Role of Artificial Intelligence in Determining Ideal Nasal Tip Projection in Diverse Populations

Abstract

Background:

Nasal tip projection ideals have been based on the Caucasian nose; however, many racial groups have underprojected tips. Several methodologies have been described to measure ideal nasal tip projection. This study aimed to compare various methods to racially diverse groups to elucidate those best suited for race-specific facial morphology.

Methods:

An openly accessible artificial intelligence image generator was used to generate images of Caucasian, Middle Eastern, and African American noses. Tip projection was manually measured using 5 different ideal tip projection methods. One-way analysis of variance with post hoc Tukey honest significant difference test evaluated differences between racial cohorts. One-sample t tests were used for comparisons to previously described aesthetic ideals, and a P value of less than 0.05 was considered statistically significant.

Results:

Twenty lateral images were analyzed per racial cohort (10 men and 10 women). All racial groups—except for male Middle Eastern noses (P = 0.01)—were measured as ideal using the Byrd nasal length proportion method. Caucasian and Middle Eastern noses measured ideal using the nasolabial tip projection method. African American female noses measured ideal using the nasofacial tip projection method. Only Caucasian noses were ideal using the lip tangent method.

Conclusions:

This study suggests that the ideal nasal length to tip length ratio of two-thirds (67%) has generalizability across these racial groups; the lip tangent is an excellent method for evaluating tip projection in Caucasians; and assessment of ideal tip projection in Middle Eastern and African American groups should use regional midface and lower face anatomy to account for differences in facial morphology.

Takeaways

Question: Do previously described ideal nasal tip projection measurement methods—which are based on the Caucasian nose—have generalizability to Middle Eastern and African American populations? Can artificial intelligence (AI) generate noses with accurate and variable racial morphology?

Findings: Using an AI image-generating platform, this study shows that AI can generate racially nuanced noses. This study also shows that various nasal tip measurement methods lead to significantly different measurements for these racial groups.

Meaning: This study suggests that different tip projection measurement methods lead to ideal measurements for different racial groups, which is heavily influenced by facial morphological differences.

INTRODUCTION

Projection of the nasal tip is a key element of nasal aesthetics and is frequently addressed surgically in rhinoplasty. An under- or overprojected nasal tip is often a central element of suboptimal cosmesis when patients are seeking corrective rhinoplasty, whether the patient notices it or not. The rhinoplasty surgeon has several surgical techniques in their armamentarium to optimize nasal tip projection (NTP) for a given patient. However, NTP is also a critical element of the racial identity of one’s nose. Significant deviations in NTP from a patient’s baseline may increase the risk of disharmony with their racial identity.

The nose is an anatomical structure on the human body with immense racial variation; however, consistent features for different racial groups have been described. When addressing NTP, rhinoplasty surgeons must be acutely aware of the nasal aesthetic ideals of the racial group the patient belongs to and weigh this against the patient’s goals and anatomy. Patients of races with classically underprojected noses may require some increased projection, but their “ideal” NTP is likely less than a Caucasian patient.

For example, the Middle Eastern nose frequently displays an underprojected and bulbous tip that may “hang” and contribute to the illusion that a prominent arching dorsal hump is much more significant than it really is.^1,2 Other features of these mesorrhine and leptorrhine noses³ include a wide bony vault, short columella, hyperactive depressor septi nasi muscles, weak cephalically oriented lateral crura with a widened genu, thicker fibrofatty tissue in 40%–90% of patients,^4,5 and Fitzpatrick III-V skin.⁶

Nasal tips of African descent are often also underprojected due to the morphology of the underlying nasal framework, including short, flat, and wide nasal bones, and retruded premaxillae and anterior nasal spines.^7–9 Other features of these often platyrrhine noses³ include a low-set deep radix, shorter medial crura, less abundant and cephalically oriented lateral crura, wide alar base width, alar flare, thick skin with abundant fibrofatty tissue, and Fitzpatrick IV-VI skin.⁶

Caucasian nasal aesthetics have been broadly applied across many racial groups, and many previous descriptions of “ideal” NTP have been centered around the Caucasian nose, which is often well projected. In fact, the vast majority of facial anatomy texts are based on Caucasians, a global minority.¹⁰ However, the goal of many racially diverse patients is distinctly not to “Westernize” their nose, but to meet their aesthetic goals while maintaining the integrity of features unique to their particular race.

