Example of a preoperative profile view (A), computer-imaged view (B), and actual 6-month postoperative photograph (C).
Example of a preoperative anteroposterior view (A), computer-imaged view (B), and actual 6-month postoperative photograph (C).
Mean accuracy scores (based on a 5-point Likert scale) assigned by the expert (surgeon) panel for the various nasal parameters.
Mean accuracy scores (based on a 5-point Likert scale) assigned by the nonsurgeon panel for the various nasal parameters.
Mehta U, Mazhar K, Frankel AS. Accuracy of Preoperative Computer Imaging in Rhinoplasty. Arch Facial Plast Surg. 2010;12(6):394-398. doi:10.1001/archfacial.2010.96
Copyright 2010 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2010
To quantitatively measure the accuracy of preoperative computer imaging (PCI) as a reflection of postoperative rhinoplasty results.
Thirty-eight patients underwent primary and revision rhinoplasty. Six-month postoperative photographs and “morphed” PCI images were graded on a 5-point Likert scale by 2 panels of judges, one composed of surgeons and the other of nonsurgeons. Twelve parameters were assessed. Results were stratified based on primary vs revision rhinoplasty and degree of difficulty of the rhinoplasties. Patient surveys were conducted to determine correlation between satisfaction and PCI accuracy.
Mean overall accuracy was 2.98 by the expert panel, indicating moderate PCI accuracy. Supratip height was the lowest-rated parameter, while measurements of the upper third were most accurate. The accuracy in primary rhinoplasty was comparable to that found in revision surgery, although tougher cases were rated lower in overall accuracy and projection score. Satisfied patients had significantly higher PCI accuracy scores, and most patients found the PCI extremely useful.
Preoperative computer imaging is a useful exercise, valued by patients during the rhinoplasty consultation. Accuracy is moderate in both primary and revision cases, although supratip edema is a limiting factor at the 6-month mark.
Preoperative computer imaging (PCI) is now widely used within the specialty of facial plastic surgery, and several published articles have attempted to determine its usefulness with respect to rhinoplasty.1- 4 It is generally accepted that PCI provides for improved communication between surgeon and patient; allows for reconciling differences between the patient's desires and the surgeon's aesthetic; aids in preoperative analysis and planning; provides a mechanism for critical self-assessment; and helps to weed out patients who have unrealistic expectations. After using PCI for several years, we believe that these benefits, though difficult to quantify, are real. What remains are 2 important questions: (1) can PCI be a reliable predictor of postoperative results in both primary and revision rhinoplasty? and (2) does the degree of patient satisfaction depend significantly on the accuracy of the PCI?
A few studies have attempted to answer these questions. In 2002, Sharp et al1 surveyed rhinoplasty patients and found that the vast majority felt that PCI enhanced the surgical process. In an overall assessment, 44% of patients stated that the actual result was more aesthetically pleasing than the PCI images, while 32% felt that the outcome was less attractive than the PCI views.
Mühlbauer and Holm2 published a report in 2005 involving predominantly primary rhinoplasty cases (115 of 120) in which both authors and patients completed a survey to assess the overall accuracy of PCI for each patient. They found that the PCI images and postoperative photographs were rated identical in 11% of cases by the authors and in 29% of cases by the patients. In most cases, the 2 groups rated the preoperative computer images and postoperative photographs as “similar.”
In 2007, Agarwal et al3 reported their findings in 25 consecutive cases rated by 12 double-blinded surgeons, providing a global assessment of accuracy. In this series, 18% of patients felt that the PCI images fully predicted the actual results. Fifty percent of respondents felt that the PCI images predicted the results but not with complete accuracy.
Adelson et al4 recently published an article quantifying the accuracy of imaging by measuring 5 specific parameters of the nasal profile: nasofrontal angle, nasolabial angle, projection, columellar-tip angle, and columellar-infratip ratio. Measurements were made on the PCI images and on the postoperative photographs, and no statistically significant difference was found in any of the assessed parameters, with the exception of the columellar-tip angle.
Based on the experience of those who have tackled this issue previously, this study represents our attempt to grade the accuracy of PCI as a predictor of actual 6-month postoperative results using unbiased panels of surgeons and nonsurgeons to assess 12 different parameters for each case. Both primary and revision rhinoplasty cases were included, and we aimed to determine if patient satisfaction correlates with the accuracy of PCI.
