Facial Reconstruction of a Wrapped
Egyptian Mummy Using MDCT
OBJECTIVE. Facial reconstruction of mummies and corpses in general is important in
Federico Cesarani1
anthropological, medical, and forensic studies. The purpose of our study was to evaluate the
Maria Cristina Martina2
role of MDCT examination for 3D facial reconstruction and report the results of multidisci-
Renato Grilletto3
plinary work performed by radiologists, anthropologists, and forensic police in reconstructing
Rosa Boano3
the possible physiognomy of an ancient Egyptian mummy.
Anna Maria Donadoni Roveri4
MATERIALS AND METHODS. Three-dimensional MDCT data were obtained from a
Valter Capussotto5
well-preserved, completely wrapped Egyptian mummy from the collection of the Egyptian Mu-
Andrea Giuliano5
seum in Torino, Italy, which dated from the XXII or XXIII dynasty (945 715 BC). Data were
Maurizio Celia5
used as a model for the rapid prototyping stereolithographic technique, a method that allows the
Giovanni Gandini2
creation of a 3D model with digital data using synthetic materials such as a resin or nylon.
RESULTS. The physical creation of the face was accomplished with boosting techniques per-
formed by the progressive layering of plasticine on the nylon model according to the anthropo-
metric data, the conditions of the soft residual dehydrated tissues, and the most accepted scientific
and anthropological criteria. CT is the only noninvasive method for obtaining fundamental data
for 3D reconstructions of the skull and the body, especially with wrapped mummies.
CONCLUSION. Our multidisciplinary cooperative study produced a model of the face of an
individual who lived nearly 3,000 years ago, which would not previously have been possible unless
we unwrapped, destroyed, and altered the conservation of the bandages and the mummy.
T studies in the scientific litera- and from the Ptolemaic and Roman periods (332 BC
395 AD) obtained detailed acquisitions of the skulls
ture of ancient human mummies
C
and highly accurate 3D reconstructions of the skulls
have supplied anthropologists
and dehydrated soft tissues.
with fundamental information. Facial recon-
Those data form the basis of our attempt to per-
Received September 8, 2003; accepted after revision
structions of wrapped mummies using CT
form the facial reconstruction of the mummy (ac-
April 19, 2004.
data have been reported [1 3], although few
cession number, SUPPL 5226, CGT 13011) of
1
Struttura Operativa Complessa di Radiodiagnostica,
of them have used the helical technique and,
Harwa, an artisan who lived during the XXII or
Presidi Ospedalieri Riuniiti, Azienda Sanitaria Locale 19,
as far as we know, MDCT has never been
Asti, Regione Piemonte, Italy. Address correspondence to XXIII dynasty (945 715 BC) that was found in
F. Cesarani, Via San Felice 46/4, Pino Torinese 10025, Italy used before. Our group previously per-
the Valley of the Queens by E. Schiaparelli during
([e-mail]).
formed a whole-body CT study of Egyptian the archaeological excavations that took place in
2
Instituto di Radiologia Diagnostica ed Interventistica,
1903 1906 [5]. The items in the collection of the
mummies using the MDCT technique [4].
Universitą di Torino, Azienda Ospedaliera San Giovanni
Egyptian Museum are numbered using four in-
Now we report the facial reconstruction of a
Battista di Torino, Ospedale Molinette, Corso Bramante 84,
dexes: CAT designates the Catalogo Antico
well-preserved, completely wrapped mummy
Torino 10126, Italy.
Torino; CGT, Catalogo Generale Torino; PROVV,
3 in cooperation with anthropologists and the
Dipartimento di Biologia Animale e dell Uomo, Laboratorio
Provvisorio; and SUPPL, Supplemento al CAT.
di Antropologia Fisica, Universitą di Torino, via Accademia forensic police, and we describe the tech-
The procedures through which the final model
Albertina 17, Torino 10123, Italy.
niques that we used to obtain the physiog-
was achieved can be divided into several steps.
4
Soprintendenza al Museo delle Antichitą Egizie di Torino,
nomy of the mummy.