Several methods of measuring and predicting ideal NTP have been described over the decades. Classically, Goode¹¹ described ideal NTP as a ratio of 0.55–0.60 between tip length (TL) and nasal length, whereas Byrd and Hobar¹² suggested this ideal ratio should be two-thirds (0.67). Other authors also described ideal NTP measurements around this time.^13–15 However, Crumley and Lanser¹⁶ cited shortcomings in several descriptions when proposing 2 proportions for ideal NTP: 1 incorporating upper lip length and another using facial height.¹⁶ More recently, the Rohrich “10-7-5 nasal analysis” includes an easy-to-use ideal NTP measurement, when 50%–60% of the tip lies anterior to a vertical line adjacent to the upper lip.¹⁷

Given the high degree of variability in nasofacial morphology across racial groups,⁶ it is intuitive that a one-size-fits-all method of measuring ideal NTP is neither realistic nor reliable. Previously described methods for ideal NTP measurement have been contrasted in the literature,^16,18 but not between different racial groups. We aimed to compare racial nuances in measuring ideal NTP and hypothesize that methodologies incorporating auxiliary facial dimensions such as upper lip length and facial height may be better suited for racially diverse populations seeking rhinoplasty.

METHODS

With institutional review board acknowledgment and designation as “non-human subject research,” an openly accessible artificial intelligence (AI) image-generator platform (catbird.ai, NewCompute, Inc., New York, NY) was used to generate images of men and women with “aesthetically ideal” noses. The AI models used were latent text-to-image diffusion models, trained with millions of real images to generate high-quality photorealistic images of people given any text input. A commercial license was obtained for image publication. Lateral profile poses were used. Basal views were attempted but were unable to be generated.

The text prompt used to generate these images was as follows: “A Caucasian man with a perfect, aesthetically ideal nose posing in lateral view.” This prompt was then minimally altered to “Middle Eastern,” “African American,” and “woman” to generate images of Caucasian, Middle Eastern, and African American men and women posing in lateral view. Asian and Hispanic noses were not included due to the company taking the image generator offline for maintenance before generating these images.

The generated images were then chosen using the following inclusion criteria: as true of a lateral or profile view as possible, clear ability to visualize the nasal and facial anatomy, no excessive low- or high-angle images, and no anatomic distortion. Ten consecutive images that met inclusion criteria for Caucasian, Middle Eastern, and African American men and women were compiled (60 images total) and were used for the analysis.

Adobe Photoshop 2023 (Adobe, Inc., San Jose, CA) was used to obtain NTP measurements for each racial cohort. Manual measurements of each method were made by the first author (A.L.W.) using pixels as a unit of length, standardizing measured lengths across all images, which did have variation in overall image size. Figure 1 outlines the different ideal NTP methodologies used. Consistent with previous descriptions, nasal length was measured from nasion to the anterior most tip-defining point, nasolabial length was measured from nasion to upper vermilion border, and nasofacial length (NFL) was measured from nasion to mandibular border with the line extending through the alar sulcus. Depending on the NTP methodology being used, “TL,” or tip projection distance, was measured from the alar sulcus, nasolabial line (NLL) (at 90°), nasofacial line (at 90°), or lip tangent line to the anterior most tip-defining point.

Fig. 1.

The 5 different ideal NTP measurement methods used in this study are shown. Nasal length proportion (left) was defined as the proportion of TL (equivalent to tip projection distance) to nasal length and included two identical NTP methods with differing proposed ideals.^11,12 Technically, Byrd and Hobar used “ideal nasal length” in their article; however, the nasion to TL was applied in this study. Nasolabial tip projection (top) was defined as the proportion of TL (from the NLL) to the nasolabial length.¹⁶ Nasofacial tip projection (right) was defined as the proportion of the TL (from the nasofacial line) to the NFL.¹⁶ Lip tangent tip projection (bottom; from Rohrich 10-7-5 nasal analysis) was defined as the proportion of TL lying anterior to a vertical line tangent to the anterior border of the upper lip. 2/3, two-thirds.

