Patient demographics

The cohort included 15 men and 15 women. Their mean age was M = 24 years (SD =  ± 2.3), mean height M = 176 cm (SD =  ± 8 cm), mean weight M = 69.6 kg (SD =  ± 14.0 kg), and mean BMI M = 22.5 (SD =  ± 3.6).

Landmark-to-Landmark Distance AnalysesComparison of vectra M5- and smartphone-based SMs

Table 2 presents the outcomes of the landmark-to-landmark distance analyses.

Table 2 Descriptive Statistics: Landmark-to-landmark distance analyses; comparison of Vectra M5- and smartphone-based SMs; values in millimeters (mm) for TrueDepth-based and photogrammetry-based measurements (1) – (16); IBM SPSS 29 was used for data analysis

The mean value for all landmark-to-landmark distances (16) of photogrammetry-based SMs to Vectra-based SMs was calculated at M = 0.8 mm (SD =  ± 0.58 mm, n = 450; Table 2). The highest deviation was found in measurement (14) (left cheilion to left cheilion) M = 1.32 mm (SD =  ± 1.02 mm, n = 30; Table 2).

The mean value for all landmark-to-landmark distances (16) between TrueDepth-based SMs and Vectra-based SMs was calculated at M = 1.1 mm (SD =  ± 0.72 mm, n = 450; Table 2). The highest deviation was found in measurement (14) (left cheilion to left cheilion) M = 1.5 mm (SD =  ± 0.95 mm, n = 30; Table 2).

All landmark-to-landmark measurements (1) – (16) remained within a clinically acceptable range, exhibiting an overall landmark-to-landmark deviation of ≤ 2 mm, when comparing both TrueDepth- and photogrammetry-based SMs with Vectra-based SMs (Table 2).

Comparison of truedepth- and photogrammetry-based SMs

Table 3 presents the outcomes of landmark-to-landmark distance analyses, when comparing TrueDepth- with photogrammetry-based SMs based on their alignment with Vectra-based SMs.

Table 3 Landmark–to–landmark distances: 95% Bland–Altman LoA for measurement (1) – (16); and Wilcoxon signed-rank test for paired samples for measurements (1) – (16); values in millimeters (mm); IBM SPSS 29 was used for data analysis

Seven out of 16 measurements exceeded the clinically acceptable 95% Bland–Altman LoA of ≤ 2 mm. However, when contrasting the mean landmark-to-landmark deviation across all distances (16) of TrueDepth- and photogrammetry-based SMs, based on their alignment with Vectra-based SMs, the results indicate a clinically acceptable 95% Bland–Altman LoA of 1.35 mm to −2.0 mm (Table 3). The Wilcoxon signed-rank test for paired samples indicated that the deviation across all landmark-to-landmark distances (16) of photogrammetry-based measurements (median = 0.66 mm) was significantly lower than for TrueDepth-based measurements (median = 0.98 mm; Wilcoxon signed-rank test for paired samples; p =  < 0.001, n = 450; Table 3). Figure 8 shows the Bland–Altman plots for the landmark-to-landmark measurements (1) – (16).

Fig. 8

Bland–Altman Plots: Comparison of TrueDepth- and Photogrammetry-based SMs. Measurements (1) – (20); Values in millimeters (mm) for measurement (1) – (16) and cubic centimeters (cc) for measurements (17) – (20). MS Excel was used to create the illustration

Volumetric analysesComparison of vectra M5- and smartphone-based SMs

Table 4 presents the outcomes of the volumetric deviation analyses.

Table 4 Descriptive Statistics: Volumetric difference between superimposed SMs; comparison of Vectra M5- and smartphone-based SMs; values in cubic centimeters (cc) for TrueDepth-based and photogrammetry-based measurements (17) – (20); IBM SPSS 29 was used for data analysis

The mean volumetric difference across all volumetric measurements (20) comparing photogrammetry-based SMs to Vectra-based SMs was calculated at M = 1.8 cc (SD =  ± 2.12 cc, n = 90). The highest deviation occurred in measurement (18) (mid-face) with M = 2.16 cc (SD =  ± 2.34 cc, n = 30; Table 4). All photogrammetry-based volumetric differences except measurement (18) (midface) remained within a clinically acceptable range, exhibiting a volumetric difference of ≤ 2 cc, when comparing photogrammetry-based SMs with Vectra-based SMs.

The mean volumetric difference across all volumetric measurements (20) for TrueDepth-based SMs compared to Vectra-based SMs was calculated at M = 3.1 cc (SD =  ± 2.64 cc, n = 90). The highest deviation was observed in measurement (18) (mid-face) with M = 4.7 cc (SD =  ± 2.86 cc, n = 30; Table 4). TrueDepth-based volumetric differences exceeded the clinically acceptable range for the overall accuracy (20), the upper- (17) and mid-face (18), exhibiting an average volumetric deviation of > 2 cc, when comparing TrueDepth-based SMs with Vectra-based SMs. However, values for the lower face (19) remained within the clinically acceptable volumetric difference of ≤ 2 cc, when comparing TrueDepth-based SMs with Vectra-based SMs (Table 4).

