PAPER

PHYSICAL ANTHROPOLOGY

Victoria A. Smith,1 M.N.S.; Angi M. Christensen,2 Ph.D.; and Sarah W. Myers,3 B.S.

The Reliability of Visually Comparing SmallFrontal Sinuses*

ABSTRACT: Several studies have investigated frontal sinus comparison for personal identification. One study addressed the statistical reliabilityof correct identification using automated digital methods and resulted in a 96% accuracy rate. Missed matches with the digital methods generallyinvolved small, less featured sinuses. This study investigates the hypothesis that human examiners may be able to more accurately identify correctmatches than digital methods, even when the comparisons involve small frontal sinuses. Participants were provided two sets of 28 radiographs andwere instructed to identify matching radiographs and list the radiographs that did not have a corresponding match. Overall, error rates were low, withcorrect associations identified at a rate of 0.983. No incorrect associations (‘‘false positives’’) were made. Correct association rates were highestamong participants ‘‘experienced’’ examining radiographs. Results support previous assertions that frontal sinus radiographs are a reliable means ofpersonal identification even when the frontal sinuses are small.

KEYWORDS: forensic science, forensic anthropology, frontal sinus, personal identification

Numerous studies, dating back to the early 1900s, have investi-gated frontal sinus uniqueness and the use of radiographic compari-son for personal identification (1–35). Many of these studies,however, involved small sample sizes in terms of both the numberof radiographs examined and the number of participants involved.One study investigated the statistical reliability of correct identifica-tion using Elliptic Fourier Analysis and Euclidean distance modelson digitized images which resulted in a 96% accuracy rate (8). Theremaining 4% largely represented the inability of the computerizedmodels to correctly match small, less featured frontal sinuses. Thisstudy investigates the hypothesis that human examiners are able toidentify correct matches more accurately than digital methods, evenwhen the frontal sinuses being compared are small or less featured.This, we believe, is both because of the fact that human examinersare more discriminating than digital methods and because humansare able to take into consideration other features present on theradiograph images. Also of interest is the effect, if any, of examin-ers’ experience level on the ability to identify matches.

Materials and Methods

Radiographs were obtained from the University of Tennesseeand are images of specimens from the William M. Bass DonatedSkeletal Collection taken as part of a previous study (see (8)) inthe following manner:

Radiographs were taken of crania with present, complete, andundamaged frontal regions. They were performed at the Universityof Tennessee Student Health Center using a HoLogic HFQ Series100 kHz High Frequency machine at the following settings:

KVP (peak kilovoltage): 58–50 kVpeak

CM (distance from tube to film): 40 cmMA (current in the X-ray tube): 75 mASEC (exposure time): 65 msec.

Crania were oriented so that the X-ray beam traversed the craniaposterior to anterior, with the frontal bone nearest the film to allowminimal distortion and maximum clarity of the frontal sinus. Thecrania were placed on a foam ⁄ cloth ring with the midsagittal planeperpendicular to the X-ray film using the median palatine suture asa guide. When the cranium was viewed from the side, a line per-pendicular to the film was in the same vertical plane as nasion andthe superior border of the external auditory meatus. Replicate radio-graphs were taken of each cranium using the same methodologybut at a different time so that replicates would not be exact copiesand therefore simulated ‘‘antemortem’’ and ‘‘postmortem’’ images.

For this study, a random sample of 60 pairs of these previouslytaken radiographs was selected from the collection. From these 60pairs, the radiographs with the smallest frontal sinuses and lackingvisible dental restorations were used to make the matching processas challenging as possible.

Participants of varying backgrounds and levels of experiencewere solicited to participate in the study including Federal Bureauof Investigation scientists and attendees of the 2008 annual meetingof the American Academy of Forensic Sciences in Washington,D.C. Participants were provided two sets of 28 radiographs labeledA through BB and 1 through 28, an answer sheet ⁄questionnaire,and a light box. They were advised that a ‘‘match’’ consisted ofone letter or letter combination plus one number and that not allradiographs necessarily had a corresponding match present (therewere actually 26 correctly matching pairs and 4 radiographs with

1Oak Ridge Associated Universities, 724 S. Saint Asaph Street, B-312,Alexandria, VA 22314.

2Federal Bureau of Investigation, Laboratory Division, 2501 InvestigationPkwy, Wuantico, VA 22135.

3Emory University, 201 Dowman Drive, Atlanta, GA 30322.*Presented at the 61st Annual Meeting of the American Academy of

Forensic Sciences, February 16-21, 2009, in Denver, CO.Received 1 July 2009; and in revised form 3 Sept. 2009; accepted 3 Oct.

2009.

J Forensic Sci, November 2010, Vol. 55, No. 6doi: 10.1111/j.1556-4029.2010.01493.x

Available online at: onlinelibrary.wiley.com

2010 American Academy of Forensic SciencesPublished 2010. This article is a U.S. Government work and is in the public domain in the U.S.A. 1413

no match present). Although the frontal sinuses were presumablythe main feature compared, participants were also asked to list anycharacteristics besides the frontal sinuses that they used to deter-mine matches.

Participants were also asked to provide information regardingtheir education and background, and to rate their level of experi-ence in both examining radiographs and performing anthropologicalor skeletal examinations. Experience ratings were defined as ‘‘Noexperience,’’ ‘‘Some experience’’ (I have done this before, but notextensively or routinely), or ‘‘Experienced’’ (I do this as part of mywork or have conducted research in this area).

Results

Sixty-five individuals participated in the study. Backgroundsincluded anthropology, biology, radiology, document examination,geology, photography, chemistry, and odontology and ranged fromstudents to highly experienced professionals. The majority of partic-ipants had little or no experience in examining either radiographsor anthropological specimens. Specifically, for examining radio-graphs, 40 participants reported having ‘‘No experience’’, 17reported ‘‘Some experience’’, and eight reported being ‘‘Experi-enced’’. For anthropology ⁄ osteology, 33 participants reported hav-ing ‘‘No experience’’, 14 reported ‘‘Some experience’’, and 18reported being ‘‘Experienced’’ (Fig. 1).

Overall, error rates were very low. The correct association rateamong all participants was 0.983, with 43 of the 65 participants(66%) correctly identifying all matches. No incorrect associations(i.e., pairing radiographs from different people, or ‘‘false positives’’)were made.

A total of 27 nonassociations (i.e., failure to identify a correctmatch, or ‘‘false negatives’’) occurred among all participants. Cor-rect association rates were generally higher among those with moreexperience (Fig. 2) and were highest for those reporting being‘‘experienced’’ examining radiographs. Participants were given anexperience score of 0 (no experience), 1 (some experience), or 2(experienced) for both their anthropology and radiograph experi-ence as well as for combined experience (up to a score of 4 if theparticipant reported being an expert in both categories). Correla-tions were calculated between the type of reported experience andthe correct association rate. The correlations were calculated foreach type of experience individually as well as combined experi-ence. Only the correlation between radiograph experience and cor-rect associations was significant:

Anthropology experience: r = 0.0817, p = 0.5176Radiograph experience: r = 0.28076, p = 0.0224Combined experience: r = 0.17928, p = 0.153

One particular frontal sinus pair was frequently not associated,accounting for 18 of the 27 total nonassociations (Fig. 3). Reasonsfor this may include the limited number of features of the frontalsinus, differences in cranium position, or radiograph quality.

Other common features participants reported using in the match-ing process included the following: overall skull shape, bonycrests ⁄ landmarks ⁄ processes, nose shape, orbit shape, venous mark-ings, other paranasal sinuses, trabecular patterns, teeth, and cranialsutures.

Conclusions

Results support previous assertions that frontal sinus radiographsare a reliable means of personal identification. No incorrect associa-tions (‘‘false positives’’) were made by participants in this study.

Correct associations were identified at a rate of 0.983, and this ratewas somewhat higher among those with more experience, espe-cially if that experience included radiographic examinations. More-over, while previous studies have evaluated the technique’sreliability using computerized models, our results indicate that tradi-tional visual comparison is highly accurate, even when frontalsinuses are small.

Acknowledgments

We thank Lee Jantz and the University of Tennessee Depart-ment of Anthropology for access to the radiographs. We alsothank the volunteers who participated in the study, as well asRichard Thomas and Brian Carroll of the FBI Laboratory forreviewing earlier drafts of this manuscript. Kc Wendler andJoshua Friedman assisted with data collection.

FIG. 1—Experience levels of participants.

FIG. 2—Correct association rates by experience level.

FIG. 3—Most frequently missed frontal sinus match.

1414 JOURNAL OF FORENSIC SCIENCES

References

1. Angual M, Derczy K. Personal identification on the basis of antemortemand postmortem radiographs. J Forensic Sci 1998;43:1089–93.

2. Asherson N. Identification by frontal sinus prints: a forensic medicalpilot survey. London, UK: H.K. Lewis & Co. Ltd, 1965.

3. Atkins L, Potsaid MS. Roentgenographic identification of humanremains. JAMA 1978;240:2307–8.

4. Brothwell DR, Molleson T, Metreweli C. Radiological aspects of normalvariation in earlier skeletons: an exploratory study. In: Brothwell DR,editor. The skeletal biology of earlier human populations. New York,NY: Pergamon Press, 1968;149–72.

5. Buckland-Wright JC. A radiological examination of frontal sinuses inearly British populations. Man 1970;5:512–7.

6. Camps FE. Radiology and its forensic application. In: Camps FE, editor.Recent advances in forensic pathology. London, UK: J. & A. Churchill,1969;149–60.

7. Cheevers LS, Ascencio R. Identification by skull superimposition. Int JForensic Dent 1977;13:14–6.

8. Christensen AM. Testing the reliability of frontal sinus outlines in per-sonal identification. J Forensic Sci 2005;50(1):18–22.

9. Cryer MH. Some variations in the frontal sinuses. JAMA 1907;48:284–98.

10. Culbert WL, Law FL. Identification by comparison of roentgenogramsof nasal accessory sinuses and mastoid processes. JAMA 1927;88:1634–6.

11. Fischman SL. The use of medical and dental radiographs in identifica-tion. Int Dent J 1985;35:301–6.

12. Gulisano M, Pacini P, Orlandini GF, Colosi G. Anatomico-radiologicalfindings on the frontal sinus: statistical study of 520 human cases. ArchItal Anat Embriol 1987;83(9):9–32.

13. Haglund WD, Fligner CL. Confirmation of human identification usingcomputerized tomography (CT). J Forensic Sci 1993;38:708–12.

14. Hanson CL, Owsley DW. Frontal sinus size in Eskimo populations. AmJ Phys Anthropol 1980;53:251–5.

15. Harris AMP, Wood RE, Nortje CJ, Thomas CJ. The frontal sinus: aforensic fingerprint? – A pilot study. J Forensic Odonto-stomatol 1987;5:9–15.

16. Joblanski NG, Shum BS. Identification of unknown human remains bycomparison of antemortem and postmortem radiographs. Forensic Sci Int1989;42:221–30.

17. Kerley ER, Snow CC. Authentification of John F. Kennedy autopsyradiographs and photographs. Final report to the Select Committee ofAssassinations. Washington, DC: U.S. House of Representatives, March9, 1979.

18. Kirk NJ, Wood RE, Goldstein M. Skeletal identification using the fron-tal sinus region: a retrospective study of 39 cases. J Forensic Sci2002;47(2):318–23.

19. Koertvelyessy T. Relationships between the frontal;sinus and climacticconditions: a skeletal approach to cold adaptation. Am J Phys Anthropol1972;37:161–72.

20. Kullman L, Eklund B, Grudin R. The value of the frontal sinus in iden-tification of unknown persons. J Forensic Odonto-stomatol 1990;8:3–10.

21. Marek Z, Kusmiderski J, Lisowski Z. Radiograms of the paranasalsinuses as a principle of identifying catastrophe victims and unknownskeletons. Arch Kriminol 1983;2:1–6.

22. Marlin DC, Clark MA, Standish SM. Identification of human remainsby comparison of frontal sinus radiographs: a series of four cases. JForensic Sci 1991;36:1765–72.

23. Messmer JM. Radiographic identification. In: Fierro MF, editor. CAPhandbook for postmortem examination of unidentified remains. Skokie,IL: College of American Pathologists, 1986;68–75.

24. Murphy WA, Gantner GE. Radiologic examination of anatomic partsand skeletonized remains. J Forensic Sci 1982;27:9–18.

25. Phrabhakaran N, Naidu MDK, Subramaniam K. Anatomical variabilityof the frontal sinuses and their application in forensic identification. ClinAnat 1999;12:16–9.

26. Quatrehomme G, Fronty P, Sapanet M, Grevin G, Bailet P. Identifica-tion by frontal sinus pattern in forensic anthropology. Forensic Sci Int1996;83(2):147–53.

27. Quatrehomme G, Sapanet M, Bailet P, Grevin G, Boublenza A, OllierA. Identification by frontal sinus: performance and difficulties. Proceed-ings of the 6th Annual Meeting of the International Association for Cra-niofacial Identification; 1995 Nov 8–11; Boca Raton, FL: InternationalAssociation for Craniofacial Identification, 1995.

28. Reichs KJ. Quantified comparison of frontal sinus patterns by means ofcomputed tomography. Forensic Sci Int 1993;61:141–68.

29. Reichs KJ, Dorion RBJ. The use of computerized axil tomography(CAT) scans in the comparisons of frontal sinus configurations. Cana-dian Soc Forensic Sci J 1992;25:1–16.

30. Ribeiro FA. Standardized measurements of radiographic films of thefrontal sinuses: an aid to identifying unknown persons. Ear Nose ThroatJ 2000;79:26–33.

31. Schuller A. Das Rontgenogram der Stirnhohle: ein Hilfsmittel fur dieIdentitatsbestimmung von Schadeln. Monatsschr Ohrenheilkd Laryngorh-inol 1921;5:1617–20.

32. Schuller A. A note of the identification of skulls by X-ray pictures ofthe frontal sinuses. Med J Aust 1943;1:554–7.

33. Strek P, Kaczanowski K, Skawina A, Pitynski K, Kitlinski Z, MrowkaD, et al. The morphological evaluation of frontal sinuses in HumanSkulls. Floia Morphol (Warsz.) 1992;51:319–28.

34. Ubelaker DH. Positive identification from the radiographic comparisonof frontal sinus patterns. In: Rathbun TA, Buikstra J, editors. Humanidentification: case studies in forensic anthropology. Springfield, IL:Charles C. Thomas, 1984;399–411.

35. Yoshino M, Miyasaka S, Sato H, Seta S. Classification system of frontalsinus patterns by radiography: its application to identification ofunknown skeletal remains. Forensic Sci Int 1987;34:289–99.

Additional information–reprints not available from author:Victoria A. Smith, M.N.S.Oak Ridge Associated Universities724 S. Saint Asaph Street, B-312Alexandria, VA 22314E-mail: victoria.smith@orise.orau.gov

SMITH ET AL. • VISUALLY COMPARING SMALL FRONTAL SINUSES 1415