One-way analysis of variance with post hoc Tukey honest significant difference test was used to evaluate NTP differences between the gender and racial cohorts. One-sample t tests were then used to compare these measurements to previously described aesthetic ideals.^11,12,16,17 R statistical software (v. 4.3.1, R Core Team, Vienna, Austria) was used for analysis, and a P value of less than 0.05 was considered statistically significant.

RESULTS

Twenty lateral facial images were analyzed per racial cohort (10 men and 10 women). Mean projection measurements are displayed in Table 1. Male versus female NTP did not differ significantly in any racial cohort using any measurement method (Table 1). No noses were measured as having ideal NTP using the Goode nasal length proportion method (Table 2), and NTP across all racial groups did not differ significantly (P = 0.053 men and P = 0.52 women) (Table 3).

Table 1.

Mean NTP and Male Versus Female NTP Comparisons

Measurement Method	Racial Group	Mean Male NTP, % (SD)	Mean Female NTP, % (SD)	P
Nasal length proportion
Goode/Byrd and Hobar TL:nasal length	Caucasian	68.3 (3.1)	66.8 (3.3)	0.3
	African American	66.4 (2.8)	64.3 (5.7)	0.31
	Middle Eastern	69.9 (3.3)	66.2 (5.5)	0.09
Nasolabial tip projection
Crumley and Lanser method I TL:nasolabial length	Caucasian	28.8 (3.7)	28.5 (1.8)	0.85
	African American	19.9 (1.8)	21.0 (3.8)	0.42
	Middle Eastern	29.3 (1.4)	28.0 (2.0)	0.11
Nasofacial tip projection
Crumley and Lanser method II TL:NFL	Caucasian	25.9 (1.3)	26.3 (1.9)	0.56
	African American	22.2 (1.2)	23.0 (2.0)	0.35
	Middle Eastern	25.6 (1.5)	26.7 (1.2)	0.1
Lip tangent tip projection
Rohrich et al Amount of tip anterior to UL tangent line	Caucasian	48.3 (8.8)	48.9 (9.1)	0.89
	African American	12.0 (10.4)	10.2 (6.8)	0.65
	Middle Eastern	44.1 (8.3)	42.4 (6.8)	0.63

TL, tip projection length; UL, upper lip.

Table 2.

Comparing NTP to Previously Described Ideals

Measurement Method	Sex	Racial Group	Significance
Nasal length proportion
Goode TL:nasal length	P versus 0.60
	Male	Caucasian	<0.0001*
		African American	<0.0001*
		Middle Eastern	<0.0001*
	Female	Caucasian	0.0001*
		African American	0.04*
		Middle Eastern	0.006*
	All	Caucasian	<0.0001*
		African American	<0.0001*
		Middle Eastern	<0.0001*
Byrd and Hobar TL:nasal length			P versus 0.67
	Male	Caucasian	0.14
		African American	0.76
		Middle Eastern	0.01*
	Female	Caucasian	0.94
		African American	0.22
		Middle Eastern	0.77
	All	Caucasian	0.25
		African American	0.20
		Middle Eastern	0.23
Nasolabial tip projection
Crumley and Lanser method I TL:nasolabial length	P versus 0.2833
	Male	Caucasian	0.73
		African American	<0.0001*
		Middle Eastern	0.06
	Female	Caucasian	0.78
		African American	<0.001*
		Middle Eastern	0.58
	All	Caucasian	0.65
		African American	<0.0001*
		Middle Eastern	0.48
Nasofacial tip projection
Crumley and Lanser method II TL:NFL	P versus 0.2364
	Male	Caucasian	<0.001*
		African American	0.005*
		Middle Eastern	0.002*
	Female	Caucasian	0.002*
		African American	0.31
		Middle Eastern	<0.0001*
	All	Caucasian	<0.0001*
		African American	0.01*
		Middle Eastern	<0.0001*
Lip tangent tip projection
Rohrich et al Amount of tip anterior to UL tangent line	In 50%–60% (95% CI)
	Male	Caucasian	Yes (0.42–0.55)
		African American	No (0.05–0.19)*
		Middle Eastern	Yes (0.38–0.50)
	Female	Caucasian	Yes (0.42–0.55)
		African American	No (0.05–0.15)*
		Middle Eastern	No (0.38–0.47)*
	All	Caucasian	Yes (0.45–0.53)
		African American	No (0.07–0.15)*
		Middle Eastern	No (0.40–0.47)*

*Statistical significance (P < 0.05). Of note, Statistical significance delineates those findings that differed from the previously described ideals. The measurements obtained using the Goode method were compared with the upper limit of the ideal range of 55%–60% and all groups had significantly more projection.

CI, confidence interval; TL, tip projection length; UL, upper lip.

Table 3.

Comparing NTP Across Racial Groups

Measurement Method	Sex	Mean Caucasian NTP, % (SD)	Mean African American NTP, % (SD)	Mean Middle Eastern NTP, % (SD)	P	Significant Pairs
Nasal length proportion
Goode/Byrd and Hobar TL:nasal length	Male	68.3 (3.1)	66.4 (2.8)	69.9 (3.3)	0.05	None
	Female	66.8 (3.3)	64.3 (5.7)	66.2 (5.5)	0.52	None
Nasolabial tip projection
Crumley and Lanser method I TL:nasolabial length	Male	28.8 (3.7)	19.9 (1.8)	29.3 (1.4)	<0.0001*	Caucasian versus African American, Middle Eastern versus African American
	Female	28.5 (1.8)	21.0 (3.8)	28.0 (2.0)	<0.0001*	Caucasian versus African American, Middle Eastern versus African American
Nasofacial tip projection
Crumley and Lanser method II TL:NFL	Male	25.9 (1.3)	22.2 (1.2)	25.6 (1.5)	<0.0001*	Caucasian versus African American, Middle Eastern versus African American
	Female	26.3 (1.9)	23.0 (2.0)	26.7 (1.2)	<0.0001*	Caucasian versus African American, Middle Eastern versus African American
Lip tangent tip projection
Rohrich et al Amount of tip anterior to UL tangent line	Male	48.3 (8.8)	12.0 (10.4)	44.1 (8.3)	<0.0001*	Caucasian versus African American, Middle Eastern versus African American
	Female	48.9 (9.1)	10.2 (6.8)	42.4 (6.8)	<0.0001*	Caucasian versus African American, Middle Eastern versus African American

*Statistical significance (P < 0.05). Of note, Caucasian and Middle Eastern noses did not differ significantly using any measurement method, whereas Caucasian and Middle Eastern noses were significantly more projected than African American noses using nasolabial tip projection, nasofacial tip projection, and lip tangent tip projection.

NFL, nasofacial length; TL, tip projection length; UL, upper lip.

All male and female racial groups—except for male Middle Eastern noses (P = 0.01)—were measured as having ideal NTP using the Byrd nasal length proportion method (Fig. 2; Table 2). No racial group had significantly different NTP using this method (Table 3).

Fig. 2.

Middle Eastern male lateral images showing the NTP measurement methods used in this study. A, NL proportion (64.1% in this model compared with ideals of 55%–60% and 67%), the TL can be visually “swung up” and compared with NL. B, Nasolabial tip projection (0.274 in this model compared with ideal of 0.2833) where the TL anterior to the NLL is compared with the NLL. C, Nasofacial tip projection (0.259 in this model compared with ideal of 0.2364) where the TL is compared with the NFL. D, Lip tangent tip projection (28% in this model compared with ideals of 50%–60%) where the anterior length of the tip is compared with total TL. NL indicates nasal length.

Caucasian and Middle Eastern, but not African American noses (P < 0.0001) were measured, as having ideal NTP using the nasolabial tip projection method (Fig. 3; Table 2). Both male and female Caucasian (P < 0.0001) and Middle Eastern (P < 0.0001) noses had significantly more nasolabial tip projection than African American noses (Table 3).

Fig. 3.

Middle Eastern female lateral images showing the NTP measurement methods used in this study. A, Nasal length proportion (66.7% in this model compared with ideals of 55%–60% and 67%). B, Nasolabial tip projection (0.286 in this model compared with ideal of 0.2833) where the TL anterior to the NLL is compared with the NLL. C, Nasofacial tip projection (0.277 in this model compared with ideal of 0.2364) where the TL is compared with the NFL. D, Lip tangent tip projection (50.5% in this model compared with ideal of 50%–60%) where the anterior length of the tip is compared with total TL.

African American female noses were measured as having ideal NTP using the nasofacial tip projection method, and African American male noses trended relatively toward ideal (Table 2), whereas Caucasian (P < 0.0001) and Middle Eastern (P < 0.0001) noses had significantly more nasofacial tip projection (Fig. 4; Table 3).

Fig. 4.

African American female lateral images showing the NTP measurement methods used in this study. A, Nasal length proportion (63% in this model compared with ideals of 55%–60% and 67%). B, Nasolabial tip projection (0.215 in this model compared with ideal of 0.2833) where the TL anterior to the NLL is compared with the NLL. C, Nasofacial tip projection (0.22 in this model compared with ideal of 0.2364) where the TL is compared with the NFL. D, Lip tangent tip projection (4.8% in this model compared with ideal of 50%–60%) where the anterior length of the tip is compared with total TL.

Using the lip tangent method, only Caucasian noses fell consistently into the ideal range for NTP (Table 2), and Caucasian (P < 0.0001) and Middle Eastern (P < 0.0001) noses were significantly more projected than African American noses (Fig. 5). Table 3 summarizes the differences between the NTP across racial groups.

Fig. 5.

African American male lateral images showing the NTP measurement methods used in this study. A, Nasal length proportion (66.7% in this model compared with ideals of 55%–60% and 67%). B, Nasolabial tip projection (0.217 in this model compared with ideal of 0.2833) where the TL anterior to the NLL is compared with the NLL. C, Nasofacial tip projection (0.216 in this model compared with ideal of 0.2364) where the TL is compared with the NFL. D, Lip tangent tip projection (22.7% in this model compared with ideal of 50%–60%) where the anterior length of the tip is compared with total TL.

DISCUSSION

The proper evaluation of a patient’s NTP is a key element of nasofacial analysis. Many iterations of what constitutes ideal NTP have been previously described, and there is likely substantial variability in the NTP measurement methodologies rhinoplasty surgeons use in their day-to-day practice. In modern American culture, rhinoplasty surgeons must be able to treat racially diverse populations and achieve excellent, racially congruous results. Understanding the racial nuances of ideal NTP is critical. This study clearly shows that different NTP measurement methods yield very different results in racially diverse populations and highlights the difficulty in applying previously described NTP ideals to racial groups with underprojected noses, particularly African Americans.

The NTP measurement methods used in this study were chosen because they have been broadly applied in the rhinoplasty literature over the past several decades. The Goode method has been used to evaluate outcomes after orthognathic surgery¹⁹ and cleft rhinoplasty,²⁰ comparing columellar strut versus caudal septal extension grafts,²¹ and comparing flexible versus classic tongue-in-groove techniques.²² The Byrd NTP ideal has been used in preoperative rhinoplasty planning,²³ evaluating the effectiveness of projection suture techniques,²⁴ and in the initial descriptions of caudal septal extension grafts.²⁵ The Crumley NTP ideal has been shown to significantly impact perceived facial aesthetics relative to other ideals.^26,27 The Rohrich 10-7-5 method of nasal analysis is more contemporary, but likely the most straightforward and comprehensive nasal analysis described to date.

Nasal length proportion yielded strikingly different NTP results in this study. No analyzed group fell into the Goode ideal range of TL being 55%–60% of nasal length—all were overprojected.¹¹ Yet, nearly all groups had ideal NTP when this proportion was increased to two-thirds (67%).¹² This not only suggests that the Goode method consistently results in underprojected tips (even the African American groups were overprojected using this method), but also that the Byrd method may be an easy and generalizable way to evaluate ideal NTP in diverse populations—although this requires further validation. Despite this, the major shortcoming of nasal length proportion in evaluating NTP is its disregard for regional midface and lower face anatomy.¹⁶

Incorporating auxiliary facial anatomy into NTP analysis¹⁶ led to Caucasian, Middle Eastern, and African American noses measuring ideal NTP in this study. Middle Eastern facial morphology displays greater nasal length with more acute nasolabial angles,²⁸ greater facial convexity angle, and greater retrogenia when compared with North American Caucasians.^29–31 (See figure, Supplemental Digital Content 1, which displays the Caucasian male lateral images showing the NTP measurement methods used in this study. Left, nasal length proportion [69.1% in this model compared with ideals of 55%–60% and 67%]. Second from left, nasolabial tip projection [0.311 in this model compared with ideal of 0.2833] where the TL anterior to the NLL is compared with the NLL. Second from right, nasofacial tip projection [0.269 in this model compared with ideal of 0.2364] where the tip TL is compared with the NFL. Right, lip tangent tip projection [58.8% in this model compared with ideal of 50%–60%] where the anterior length of the tip is compared with total TL, http://links.lww.com/PRSGO/D637.) (See figure, Supplemental Digital Content 2, which displays the Caucasian female lateral images showing the NTP measurement methods used in this study. Left, nasal length proportion [67.9% in this model compared with ideals of 55%–60% and 67%]. Second from left, nasolabial tip projection [0.312 in this model compared with ideal of 0.2833] where the tip TL anterior to the NLL is compared with the NLL. Second from right, nasofacial tip projection [0.247 in this model compared with ideal of 0.2364] where the TL is compared with the NFL. Right, lip tangent tip projection [56.4% in this model compared with ideal of 50%–60%] where the anterior length of the tip is compared with total TL, http://links.lww.com/PRSGO/D638.)

However, Caucasian and Middle Eastern faces both derive from Caucasoid facial structure¹⁰ and share similar midfacial to lower third height ratios and upper lip protrusion.³⁰ It is fitting that Caucasian and Middle Eastern noses did not differ significantly in NTP using any measurement method (Table 3) and were both found to have ideal nasolabial tip projection, but overprojected nasofacial tip projection. This suggests that Caucasian and Middle Eastern midface morphologies better accommodate their respective NTP; and that the tip of these groups may seem overprojected when the lower face is incorporated into analysis.

African American facial morphology displays shorter noses, shorter middle and lower facial thirds, and fuller lips with greater lip and midface protrusion when compared with Caucasian faces.^28,32,33 It is appropriate that African American noses were found to have underprojected nasolabial tip projection, but more ideal nasofacial tip projection. This suggests that shorter African American midfaces (with fuller, more anterior lips) push the nasion to the upper vermilion line more anteriorly—significantly decreasing the length of the tip from this line and, therefore, NTP (Figs. 4, 5). Comparing NTP to the shorter height of the lower face in African Americans leads to larger nasofacial tip projection measurements, increasing NTP toward more ideal ratios. In summary, Middle Eastern (and Caucasian) NTP may be optimized when compared with midfacial morphology and African American NTP may be optimized when compared with lower facial morphology.

This study suggests that the lip tangent method¹⁷ is an efficacious and straightforward method to evaluate ideal NTP in Caucasian patients; however, it leads to consistently underprojected NTP measurements in Middle Eastern and African American noses. Middle Eastern noses consistently display drooping or plunging tips,^1,34 which means that, despite a longer nasal length,²⁸ the vector at the tip region turns inferiorly (Figs. 2, 3). Thus, this loss of anterior length leads to tip underprojection relative to the lip tangent line. This method also does not apply well to African American noses because of their relatively fuller and more anteriorly positioned lips (Fig. 5).^32,33

A unique aspect of this study is the use of AI to generate images of “ideal” and racially diverse noses. The authors do not intend to claim that the generated noses are truly ideal in every way—or that they represent all individuals from these variegated racial groups. The authors caution against overgeneralization of nasal phenotypes for racial groups based on these images. For example, African noses may be platyrrhine, mesorrhine, or leptorrhine (as well as thick- or thin-skinned), and rhinoplasty surgeons must tailor surgical plans to a given patient’s morphology and their goals. However, the findings in this study are important because they highlight real measured differences in ideal NTP between Caucasian, Middle Eastern, and African American noses with phenotypes representative of clinically significant proportions of these racial groups.