The first step was the acquisition of data using
via Accademia delle Scienze 6, Torino 10123, Italy.
an MDCT scanner (LightSpeed QX/i, GE Health-
5
Gabinetto Interregionale di Polizia Scientifica per il
care). Preliminary anteroposterior and lateral scout
Piemonte e la Valle d Aosta di Torino, c.so Vinzaglio 10,
Materials and Methods
images were obtained to optimize the field of
Torino 10121, Italy.
Our previous MDCT study [4] of the whole bodies view. For the dedicated acquisition of the head,
AJR 2004;183:755 758
of 13 completely wrapped Egyptian mummies from these parameters were used: helical scanning, 0.8-
0361 803X/04/1833 755
the collection of the Egyptian Museum in Torino, dat- sec rotation time, 1.25-mm slice thickness, 7.5
ing from the III and IV dynasties (2650 2450 BC) mm/sec table speed, 0.7-mm reconstruction inter-
� American Roentgen Ray Society
AJR:183, September 2004 755
Cesarani et al.
val, large field of view, 120 kV, and 140 mA. (Duraform). The reconstruction was performed in
The anthropological indexes measured on
Three hundred fifty-five images were obtained, steps of 0.1 mm, working from the neck to the ver-
the axial and multiplanar reconstructions are
and the scanning time was 27.4 sec. tex. After the model cooled, excess powders were re-
reported in Table 1.
The second step consisted of postprocessing moved and a computerized system checked the
The facial reconstruction allowed the
performed with a Precision 530 workstation with resulting model against the mathematic model to de-
identification of a man approximately 45
dual processors, Intel Xeon 1.7 GHz, and 2,048 tect possible inaccuracies.
years old at the time of death. We emphasize
MB of random access memory (Dell Computer) The final facial reconstruction was then per-
that the reconstruction did not allow estab-
with version 2.5 Vitrea software (Vital Images). formed on the skull cast by the combined work of
lishing the fatty layers of the face. The mus-
After the analysis of the external aspects of the the anthropologist and the forensic artist, using the
cle insertions are clearly visible on the skull
head, which was automatically reconstructed by Manchester protocol reconstruction method [6] and
the software, the virtual removal of the bandages the information given by the residual soft tissues. In bones reconstructed according to the CT data
was performed. In this case the bandages were sep- particular, pegs were positioned at marked points on
and helped the forensic artist and anthropol-
arated from the superficial dried tissues of the the resin model of the skull according to the protocol
ogist to infer the possible thickness of the
head, delimiting the layers of the wrappings from parameters to determine the correct thickness of the
muscles, but fat does not leave signs on the
the whole circumference of the head on the axial soft tissues. The thickness of the pegs was estab-
skull. Because of this reason, dehydration,
scans and excluding them from the reconstruction. lished by the anthropologist according to the marks
and the embalming procedures, we do not
Several attempts at virtual removal of the bandages of muscle insertions on the skull model. Among the
know how fatty the face was.
were performed to obtain the best delimitation of pegs, strips of plasticine were laid and the empty
The reconstruction parameters estab-
the soft tissues (Fig. 1A). A 3D reconstruction of spaces were filled with modeling material (Fig. 1E).
lished for an average individual were attrib-
the cranial bones was also obtained using preset al- The model with the residual soft tissues gave impor-
uted to the face of Harwa according to the
gorithms (Fig. 1B). tant information regarding the morphology of the
The third step consisted of generating the com- nose, the mouth, and the ears. CT data of the whole bodies previously stud-
puterized CT data to obtain two models made of We used CT data from the whole bodies previ- ied [4] and the cranial indexes. The results
resin. The first was a highly precise reproduction ously studied [4] to reconstruct parameters for an
are also strengthened by the studies of
of the skull (Fig. 1C), and the second was the face of average individual. We decided not to add beard,
Chantre [7], Thomson and Randall-Maciver
the mummy with its residual soft tissues still present mustache, hair, or makeup to limit artistic inter-
[8], and Marro [9], according to which a
under the bandages (Fig. 1D). For the skull model, pretation and give precedence instead to the sci-
gradual increase of the cephalic index of an-
the bone structures were easily depicted from the entific data.
cient Egyptians toward mesocephaly oc-
soft tissues according to their density, but for the sec- Recently, a plaster cast was made to preserve
curred in the Greek and Roman periods.
ond model an accurate delimitation of the soft tis- the model and allow its exhibition.
