: In fact, single regression using the femoral length gives a value of E%= 3.2% and a similar value (2.9%) is obtained by performing multiple regression based on the data of all three bones. In order to estimate stature from the lengths of multiple bones, a multiple regression analysis was performed considering femoral length together with the Ba-N and Ba-NB distances. In particular, the multiple regression analysis was performed considering three cases: femoral length and Ba-N distance, femoral length and Ba-NB distance, and femoral length with both Ba-N and Ba-NB distances. Fig. 5. Values of estimated stature (h-) as a function of real stature (hm) obtained from femoral length (left panel), Ba-N distance (middle panel), and Ba-NB distance (right panel). The line of equality is also shown for each plot.
The values of E% were calculated in all three cases of multiple regression (Table 3). They show a very small difference compared to the best result obtained by performing single linear regression: In fact, single regression using the femoral length gives a value of E%= 3.2% and a similar value (2.9%) is obtained by performing multiple regression based on the data of all three bones. In order to estimate stature from the lengths of multiple bones, a multiple regression analysis was performed considering femoral length together with the Ba-N and Ba-NB distances. In particular, the multiple regression analysis was performed considering three cases: femoral length and Ba-N distance, femoral length and Ba-NB distance, and femoral length with both Ba-N and Ba-NB distances. Fig. 5. Values of estimated stature (h-) as a function of real stature (hm) obtained from femoral length (left panel), Ba-N distance (middle panel), and Ba-NB distance (right panel). The line of equality is also shown for each plot.
Single and multiple regression equations for males and females sample. Table 4
Single and multiple regression equations for males and females sample. Table 4
Fig. 6. Histograms of the differences between real height (hm) and stature estimated from femoral length (left panel), Ba-N distance (center), and Ba-NB distance (right panel).
Fig. 6. Histograms of the differences between real height (hm) and stature estimated from femoral length (left panel), Ba-N distance (center), and Ba-NB distance (right panel).
Fig. 7. Left panel: values of estimated stature (h.) as a function of real stature (h,) obtained by multiple regression using femoral length and Ba-NB distance, the line of equality is also shown. Center: Bland-Altman plot showing the comparison of h,, and he, the two dotted lines represent the mean + standard deviation of he. Right panel: histograms of the differences between h,, and hg.
Fig. 7. Left panel: values of estimated stature (h.) as a function of real stature (h,) obtained by multiple regression using femoral length and Ba-NB distance, the line of equality is also shown. Center: Bland-Altman plot showing the comparison of h,, and he, the two dotted lines represent the mean + standard deviation of he. Right panel: histograms of the differences between h,, and hg.
Fig. 8. Left panel: values of estimated stature (h,) as a function of real stature (h,,) obtained by multiple regression using femoral length and Ba-N distance, the line of equality is also shown. Center: Bland-Altman plot showing the comparison of h,, and he, the two dotted lines represent the mean + standard deviation of he. Right panel: histograms of the differences between h and he. Applying a multiple regression analysis, the best values were obtained using the data of all the three bones together, which resulted in an absolute error of 2.9%. However, we did not notice a significant increase in accuracy using the formula with multiple Above all, femoral length showed the maximum reliable correlation with height (Reem = 0.71); also Ba-N and Ba-NB showed quite correlation (Rgan = 0.53, Rgang = 0.50), and, because of this,
Fig. 8. Left panel: values of estimated stature (h,) as a function of real stature (h,,) obtained by multiple regression using femoral length and Ba-N distance, the line of equality is also shown. Center: Bland-Altman plot showing the comparison of h,, and he, the two dotted lines represent the mean + standard deviation of he. Right panel: histograms of the differences between h and he. Applying a multiple regression analysis, the best values were obtained using the data of all the three bones together, which resulted in an absolute error of 2.9%. However, we did not notice a significant increase in accuracy using the formula with multiple Above all, femoral length showed the maximum reliable correlation with height (Reem = 0.71); also Ba-N and Ba-NB showed quite correlation (Rgan = 0.53, Rgang = 0.50), and, because of this,
Fig. 9. Left panel: values of estimated stature (h.) as a function of real stature (h,) obtained by multiple regression using femoral length and both Ba-NB and Ba-N distances, the line of equality is also shown. Center: Bland-Altman plot showing the comparison of hm and he, the two dotted lines represent the mean + standard deviation of he. Right panel: histograms of the differences between h,, and h,.
Fig. 9. Left panel: values of estimated stature (h.) as a function of real stature (h,) obtained by multiple regression using femoral length and both Ba-NB and Ba-N distances, the line of equality is also shown. Center: Bland-Altman plot showing the comparison of hm and he, the two dotted lines represent the mean + standard deviation of he. Right panel: histograms of the differences between h,, and h,.
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved