MRI and Multinuclear MR Spectroscopy of 3, 200-Year-old Egyptian mummy brain

AJR:189, August 2007 W105

AJR 2007; 189:W105–W110

0361–803X/07/1892–W105

© American Roentgen Ray Society

Karlik et al.MR Spectroscopy of Mummy Brain

N e u ro r a d i o l o g y • O r i g i n a l R e s e a rc h

MRI and Multinuclear MR Spectroscopy of 3,200-Year-Old Egyptian Mummy Brain

Stephen J. Karlik1

Robert Bartha2

Karen Kennedy1

Rethy Chhem1

Karlik SJ, Bartha R, Kennedy K, Chhem R

Keywords: adipocere, Egypt, MRI, mummy, NMR spectroscopy

DOI:10.2214/AJR.07.2087

Received November 21, 2006; accepted after revision March 27, 2007.

1Paleoradiology Research Unit, Department of Diagnostic Radiology and Nuclear Medicine, Schulich School of Medicine and Dentistry, University of Western Ontario, London Health Sciences Center, University Hospital, 339 Windermere Rd., London, ON N6A 5A5, Canada. Address correspondence to S. J. Karlik ([email protected]).

2Imaging Research Laboratory, Robarts Research Institute, London, ON, Canada.

WEBThis is a Web exclusive article.

OBJECTIVE. Our objective was to present the MR and MR spectroscopy imaging findingsof a 3,200-year-old preserved brain from an Egyptian mummy.

MATERIALS AND METHODS. In this work, the morphology of the intact specimenwas examined by MRI at 1.5 T. Chemistry of the intact specimens was studied by proton spec-troscopy at 1.5 T and sodium nuclear MR (NMR) spectroscopy at 4.0 T. Biopsies from the tem-poral lobes were analyzed by proton and phosphorus NMR spectroscopy (14 T) or rehydratedand stained for paleohistologic study.

RESULTS. MRI showed a heterogeneous brain with convolutions, gyri, and air pockets.Paleohistology showed a uniform, disorganized cerebral substance with numerous eosinophilicstructures and argentophilic granules. Spectroscopic studies identified bound sodium ions inthe specimen and phosphate and free fatty acids in extracts.

CONCLUSION. MR techniques are a nondestructive method for the analysis of adipo-cere observed in a preserved mummy’s brain.

maging mummified samples re-mains a technologic and logisticchallenge. CT has been used to suc-cessfully image mummified re-

mains noninvasively [1–7]. It was possible touse CT to differentiate cerebral gray and whitematter in high-altitude Argentine mummies [4],and Hoffman and Hudgins [5] were able to vis-ualize the desiccated brain in female Egyptianmummies from which the intracranial contentshad not been removed. However, MRI, basedon water protons, has met with only partial suc-cess, with a limited subset of mummified tis-sues. In early reports of mummy MRI, no signalor image was obtained from a variety of pulsesequences [3, 8]. To obtain MR images fromthe extremities of a Peruvian mummy, Piepen-brink and coworkers [9] needed to rehydrate thespecimen in 20% acetone in water. AlthoughMRI was able to visualize the brain after acorpse had lain 13 months in a river bed, thebrain was not fully mummified and showed ev-idence of putrefaction [10].

An issue that complicates MRI detectabilityof mummified tissue is the conversion of nor-mal water-laden tissues to adipocere, a firmwaxy substance [11]. The formation of adipo-cere in a desiccated environment (Egyptiandesert) compounds the problem compared withmoisture-laden peat or bog conditions [12].

Modern chemical analysis within the context offorensic sciences shows that adipocere is the re-sult of hydrolysis of triglycerides into glycerinand free fatty acids [13]. Although this suggeststhat MRI of an ancient desiccated mummybrain would detect only lipid materials, wefound that bound water remained in the sampleand that it was possible to generate images ofsuch an ancient brain (Nakht, a 16-year-oldweaver from the XX Dynasty) by using fast im-aging techniques. Furthermore, multinuclearspectroscopy was able to detect sodium andphosphate ions and numerous organic bio-chemicals from temporal lobe extracts.