Details of the soft tissues were reproduced
sues of the face had to be performed on each CT
with high accuracy. Also, a small lesion was
slice. This was achieved by cooperation between the
Results
radiologist and the computer-assisted design techni- reproduced on the superficial surface of the
cian to exclude the bandages and debris from the re- The embalming procedures used to pre- left temporal area. Already visible on the ax-
construction. We used a dedicated software system
serve the body were identified from the eval- ial scans as a small isodense homogeneous
for interfacing medical scanner information from
uation of axial scans and multiplanar lesion without calcifications, it is probably a
CT or MRI with rapid prototyping (Mimics 7 soft-
reconstructions. The absence of brain mate- nevus. The reconstructions show it clearly
ware, Materialize). These data were then transferred
rial, the persistence of only thin meningeal (Fig. 1F), along with other details of the face.
to a Vanguard selective laser sintering system (3D
sheets, and the interruption of the ethmoidal
Systems Italia) for the reproduction of the resin
cells revealed that the brain was removed
model. The selective laser sintering system is a de-
Discussion
through the nasal cavity. The teeth are in
vice that produces a model polymerizing thin layers
of plastic material using the computer-assisted de- poor condition, and no other evidence of dis- Facial reconstruction from human remains
sign file of an object. We used Polyamide powders ease can be seen. is fundamental in forensic science for identi-
TABLE 1 Mummy of Harwa Cranial Chronology During XXII and XXIII Dynasties (945 715 BC)
Indexa Description Values Features
8/1 Maximum cranial breadth � 100 / maximum cranial length = cranial index 76.75 Mesocephalic
17/1 Cranial height � 100 / maximum cranial length = cranial length height index 70.85 Medium-height skull; orthocephalic
17/8 Cranial height � 100 / maximum cranial breadth = cranial breadth height index 92.30 Medium-height, medium-large skull;
metriocephalic
9/10 Minimum frontal breadth � 100 / maximum frontal breadth 75.20 Convex forehead
9/8 Minimal frontal breadth � 100 / maximum cranial breadth = frontoparietal index 63.63 Narrow forehead; stenometopic
47/45 Total facial height � 100 / bizygomatic breadth = total facial index 92.46 Slender or narrow face; leptoprosopic
48/45 Upper facial height � 100 / bizygomatic breadth = upper facial index 61.88 Very slender or narrow face; hyperleptenic
52/51 Orbital height � 100 / orbital breadth = orbital index 98.28 Narrow orbits; hypsiconchic
54/55 Nasal breadth � 100 / nasal height = nasal index 53.99 Wide-nosed; platyrrhinic
Note. The mummy of Harwa is in the Egyptian Museum, Torino, Italy; accession number SUPPL 5226, CGT 13011). Table parameters are according to Bass [15].
a
Anthropological indexes are measured on axial and multiplanar reconstructions.
756 AJR:183, September 2004
MDCT of a Wrapped Egyptian Mummy
ABC
DE F
Fig. 1. Mummy of Harwa from Egyptian Museum, Torino, Italy (accession number SUPPL 5226, CGT 13011), dating from XXII or XXIII dynasty, 945 715 BC. Photographs
show reconstruction process and results.
A, First 3D reconstruction of face of mummy includes residual soft tissues after virtual removal of bandages.
B, Three-dimensional skull reconstruction used preset algorithms.
C, Resin model of skull was obtained using selective laser sintering system.
D, Resin model of face with its residual dehydrated soft tissues was obtained by accurately separating tissues from bandages and debris using computer-assisted system.
Hard reconstruction was then performed using selective laser sintering system.