At the initial consultation with all patients requesting rhinoplasty between March and October 2008, the senior surgeon (A.S.F), not a nurse or other staff member, took standardized photographs and archived them into United Imaging software (United Imaging Inc, Winston-Salem, North Carolina). Digital alteration of these images (anteroposterior [AP] and left and right profile views) was then performed by the surgeon, with the input of the patient (Figure 1 and Figure 2). The images were displayed in the operating room during the case and used for guidance. At 6-month follow-up visits, photographs were again taken and archived. Only those patients who returned for this 6-month follow-up visit were included in the evaluation, although any long-term follow-up visit greater than 5 months from the time of surgery was considered clinically acceptable.
The preoperative views, CPI views, and 6-month postoperative views were assembled and sent on compact discs to 2 panels of judges, one consisting of 3 surgeons and the other of 3 nonsurgeons. The panelists chosen had no personal or financial relationship with us and were assured that their responses would remain anonymous to maintain their objectivity.
Surveys were completed by the 2 panels to grade the accuracy of the PCI images along 12 parameters on a standard 5- point Likert scale (1 indicating not at all accurate, and 5 indicating identical). The parameters assessed were (1) overall accuracy, (2) projection, (3) length, (4) rotation, (5) supratip height, (6) profile height, (7) columellar show, (8) width of the tip, (9) middle vault width, (10) width of the nasal bones, (11) straightness of the nasal pyramid, and (12) whether the PCI image or postoperative result was more attractive. This 12th parameter was scored differently, with 1 meaning that the PCI view was much more attractive a 5 meaning that the postoperative photograph was the much more attractive view. The expert (surgeon) panel was given additional pertinent patient historical and physical examination information (eg, previous surgery, status of septal and/or auricular cartilage) and asked 2 additional questions: (1) the degree of difficulty of the rhinoplasty or revision rhinoplasty and (2) whether the AP or profile PCI view was more accurate when compared with the postoperative views.
Patients were subsequently given surveys that included questions regarding (1) overall happiness with the outcome (subdivided into the ability to breathe and the appearance of the nose); (2) how closely the actual result resembled the PCI view; (3) how the PCI images and actual results differed (open-ended question); (4) whether the actual result was better or worse than the PCI view; (5) the utility of the imaging in expression of their aesthetic, development of trust in the surgeon, and understanding the rhinoplasty process; and (6) whether they would recommend the PCI process to a friend undergoing rhinoplasty.
Descriptive statistics were used to describe the distribution of the data. Separate analyses were performed for ratings given by the expert panel and those given by the nonsurgeon panel. Cohen κ values were calculated to evaluate the degree of agreement among the experts and nonexperts independently. Data were then stratified on the basis of primary vs revision rhinoplasty and degree of difficulty of the rhinoplasty, as graded by the expert panel. Cases scored 1 to 3 were classified as one group, and cases scored 4 or higher were classified as another. The t test was used to evaluate for differences in these subgroup analyses. All statistical analysis was conducted using SAS software, version 9.1 (SAS Institute Inc, Cary, North Carolina), and P < .05 was considered significant.
Thirty-eight patients met the inclusion criteria during the enrollment period and underwent surgery between February and December 2008. Twenty-two patients underwent primary rhinoplasty (58%) while 16 underwent revision. Patient ages ranged from 16 to 57 years, and 5 patients were male (13%). The median time to follow-up visit was 8.3 months, with a range of 5.1 to 12.5 months. To date, 3 of the 38 patients required a second surgical intervention by a senior surgeon (A.S.F.) and in all 3 cases, photographs were taken 6 months after the most recent surgery.
The overall accuracy of the preoperative imaging for the entire group was given a mean (SD) score of 3.10 (0.91) by the expert panel, equating to moderately accurate (Figure 3). The lowest-rated parameter was supratip height (2.82 [1.06]), while the highest scores were given to the width of the nasal bones (3.72 [0.96]). The PCI views and actual results were found to be equally attractive on the whole (2.98 [0.99]). Cohen κ values ranged from −0.26 to 0.55.
The experts found that the AP image was more accurate than the profile in 76% of cases (n = 29), while in only 12% of the cases were the profile views rated more accurate (n = 5). The AP and profile views were rated equally accurate in the remaining 11% of cases (n = 4). The mean (SD) degree of difficulty was 3.18 (0.84). Interestingly, a score of 1 or “easy” was given only twice by the expert panel, and in each case by only a single panelist.
The nonsurgeon panel scored the overall accuracy a mean (SD) of 3.55 (0.90) (Figure 4). The lowest scores were again given to the height of the supratip (3.30 [1.06]), while the highest-rated parameter was straightness of the nasal pyramid (3.86 [0.86]).
The expert panel was considerably tougher grading than the nonsurgeons. This was true of the overall assessment of accuracy as well as all the parameters related to the profile views: projection, length, rotation, supratip, profile height, and columellar show (P < .05 for all comparisons). However, there was no statistically significant difference in the ratings related to the width of the various segments of the nose or whether the PCI view or actual result was more attractive.
The cases were then subdivided into primary and revision surgery. Overall, the ratings were extremely similar from both surgeons and nonsurgeons. The only parameter that reached statistical significance for the difference was profile height in the nonsurgeon ratings (primary vs revision, 3.41 vs 3.82) (P = .04).
When the cases were divided into easy (scores 1-3) and difficult (scores 4 or 5) groups based on the degree of surgical difficulty, the differences in rating for a few parameters were statistically significant. These included, from the expert panel, the mean global assessment of accuracy (3.20 vs 2.71) and projection (3.27 vs 2.67). Three parameters were statistically different in the nonsurgeons' ratings: overall accuracy (3.66 vs 3.17), projection (3.77 vs 3.17), and rotation (3.51 vs 3.00).
Patient surveys were completed by 11 of 38 participants in this study. Overall happiness with the outcome was rated 4 or 5 by 81% of respondents (n = 9). When asked to rate the accuracy of their own PCI images, the mean (SD) score was 3.4 (0.67). Patients found the PCI process to be generally useful, with mean scores of 4.40, 4.60, and 4.30 for understanding the surgeon's aesthetic, developing trust, and understanding the surgery, respectively. Nine of eleven patients reported that they would highly recommend the imaging process to a friend or family member undergoing rhinoplasty.
The patients were then subdivided into happy (scores 4 and 5) and unhappy (scores 1-3) to determine the correlation, if any, between happiness and imaging accuracy. The mean expert panel scores for all categories were significantly higher for the happy group (3.30 vs 2.76) (P < .001). When specific parameters were examined, rotation (3.22 vs 2.17) (P = .01) and tip width (3.40 vs 2.33) (P = .002) scored higher in the happy group. In the nonsurgeon panel, the mean scores for all categories combined were also higher in the happy group (3.65 vs 2.92) (P < .001). Several parameters, including overall accuracy, length, rotation, supratip height, and tip width were rated more highly in the happy group (P < .05 for all).
Computer imaging software that enables the user to alter images is now readily available to the public and surgeons alike. In the senior author's practice (A.S.F.), such software has been used, primarily in rhinoplasty cases, to elicit patients' aesthetic goals and to communicate various principles, limitations, and pitfalls from surgeon to patient. In addition to being used as a vehicle for communication with the patient, imaging software might also have improved preoperative planning and outcomes. To be a useful tool, PCI must first be determined to yield 2-dimensional results that are reproducible with surgical techniques in both primary and revision rhinoplasty.
To our knowledge, this is the first study to tackle the issue using unbiased panels of experts and lay people while assessing each parameter of the nose individually and collectively. The lay panel was included in this study because their impressions of the rhinoplasty PCI imaging and postoperative photographs may more closely represent patient impressions, without the bias inherent in questioning patients regarding the accuracy of their own images as a primary measure. The inclusion of both primary and revision cases is also important because it allowed us to demonstrate that imaging can be reasonably accurate in even the most challenging cases. There was some variability in the scores assigned by the expert and nonsurgeon panels for each parameter, although most mean scores fell in the 3 to 4 range, indicating moderate accuracy.
It is important to recognize that for the individual surgeon, the accuracy of the PCI has as much to do with surgical skill as it has to do with imaging skill. In any practice, there is an evolution in one's ability to efficiently and adeptly create a realistic surgical target while taking into account the wishes of the patient. It is of utmost importance that the surgeon be the one performing the imaging in a responsible manner to generate realistic and surgically achievable images. The imaging must not be used as a marketing gimmick to convince the patient into signing up for surgery. It should also be made abundantly clear that the images are not a guarantee of results but rather a goal toward which the surgeon will strive.
Supratip fullness was a persistent limitation of this study, as we had hypothesized at the outset, owing to the timing of the follow-up photographs. In our experience and the experience of numerous rhinoplasty surgeons, edema of the tip and infratip take the longest to resolve owing to its dependent position and lymphatic distribution.4,5 This edema tends to diminish over the initial 12 to 18 months following rhinoplasty but is often still evident at the 6-month mark. The accuracy of imaging should ideally be assessed at 1 to 2 years after surgery, which we plan to address in a follow-up to this study. Not unexpectedly, the surgeons were, in the aggregate, tougher graders than the nonsurgeons, although we find it worth mentioning that the supratip was found to be the lowest-rated parameter by both groups.
A quite unexpected finding was that the expert panel felt that the AP PCI view was more accurate than the profile PCI view in the vast majority of cases. In our experience, the AP view is the most challenging to image effectively because the changes are made without the benefit of the homogeneous blue background. Any morphing may inadvertently affect the appearance of the patient's lips, cheeks, and eyes. Perhaps owing to the lighting used in the preoperative photographs and their 2-dimensional nature, small differences might be difficult to distinguish on the front view.
The stratification into primary and revision rhinoplasties did not yield particularly interesting findings, although the degree of difficulty stratification was quite intriguing. The PCI views in difficult cases were rated globally as less accurate than those in the easy cases by both panels. Both panels also agreed that projection was the most highly affected parameter. This was somewhat surprising to us because we had anticipated that length or rotation would be the least accurate parameter in the more challenging cases.
Although there may be several variables by which we measure a successful rhinoplasty, none is more important than a satisfied patient. Owing to the subjective nature of aesthetics, it is often difficult for the surgeon to understand precisely what a particular patient has envisioned for his or her own nose. Computer imaging software designed for the purpose of altering digital images provides an opportunity for a patient to detail desired changes, which involves them in the decision-making process.
However, does PCI lead to improved surgical outcomes (ie, happier patients)? One would certainly think so, but perhaps this is true only if the imaging is used effectively. The subjective nature of aesthetics, along with the multiple variables that help to color an individual's perception of their surgical result, make it difficult at best to answer this basic question.
We attempted to address this through the use of patient surveys, though our participation rate was admittedly somewhat low in this initial study. We did, however, find that patients were generally quite satisfied with their surgical outcomes, with scores of 4 or 5 in over 80% of respondents (n =9 ), even if their ratings of the accuracy of the imaging were slightly lower.
Patients found the PCI process to be extremely useful in several respects and stated that they would highly recommend the process to anyone undergoing the surgery. A more meaningful question, however, might be whether simply applying this technology is helpful in producing improved outcomes (perceived physician-patient relationship, communication, and confidence) regardless of the actual accuracy of the process. Simply randomizing patients into 2 groups, one with PCI as a component of the consultation, and one without, may help to answer this question. We had considered this when designing this study but ultimately came to believe that imaging has become such a critical process that we could not exclude half the patients from participating.
Importantly, we did find that the PCI accuracy for patients who rated their overall happiness as lower (1-3) also received significantly lower mean scores across the board by both panels. This would lead one to believe that some correlation exists between accuracy of imaging and patient satisfaction, although there is certainly a need to explore this further in a study with a larger sample size.
In summary, imaging is a critical component of our rhinoplasty consultation for a multitude of reasons. It is a reasonably accurate process, with supratip edema being the primary limiting factor 6 months postoperatively. Projection seems to be the most challenging parameter to image accurately in the more difficult cases. Finally, there seems to be a correlation between accuracy of imaging and the patient's overall satisfaction level.
Correspondence: Andrew S. Frankel, MD, The Lasky Clinic, 201 S Lasky Dr 4937, Beverly Hills, CA 90212 (firstname.lastname@example.org).
Accepted for Publication: March 8, 2010.
Author Contributions: Drs Mehta and Frankel had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Mehta, Mazhar, and Frankel. Acquisition of data: Mehta and Frankel. Analysis and interpretation of data: Mehta and Mazhar. Drafting of the manuscript: Mehta, Mazhar, and Frankel. Critical revision of the manuscript for important intellectual content: Mehta and Frankel. Statistical analysis: Mehta and Mazhar. Administrative, technical, and material support: Mehta and Frankel. Study supervision: Frankel.
Financial Disclosure: None reported.