Materials and MethodsDescription of Mummy Brain

Nakht was a male teenage weaver who lived inthe XX Egyptian Dynasty. The two cerebral hemi-spheres of the brain were removed in 1975 when adissection of the mummy was performed [1, 14,15]. The specimens, shown in Figure 1, had a firmwaxy texture consistent with conversion to an adi-pocerous material [16].

MRI ScanningImages were acquired using a 1.5-T TwinSpeed

MR scanner (GE Healthcare) with two 3-inch(7.62-cm) diameter surface coils and a phased-ar-ray combiner. A set of three-plane fast gradient-

I

Karlik et al.

W106 AJR:189, August 2007

echo images was acquired to localize the brain. Be-cause the sample had a solid consistency, MRI pa-rameters were selected that were akin to those usedfor imaging solid anatomic structures and had veryshort TR and TE. Two-dimensional coronal T1-weighted fast spoiled gradient-recalled (FSPGR)acquisition in the steady state (GRASS) imageswere then acquired (TR/TE, 9.9/4.4; slice thick-ness, 5 mm; flip angle, 20°; field of view, 16 cm; 40averages; in-plane resolution, 0.625 × 0.833 mm).These were followed by axial 3D FSPGR (inver-sion time, 300 milliseconds; 9.9/4.4; flip angle,20°; field of view, 16 cm; 8 averages; and slicethickness, 5 mm). A TE of 4.4 milliseconds pro-duced the greatest signal intensity. Acquisition ofaxial T2-weighted images was also attempted usingsingle-shot fast spin-echo (714/63; slice thickness,5 mm; in-plane resolution, 0.78 × 0.78 mm; num-ber of excitations, 0.57).

MR SpectroscopyStimulated echo acquisition mode (STEAM)-lo-

calized 1H spectra were acquired on the 1.5-TTwinSpeed MR scanner (1,200/30; averages,1,024) with and without water suppression using aneight-channel head and neck coil. The voxel (41mL) encompassed portions of both hemispheresand a reference tube containing saline. The water-suppressed spectrum was not interpretable becausethe peaks of interest overlapped with the phantomwater signal (4.7 ppm). The scanner would notshim or execute the acquisition in the absence of thephantom. The free induction decay (FID) valueswere transferred from the scanner and processedusing fitMAN software [17]. The saline phantom

spectrum was line-broadened using a 5-Hz expo-nential filter to mimic the altered field homogeneityproduced by the presence of the mummy brain. Anonlocalized 23Na spectrum (10-kHz bandwidth,256 averages) was acquired on the mummy brainusing a 4.0-T Varian Unity Inova whole-body MRIscanner with a Siemens Sonata 16-element hybridbirdcage radiofrequency coil in the absence of aphantom. Finally, one-dimensional 1H and 31Pspectra were acquired from mummy tissue extractson a 14-T Varian Inova actively shielded spectrom-eter. Five-milligram portions of temporal tissuewere extracted for 14 days into 1 mL of deuteriumoxide (D2O) or deuterated chloroform (CDCl3) andthe clear supernatants were decanted into 5-mm nu-clear MR (NMR) tubes for analysis.

HistologyDissected temporal brain samples were rehy-

drated in phosphate-buffered saline and 10% glyc-erol for 14 days before processing and paraffin em-bedding [18]. Five-micron sections were cut andstained with H and E, solochrome cyanin R, andBielschowsky method silver stain.

ResultsThe first-ever successful MR images of an

ancient mummy brain are illustrated inFigure 2 (coronal view) and Figure 3 (axialview). Obtaining these images was techni-cally challenging, and the specimens could beviewed only in the T1-weighted sequence.Despite the waxy appearance of the brainhemispheres consistent with the formation offatty adipocere, images were formed from re-

sidual water in the tissue. Presumably, thesewater molecules had very short T1 relaxationtimes because no signal saturation was notedin the T1-weighted images despite very shortTR and TE; short T2 relaxation times werealso presumed because no signal was detectedin the T2-weighted images. The brain imageswere quite homogeneous in signal intensityand contained surprising internal structure(Figs. 2 and 3). Coronal views (Fig. 2) ap-peared almost laminar in appearance, perhapsreminiscent of the original cerebral gyri. Theinternal structure in both imaging planes wascomplex, with the low-signal areas likely tobe air spaces resulting from tissue shrinkage.

The localized proton spectrum of the sam-ples (Fig. 4) revealed two principal water sig-nals in the mummy brain from which the im-age was obtained appearing at 4.31 and 4.86ppm. The peak observed at 4.70 ppm in thiswater-unsuppressed spectrum corresponds towater from the saline phantom (Fig. 4), whichwas not present for the acquisition of the im-ages shown in Figures 2 and 3. Neither of theresonances at 4.31 and 4.86 ppm nor thesmaller peaks observed at 3.64 and 5.23 ppmwere indicative of lipid protons.

All three pathologic stains revealed a pre-dominantly homogeneous appearance (Fig. 5),although numerous hole artifacts were seen inthe tissue that could be due to mummification,rehydration, or fixation and processing for his-tology. In the solochrome cyanin R–stainedslides, a uniform composition of the tissue(Fig. 5A) was accompanied by a few hy-

Fig. 1—Photograph of brain specimen from teenage boy who lived in XX Egyptian Dynasty.

Fig. 2—Coronal views of two hemispheres obtained at 1.5-T using fast spoiled gradient-recalled sequence, two 3-inch (7.62-cm) coils, and phased-array combiner. Brain has heterogeneous appearance with internal laminar structure and internal pockets with low signal intensity.

MR Spectroscopy of Mummy Brain


postained globular areas, some of which dis-played a granular appearance (Figs. 5B and5C). These could be artifactual. In this stain,myelin lipids would have appeared blue, butthere was no evidence of such staining. In the Hand E–stained sections, eosinophilic structureswere observed (Figs. 5D–5F). These could befound surrounding some of the holes (Fig. 5D)but were also linear in morphology (Figs. 5Eand 5F), reminiscent of vascular structures. Nu-merous argentophilic granules were observedin the Bielschowsky method–stained slides(Figs. 5G and 5H). These particles were seen inisolation or en masse in large concentrations

and could represent calcification resulting fromcalcium desequestration during mummifica-tion. Surprisingly, even after 3,000 years, themummified and adipocerous brain retainedsome residual morphologic character on a mi-croscopic level. Because the sample appearedhomogeneous on imaging, these microscopicfeatures do not appear to contribute to the MRimages.

The 4.0-T nonlocalized sodium spectrumfrom the two hemispheres (Fig. 6, upperpanel) revealed a single symmetric broad res-onance (2,071 Hz, full width at half maxi-mum). No phantom was used for this study

Fig. 3—Axial images of mummy brain hemi-spheres obtained at 1.5 T show variable signal intensity, irregular pock-ets of low signal intensity, and complex internal structure.

Fig. 4—Localized 2D stim-ulated echo acquisition mode proton spectrum (1.5 T) of intact mummy brain hemispheres (upper panel) and saline phantom (lower panel) obtained without water suppres-sion. Circular saline phan-tom can be clearly seen in inset and its signal is at 4.7 ppm in both spectra. Two additional large peaks appear adjacent to water at 4.31 and 4.86 ppm with small shoulders at 3.64 and 5.23 ppm.

\

and there was an absence of signal when thebrain was removed (Fig. 6, lower panel). Thisresult suggests that the sodium ions were in asingle-bound environment with a short corre-lation time (τc) in the intact specimen.

NMR spectroscopy on the tissue extractionsrevealed diverse soluble components. The 31Pspectrum revealed a single symmetric peak(0.809 ppm) in the D2O extract (Fig. 7). Thisobservation is consistent with the recovery ofphosphate ions from the mummified tissue.

The proton spectrum at 600 MHz in CDCl3showed numerous components (Fig. 8). Be-cause we did not treat the samples to removemetal ions, the narrow linewidths observedsuggest that the samples did not have paramag-netic contamination. Analysis of the one-di-mensional spectrum revealed numerous reso-nances consistent with the presence of freefatty acids [19]. Peaks at 0.89, 1.24, and 2.31ppm were consistent with the identification ofpalmitic and stearic acids and hydroxystearicacid and likely contributed to these resonancesand the multiplet observed at 1.60 ppm. Mul-tiplets at 1.99 and 5.35 ppm (double bond)identified residual unhydrogenated oleic acid,and these assignments are annotated on thespectrum (Fig. 8).