E, Facial reconstruction was performed according to Manchester protocol method. Pegs are positioned at marked points on resin model of skull (anatomic right side). On
left side of model, strips of plasticine are laid and empty spaces are filled with modeling material to reproduce final aspect of soft tissues.
F, Final model shows facial reconstruction according to funerary mask style. Note accurate reproduction of details, such as left temporal skin lesion.
fying bodies, but it has also been of interest menting data collected by traditional physical the facial anatomy over the skull. The Ameri-
in archaeology and paleoanthropology. It is and forensic anthropology. can method relies almost entirely on the use of
one of the ways in which archaeologists Three main approaches for facial recon- average soft-tissue measurements at specific
seek to characterize the individuals of cer- struction are known [6]. The Russian method anatomic points and building the clay up to the
tain societies and has the potential for supple- as developed by Gerasimov involves building point at which the measurements are matched.
AJR:183, September 2004 757
Cesarani et al.
The third, known as the Manchester method, The average depths of facial tissues at a References
which is the one we followed, uses soft-tissue number of sites on the skull are well known,
1. D Emilia P. Tremila anni, ma non li dimostra. Ge-
measurements and careful rebuilding of the and these are used as the basis for recon- nius 1984;2:29 41
2. Brillante G. La mummia che visse due volte.
anatomic structures. struction. On the other hand, Prag and Neave
Newton - Editrice Rizzoli Corriere della Sera
Several solid replica facial reconstructions of [6] report that large areas of the face have no
1998;11:52 58
mummies from CT data have been reported in underlying bone structure (e.g., the eyes, the
3. Wilkinson C, Brier B, Neave R, Smith D. The fa-
the literature. The first example was performed mouth, the shape of the nose) that can be
cial reconstruction of Egyptian mummies and
by the Japanese archaeologist Sakuji Yoshimura used to predict facial shape, and therefore
comparison with the Fayum portraits: mummies
[1], who reconstructed the face of the Egyptian much of the final reconstruction is guess- in a new millennium. In: Lynnerup N, Andreasen
C, Berglund J, eds. Mummies in a new millen-
mummy Crisalis in 1983. With the aid of a work. In our case, data on soft dehydrated
nium: proceedings of the 4th World Congress on
computer program of the Institute of Forensic tissues from the 3D reconstructions with the
Mummy Studies, Nuuk, Greenland, September 4
Anthropology in Kyoto, starting from 49 CT workstation gave us fundamental informa-
10, 2001. Copenhagen, Denmark: Greenland Na-
axial slices, the researchers virtually unwrapped tion about the morphology of the nose, of the
tional Museum and Archives and Danish Polar
the head and reconstructed the face of a young ears, and partially of the lips, which are fre-
Center, 2002:141 146
woman who lived in Egypt at the time of Tut- quently influenced by subjectivity and are
4. Cesarani F, Martina MC, Ferraris A, et al. Whole-
mosis III (XVIII dynasty, 1479 1425 BC). It is difficult to reconstruct accurately from cra- body three-dimensional multidetector CT of 13
Egyptian human mummies. AJR 2003;180:597 606
now shown with heavy makeup and a hairstyle. nial information only.
5. Schiaparelli E. Relazione sui lavori della Mis-
In 1998, the body of Sensaos, who died in In particular, we emphasize the role played
sione Archeologica Italiana in Egitto (anni 1903
109 BC, was scanned with a Tomoscan Ex- by virtual unwrapping. Attardi et al. [10] with
1920), vol. I. Esplorazione della Valle delle
pander (Philips) in Holland [2]. In this case, an Egyptian mummy and zur Nedden et al.