Furthermore, all noses generated were photorealistic, racially appropriate, and aesthetically pleasing. The noses for different racial groups differed morphologically, as summarized in Table 3, which supports the notion that AI can accurately generate racially diverse noses. Nearly all racial groups were measured as having ideal NTP using the Byrd nasal length proportion method, and several others measured ideal using various studied methods. However, Caucasian noses were more projected than Middle Eastern noses using the lip tangent method, and both were significantly more projected than African American noses using nasolabial tip projection, nasofacial tip projection, and the lip tangent method. These results provide the rhinoplasty surgeon with a broader armamentarium of ideal NTP measurements that can be applied to patients of these racial groups to compare the patient’s preoperative anatomy, a “ballpark” ideal projection point relative to their facial features appropriate for them, and their ultimate postoperative goals.

Given that the AI models are trained using millions of images of real humans, the use of this technology is a fascinating insight into the current American cultural facial aesthetic ideals of these different racial groups. AI image generation has been previously used to shed light on racial differences in aesthetic breast ideals,³⁵ and AI is being increasingly used in preoperative planning within plastic surgery, including rhinoplasty.³⁶ It is also important to realize that the widespread use of AI-generated “deepfake” images on social media platforms will inevitably continue to influence what prospective patients may deem aesthetically ideal and what they seek during plastic surgery consultations.³⁷

The potential future applications of AI in studying aesthetic ideals are vast, given the ease at which large quantities of models can be created instantly.³⁸ In terms of nasal analysis in racially diverse populations, this study highlights several new concepts with its application of AI. Rhinoplasty surgeons now have at their disposal a tool to help develop a thorough understanding of the racial nuances of nasal analysis. Trainees and surgeons who infrequently operate on patients from a certain racial demographic can easily generate hundreds of images of noses of that race for analysis, which may help them prepare for surgery. Surgeons can use AI to generate images of racially congruous noses during patient consultations to better understand and inspire patient goals and help explain certain inherent anatomic limitations without the ethical concerns of sharing photographs of patients’ faces. It is also possible for surgeons to train AI models with patient photographs that can generate images of potential postoperative outcomes for specific patients. Newer AI technologies such as “infill” or “generative fill” can selectively alter various nasal anatomy of a patient photograph to help patients visualize desired postoperative outcomes, without the need for manual editing on expensive 3-dimensional imaging systems.³⁹ However, this study also raises ethical concerns given the potential for false advertising by surgeons who use AI images to display unrealistic surgical outcomes or appear more experienced in rhinoplasty.

The use of AI in this study is also the source of its main limitations. As previously mentioned, these images do not perfectly represent these racial groups—something likely impossible to do even with real photographs. The exact details as to how the used AI models were trained is not known, and therefore, it was not possible to assess for any biases present in their respective datasets. It was also impossible to completely control for the individual’s positioning, image angle, facial hair, and lighting, which likely influenced the results to a minor degree. Furthermore, the AI models were unable to generate basal views of the nose, which can provide vital information regarding NTP. This study, which used images of youthful faces, also fails to address the changes nasofacial structures undergo with aging, such as lip lengthening and convexity, which would impact these results. However, AI can generate faces of various ages when prompted to do so, which is an interesting future direction of study.

Although the NTP measurements were made by the same individual (A.L.W.) using the same methods, subjectivity exists in determining various nasofacial landmarks—particularly the nasion and anterior most tip-defining point. The number of images analyzed per racial group is relatively low; however, the nasofacial analyses undertaken were tedious and limited the number of images that could be reasonably analyzed by hand. This study also failed to include other important racial groups such as Hispanic or Asian; however, analysis of these groups and others is a valuable future direction of study that is forthcoming.

CONCLUSIONS

Race is a defining element of modern facial and nasal aesthetics, and there is a high degree of racial variation in NTP. Many different methods of measuring ideal NTP have been described and used in the literature. This study shows that different methodologies lead to very different results in racially diverse groups. This study also suggests that the ideal nasal length to tip projection length ratio of two-thirds (67%) has generalizability across racial groups; that the lip tangent is an excellent method for evaluating tip projection in Caucasians; and that assessment of ideal NTP in Middle Eastern and African American groups—with less projected tips—should use regional midface and lower face anatomy, respectively, to account for differences in overall facial morphology.

DISCLOSURE

The authors have no financial interest to declare in relation to the content of this article.