There were additional singlets observed at0.66 and 1.01 ppm, and numerous multipletsof unknown origin at 1.84, 1.95, 4.60, and7.05 ppm. A singlet was observed at 7.25 ppmdue to residual protons in the CDCl3 solvent.Thus, proton spectroscopy of the CDCl3 ex-tract revealed the presence of free fatty acids,which are characteristic of adipocere [11], aswell as several other biochemicals. When ad-ipocere was found on examination of a corpseafter 3–4 months in an aqueous environment,two 9-chloro-10-methoxy(9-methoxy-10-chloro) fatty acids were found to constitute7.2% of the lipids [20]. We cannot confirm thepresence of these fatty acids because twocharacteristic singlets (3.4 and 3.7 ppm) andtwo multiplets (3.3 and 4.0 ppm) were absentfrom the spectrum of our extract.

DiscussionUsing a strategy that was optimized for tis-

sues with very short water relaxation times, ithas been possible to image the internal struc-tures of a mummified Egyptian brain. Becauseadipocere has been considered to be composedof lipid materials, we assumed that the imageswould be based on the hydrolyzed lipids. Un-expectedly, the desiccated tissue continued tobear considerable bound water in at least twoprincipal environments, and none of the proton

Karlik et al.


A B C

D E F

G H

Fig. 5—Solochrome cyanin R–, H and E– and Bielschowsky method silver–stained sections of rehydrated right temporal tissue dissected from mummy brain. (All micrographs ×400)A–C, Solochrome cyanin R staining. Tissue appears uniform in solochrome cyanin R sections (A) except for occasional lacunar areas showing pale staining (arrows, B and C).D–F, Structures in H and E–stained sections (arrows) appear linear in nature and appear to line some of tissue holes. Rest of tissue was uniformly stained.G and H, Bielschowsky method–stained sections also show dense deposits outlining margins of some holes in tissue (G) with substantial deposits of granular material throughout tissue (H).

NMR peaks observed from the specimen cor-responded to free lipids. Because bound waterhad a very short T1 relaxation time, it was onlypossible to visualize the specimens with se-quences that were optimized for fast imagingwith very short TRs and TEs.

Paleohistology revealed primarily homo-geneous tissue on rehydration. All the sam-ples showed residual tissue shrinkage, whichresulted in tissues that had holelike artifacts.Despite this limitation, there were structuralanomalies observed in the solochrome cyaninR– and H and E–stained sections. The exactidentification of these hypostained lacunarstructures and the thin linear eosinophilicthreads is unknown. Furthermore, the silverstain revealed a plethora of staining, withsome semicircular structures observed thatwere similar to those stained with eosin.

Again, the exact nature of the structures iden-tified is unknown at this time, but calcifica-tion is a possible interpretation. Calcium ionswould be released from the tissue because thenormal proteins, subcellular structures, andmembranes were altered during mummifica-tion. Similar granular structures have been at-tributed to Nissl bodies in cresyl vio-let–stained sections from a desiccatedmummified brain [19]. The uniformity of theMR images from the specimen was reflectiveof the absence of myelin staining in the solo-chrome cyanin R sections.

Spectroscopy both on the specimens and onextracts revealed additional compositional in-formation. Sodium ions were detected in thespecimen and phosphate was extracted usingD2O. In addition, the proton NMR studies ofthe extract were informative regarding the het-

erogeneity of available compounds in the bi-opsied material. Previous studies on the com-position of adipocere identify certain specificcompounds because the triglycerides are hy-drolyzed to glycerol and free fatty acids. Atfirst, stearic and oleic acids, then their metab-olites palmitic, hydroxystearic, and oxostearicacids are typically identified in this material[11]. Proton spectroscopy of the extracts wasconsistent with the detection of these com-pounds, but gas chromatography coupled tomass spectrometry would be needed to verifythe identity of all components of this complexextract. We were unable to confirm the pres-ence of 9-chloro-10-methoxy(9-methoxy-10-chloro) fatty acids [20] in our extract. Thecharacteristics of this sample of adipocereproduced in a short time (3–4 months) underwater may be different from those produced

MR Spectroscopy of Mummy Brain


when the mummy brain was desiccated over3,000 years. Nonetheless, either insufficientquantities or the solidified structure of the in-tact specimen prevented detection by protonspectroscopy at 1.5 T using our acquisition pa-rameters.