Regine. Torino: Regio Museo di Antichitą, 1923
the number of slices used for the head was [14] with the Ice Man, using duplication of
6. Prag J, Neave R. Making faces using forensic and
190 and for the rest of the body was 558. the skull by means of stereolithography, archaeological evidence. London, England: Brit-
ish Museum Press, 1997
Data were elaborated with software and a started from unwrapped mummies. In our
7. Chantre E. Recherches anthropologiques en
procedure known as multijet modeling that case, we could not follow the method pro-
Egypte. Lyon, France: Rey, 1904
produced the face of the young Egyptian posed by Attardi et al. for constructing a hy-
8. Thomson A, Randall-Maciver R. The ancient
woman in 36 hr [5]. Richard Neave [3] re- brid model produced from the hard tissues of
races of the Thebaid. Oxford, England: Oxford
constructed her face with the assistance of a the mummy and the soft tissues of the head
University Press, 1905
makeup artist. Now the face shows thick lips, because the original head was completely
9. Marro G. Osservazioni morfologiche ed osteo-
quite a large nose, and small ears appropriate covered by wrappings. Only the virtual re- metriche sopra lo scheletro degli egiziani antichi:
necropoli di Assiut, 2500 3000 BC. I. Rivista di
to the typical somatic type of the Nubian moval of the bandages allowed the identifi-
antropologia, vol. XVIII, fasc. 1 2. Rome, Italy:
population to which she belonged. cation of the soft tissues so that the
Scansano Tipografia editrice degli Olmi,
Later, the same authors [3] compared the re- reconstruction of the face and soft residual
1913:63 109
sults of the computer facial reconstruction of tissues could be used as reference.
10. Attardi G, Betr� M, Forte M, et al. 3D facial re-
four Egyptian mummies from Fayum with We avoided subjective interpretations by
construction and visualization of ancient Egyp-
their corresponding portraits. They found good giving no hair, beard, or additional color to tian mummies using spiral CT data: soft tissue
reconstruction and textures application. Los An-
correspondence between the reconstructions the skin.
geles, CA: ACM SIGGRAPH Publications,
and the portraits in all instances except one. Multidisciplinary cooperation already re-
1999:223 239
An article on computer-aided facial recon- ported by other authors [10, 14] was essential
11. Cavalcanti MG, Haller JW, Vannier MW. Three-
struction with dedicated software of an un- during the different phases of reconstruction.
dimensional computed tomography landmark
wrapped mummified Egyptian head of Although we followed a strict scientific pro-
measurement in craniofacial surgical planning:
uncertain period was recently presented by tocol, variability in the final results can be
experimental validation in vitro. J Oral Maxillo-
Attardi et al. [10]. According to the authors, admitted because of some subjective inter- fac Surg 1999;57:690 694
12. Rocha Sdos S, Ramos DL, Cavalcanti Mde G.
this system provides faster execution time and pretation of the forensic artist.
Applicability of 3D-CT facial reconstruction for
a more scientific approach than traditional fa- Facial reconstruction of mummies and
forensic individual identification. Pesqui Odontol
cial reconstruction because no manual skill is corpses is important in anthropological,
Bras 2003;17:24 28
required and no artistic freedom is given. medical and forensic studies. CT is the only
13. Tyrrell AJ, Evison MP, Chamberlain AT, Green
It becomes clear on reviewing the litera- noninvasive method to obtain fundamental
MA. Forensic three-dimensional facial recon-
ture that the continuous development of ac- data for 3D reconstructions of the head and
struction: historical review and contemporary de-
velopments. J Forensic Sci 1997;42:653 661
quisition and reconstruction techniques has body, especially of wrapped mummies. Our
14. zur Nedden D, Knapp R, Wicke K, et al. Skull of
progressively improved the accuracy of re- multidisciplinary study produced a model of
a 5300-year-old mummy: reproduction and inves-
sults. The accuracy of MDCT, with the ac- the face of a man who lived nearly 3,000
tigation with CT-guided stereolithography. Radi-
quisition of thin overlapping slices and 3D years ago, which we would otherwise have
ology 1994;193:269 272
reconstructions as a method that faithfully never seen unless we unwrapped, altered,
15. Bass W. Human osteology: a laboratory and field
reproduces human anatomy, has been widely and destroyed the integrity of the bandages
manual of the human skeleton. Columbia, MO:
reported in the literature [11 13]. and the mummy. American Archaeological Society, 1971
758 AJR:183, September 2004