Comparison of TrueDepth- and Photogrammetry-based SMs

Table 5 presents the outcomes of volumetric deviation analyses, when comparing TrueDepth- with photogrammetry-based SMs based on their alignment with Vectra-based SMs.

Table 5 Volumetric measurements: 95% Bland–Altman LoA for measurements (17) – (20); and Wilcoxon signed-rank test for paired samples for measurements (17) – (20); values in cubic centimeters (cc); IBM SPSS 29 was used for data analysis

All volumetric measurements exceeded the ≤ 2 cc 95% Bland–Altman LoA, with the highest deviation identified in the mid-face, ranging from 4.73 cc to −9.81 cc (Table 5). The Wilcoxon signed-rank test for paired samples revealed a significant difference in volumetric distances in the upper face and mid-face between the two approaches (Table 5). When contrasting the volumetric differences across all regions (20) of TrueDepth- and photogrammetry-based SMs based on their alignment with Vectra-based SMs, the results indicated a clinically unacceptable 95% Bland–Altman LoA of 4.9 cc to −7.6 cc (> 2 cc) (Table 5). The Wilcoxon signed-rank test for paired samples indicated that the deviation across all volumetric distances (20) of photogrammetry-based measurements (median = 1.14 cc) was significantly lower than for TrueDepth-based measurements (median = 2.12 cc) (Wilcoxon signed-rank test for paired samples; p =  < 0.001, n = 90; Table 5).

Figure 8 shows the Bland–Altman plots for the volumetric distances (17) – (20).

Inter-Observer ReliabilityPhotogrammetry-based measurements

Table 6 presents the inter-observer reliability of photogrammetry-based measurements. All photogrammetry-based landmark-to-landmark measurements demonstrated good to excellent correlation, with ICC values ranging from 0.70 to 0.97. Landmark-to-landmark measurements showed a clinically acceptable 95% Bland Altman LoA of ≤ 2 mm. The Wilcoxon signed-rank test revealed no statistically significant differences between the two observers for measurements (1) – (16) (Table 6).

Table 6 Inter-observer reliability of photogrammetry-based measurements: Bland–Altman analysis, Wilcoxon signed-rank test and Intraclass Correlation Coefficient (ICC). Median values and mean bias in millimeters (mm) for measurements (1) – (16) and in cubic centimeters (cc) for measurements (17) – (20). OP 1 = Observer 1, OP 2 = Observer 2; IBM SPSS 29 was used for data analysis

Volumetric assessments conducted by the two observers exhibited excellent correlation, with ICC values ranging from 0.96 to 0.97. All photogrammetry-based volumetric measurements, except for measurement (18) (midface), displayed a 95% Bland Altman LoA of ≤ 2 cc. However, the Wilcoxon signed-rank test for paired samples indicated that the deviation across all volumetric distances (20) differed significantly between the two observers (Wilcoxon signed-rank test for paired samples; p = 0.007, n = 90; Table 6).

Figure 9 presents the Bland–Altman plots illustrating the inter-observer reliability of photogrammetry-based measurements.

Fig. 9

Bland–Altman Plots: Inter-observer reliability of photogrammetry-based measurements. Measurements (1) – (20); Values in millimeters (mm) for measurement (1) – (16) and cubic centimeters (cc) for measurements (17) – (20). MS Excel was used to create the illustration

TrueDepth-based measurements

Table 7 presents the inter-observer reliability of TrueDepth-based measurements. The majority of landmark-to-landmark measurements ((1) – (8) and (10) – (16)) demonstrated good to excellent correlation, with ICC values ranging from 0.64 to 0.97. Measurement (9) showed fair correlation between the two observers. All landmark-to-landmark measurements displayed clinically acceptable 95% Bland–Altman LoA of ≤ 2 mm. The Wilcoxon signed-rank test revealed no statistically significant differences between the two observers for measurements (1) – (15). However, the Wilcoxon signed-rank test indicated a statistically significant difference for the deviation across all landmark-to-landmark distances (16) (Wilcoxon signed-rank test for paired samples; p < 0.001, n = 90; Table 7).

Table 7 Inter-observer reliability of TrueDepth-based measurements: Bland–Altman analysis, Wilcoxon signed-rank test and Intraclass Correlation Coefficient (ICC). Median values and mean bias in millimeters (mm) for measurements (1) – (16) and in cubic centimeters (cc) for measurements (17) – (20). OP 1 = Observer 1, OP 2 = Observer 2; IBM SPSS 29 was used for data analysis

For volumetric assessments conducted by the two observers, excellent correlation was observed for measurements (17) (upper face), (18) (midface), and (20) (overall volume). The Wilcoxon signed-rank test revealed no statistically significant differences between the two observers for all volumetric measurements (Table 7). However, all TrueDepth-based volumetric measurements exceeded the clinically acceptable 95% Bland–Altman LoA of ≤ 2 cc between the two observers (Table 7).

Figure 10 displays the Bland–Altman plots for inter-observer reliability of TrueDepth-based measurements.

Fig. 10

Bland–Altman Plots: Inter-observer reliability of TrueDepth-based measurements. Measurements (1) – (20); Values in millimeters (mm) for measurement (1) – (16) and cubic centimeters (cc) for measurements (17) – (20). MS Excel was used to create the illustration