The MRI and MR spectroscopy results ob-tained at 1.5 T have potential limitations fordetection of the fatty component of adipocere.The TR and TE values that were successful inimaging residual bound water may not havebeen optimal for detection of the lipid compo-nents identified in the chloroform extract. Ul-trashort TE imaging [21–23] may have the

ability to detect the fatty components of adi-pocere, particularly if the hydrolyzed fattyacids have a short τc. Although the scannerparameters permitted a minimum TE of 1.8milliseconds at a TR of 9.9 milliseconds, ourTE choice of 4.4 milliseconds produced thehighest signal for imaging. We were unable toproduce any images using T2-weighted se-quences that would also need to be furtheroptimized to visualize very short T2 species.Similarly, the in vivo proton spectrum wasperformed at a TE of 30 milliseconds, al-though the scanner, which was optimized forpatient brain studies, permitted TEs of 15, 30,

Fig. 6—Upper panel shows 23Na spectrum obtained at 4 T of intact specimens. There is single broad resonance with linewidth of 50 ppm. Lower panel is identical acquisition without brain in coil.

Fig. 7—Proton-coupled 31P spectrum obtained at 14 T from deuterium oxide extract of temporal brain shows single resonance at 0.9 ppm with linewidth of 20 Hz. Inset shows spectrum from −3 to 3 ppm.

Fig. 8—Proton spectrum at 14 T of deuterated chloroform extract of temporal brain divided into 0.5- to 2.5-ppm and 4.0- to 7.5-ppm sections.

or 144 milliseconds. It is possible that studiesusing much shorter TE values could have de-tected lipid components with shorter T2 or τcvalues [24].

Adipocere is a generic term for the endproduct of the alteration of the soft tissues ofcorpses into a grayish-white, soft, creamlikesubstance, which over time becomes aharder solid and resistant mass. The full de-composition of this material has not been de-scribed, but chemical studies are clear thatoleic, palmitic, and stearic acids are initiallythe main constituents, followed by triglycer-ide hydrolysis, beta oxidation, and hydrationof fatty acids to also yield hydroxystearicand oxostearic acids. These constituentchanges, accompanied by saponificationwith calcium and magnesium, can cause so-lidification of the tissues [11]. This was thetype of specimen examined in these studies.The process of mummification under aridand hot conditions in the Egyptian desertproduced an unyielding material that waswaxlike in consistency. According to Elliot-Smith [25], all of 500 skulls from a prehis-torical burial site in Egypt had a preservedbrain because of burial in sand in a well-drained location. We observed the same phe-nomenon with this specimen: mummifica-tion by drying with shrinkage [16]. This wasalso seen in a naturally mummified brainfrom South Africa [18], in which CT exam-ination was correlated with a total loss of in-ternal structure, as we have seen with MRI.This type of MRI may also be effective in ad-ipocere produced under other ambient con-ditions because we appeared to image the re-tained water molecules.

In this study, MRI and multinuclear spec-troscopy were used to study cerebral speci-mens from an ancient mummy to show theirmorphology and chemical composition andthe taphonomic changes that occurred in thedesiccated brain. MRI findings were consis-tent with the uniform appearance of stainedsections of specimens extracted from the tem-poral lobes of the mummy’s brain. At a mi-croscopic level, the tissue retained somestructures reminiscent of vessel morphologyand possible calcium deposits that do not ap-pear visible on the MR images. Although thetissue contained a variety of free fatty acids,they were not detectable in the intact speci-men with the acquisition parameters used. Weview these studies as works in progress andaffirm that MRI and MR spectroscopy can in-deed be used to provide additional new in-sight into mummified tissues.

Karlik et al.


AcknowledgmentWe thank Roberta Shaw of the Royal On-

tario Museum, Toronto, for having facilitatedaccess to the specimen.

References1. Lewin PE, Harwood-Nash D. Computerized axial

tomography in medical archeology. Paleopathol-

ogy Newsletter 1977; 17:8–9

2. Shin DH, Choi YH, Shin KJ, et al. Radiological

analysis on a mummy from a medieval tomb in Ko-

rea. Ann Anat 2003; 185:377–382

3. Notman DN, Tashjian J, Aufderheide AC, et al.

Modern imaging and endoscopic biopsy techniques

in Egyptian mummies. AJR 1986; 146:93–96

4. Previgliano CH, Ceruti C, Reinhard J, et al. Radio-

logic evaluation of the Llullaillaco mummies. AJR

2003; 181:1473–1479

5. Hoffman H, Hudgins PA. Head and skull base fea-

tures of nine Egyptian mummies: evaluation with

high-resolution CT and reformation techniques.

AJR 2002; 178:1367–1376

6. Cesarani F, Martina MC, Ferraris A, et al.

Whole-body three-dimensional multidetector

CT of 13 Egyptian human mummies. AJR 2003;

180:597–606

7. Hoffman H, Torres WE, Ernst RD. Paleoradiology:

advanced CT in the evaluation of nine Egyptian

mummies. RadioGraphics 2002; 22:377–385

8. Kircos LT, Teeter E. Studying the mummy of Peto-

siris. News and Notes: The Oriental Institute (Uni-

versity of Chicago) 1991; 131:1–6

9. Piepenbrink H, Frahm J, Haase A, Matthaei D. Nu-

clear magnetic resonance imaging of mummified

corpses. Am J Phys Anthropol 1986; 70:27–28

10. Jackowski C, Thali M, Sonnenschein M, et al. Ad-

ipocere in postmortem imaging using multislice

computed tomography (MSCT) and magnetic res-

onance imaging (MRI). Am J Forensic Med Pathol

2005; 26:360–364

11. Fiedler S, Graw M. Decomposition of buried

corpses, with special reference to the formation of ad-

ipocere. Naturewissenschaften 2003; 90:291–300

12. Fiedler S, Schneckenberger K, Grwa M. Character-

ization of soils containing adipocere. Arch Environ

Contam Toxicol 2004; 47:561–568

13. Neureiter F, Pietrusky F, Schutt E. Handwoter-

buch der Gerichtlichen Medizin. Berlin, Ger-

many: Springer, 1940

14. Lewin PK. Mummies I have known: a pediatrician’s

venture in the field of paleopathology. Am J Dis

Child 1977; 131:349–350

15. Lewin PK, Harwood-Nash D. X-ray computed to-

mography on an ancient Egyptian brain. IRCS Med-

ical Science 1977; 5:78

16. Oakley KP. Ancient preserved brains. Man 1960;

60:90–91

17. Bartha R, Drost DJ, Williamson PC. Factors affect-

ing the quantification of short echo in-vivo 1H MR

spectra: prior knowledge, peak elimination, and fil-

tering. NMR Biomed 1999; 12:205–216

18. Eklektos N, Dayal MR, Manger PR. A forensic

case study of a naturally mummified brain from

the bushveld of South Africa. J Forensic Sci 2006;

51:498–503

19. Pouchert CJ, Behnke J. Aldrich library of 13C and1H FT NMR spectra. Milwaukee, WI: Aldrich

Chemical Company, 1993

20. Takatori T, Yamaoka A. Separation and identifi-

cation of 9-chloro-10-methoxy(9-methoxy-10-

chloro) hexadecanoic and octadecanoic acids in

adipocere. Forensic Sci Int 1979; 14:63–75

21. Waldman A, Rees JH, Brock CS, et al. MRI of the

brain with ultra-short echo-time pulse sequences.

Neuroradiology 2003; 45:887–892

22. Robson MD, Bydder GM. Clinical ultrashort echo

time imaging of bone and other connective tissues.

NMR Biomed 2006; 19:765–780

23. Tyler DJ, Robson MD, Henkelman RM, et al. Mag-

netic resonance imaging with ultrashort TE (UTE)

pulse sequences: technical considerations. J Magn

Reson Imaging 2007; 25:279–289

24. Liimatainen T, Hakumaki J, Tkac I, Grohn O. Ultra-

short echo time spectroscopic imaging in rats: im-

plications for monitoring lipids in glioma gene ther-

apy. NMR Biomed 2006; 9:554–559

25. Elliot-Smith G. On the natural preservation of the

brain in the ancient Egyptians. J Anat Phys 1902;

36:375–381

MRI and Multinuclear MR Spectroscopy of 3, 200-Year-old Egyptian mummy brain

Documents

Transcript of MRI and Multinuclear MR Spectroscopy of 3, 200-Year-old Egyptian mummy brain