THE ANATOMICAL RECORD 231593-598 (1991)
Morphometry and Magnetic Resonance lmaging of the Human Brain in Normal Controls and Down’s Syndrome SERGE WEIS Institute of Neuropathology, Ludwig-Maximilians University of Munich, Thalkirchnerstrasse 36, 0-8000 Munich, Germany
ABSTRACT
A new powerful stereological tool for exact quantification of brain structures on magnetic resonance imaging (MRI) scans was used. Applying Cavalieri’s principle, unbiased estimation of volume can be obtained. The method was applied to estimate the volume of different brain structures from normal controls. Data were used for comparison with data obtained by analyzing the brains of persons with Down’s syndrome. A normalization procedure based on volume of cranial cavity is introduced and its advantages discussed, as is the coefficient of error a s a n indicator for the precision of the measurement.
Morphometry gives reliable, objective, and reproducible data that allow one to easily compare results derived from brains of normal control persons as well as from patients with different clinical conditions. Stereology, a morphometric discipline, is a body of mathematical methods relating three-dimensional parameters defining the structure to two-dimensional measurements obtainable on sections of the structure (Weibel, 1979). By these means, structural changes of the brain occurring under various conditions can easily be assessed. The first step in a morphometric study is to assess at the macroscopic level the size of different brain structures as well as their changes in size. However, morphometric investigations in gross anatomy of human brains were not frequent during the last decades and examined tissue derived from autopsy. The analyses served mainly for correlations between the volume of brain structures and whole brain volume (Klekamp, 1987; Paul, 1971; Schlenska, 1969; Wessely 1970; Zilles, 1972), although in rare instances reports focused on how age changes affect different brain regions (Eggers et al., 1984; Paul, 1971; Zilles, 1972). Many more papers were published dealing with quantitative comparative changes during brain evolution in various species. Furthermore, allometric analyses of different brain structures were reported in these studies (for review see Hofman, 1989). The general impression that gross anatomy had become a superfluous research field cannot be denied. However, with the advent of new imaging techniques in neuroradiology, i.e., axial computerized tomography (CT) and nuclear magnetic resonance imaging (MRI), anatomists should direct their interest toward these new facilities. These methods allow one to analyze the brains of living human beings at a high power of resolution. Investigations of autopsy brains may be affected by possible postmortem problems like preterminal events, postmortem delay time, fixation and weighing, and emptying of ventricles. All of the mentioned problems can be avoided when using CT or MRI scans. However, CT scans are quite inadequate for running morphometric evaluations since resolution and delineation of dif0 1991 WILEY-LISS,
INC.
ferent brain structures are limited. In contrast, a higher resolution and proper delineation of grey and white matter as well as of brain nuclei are provided by MRI (Drayer, 1988; Wahlund et al., 1990). In this study, Down’s syndrome (DS) was chosen a s a clinical entity. DS is a major condition of mental retardation resulting from meiotic nondisjunction of chromosome 21, or from translocation of parts of this chromosome, or from mosaicism (Cooper and Hall, 1988; Loesch and Smith, 1976). Morphometric data based on stereological methods are lacking so far. The only attempt to quantify brains of living persons with DS was based on CT (Schapiro e t al., 1987, 1989). Descriptions based on brain autopsies showed that neuropathological features of DS are decreased brain weight, reduced size of frontal lobes, brainstem, and cerebellum, and shortened fronto-orbital diameter (Scott et al., 1983). In the present study, data are presented that were obtained by quantitative analyses of the human brain using MRI. Cavalieri’s principle, which is a powerful stereological tool for estimating volume of brain structures, was applied. Furthermore, problems related to normalization of brain volume to brain height were considered and a normalization procedure based on cranial cavity volume is proposed. MATERIALS AND METHODS Sample
Inversion recovery MRI scans (TR IR 1800, TE 80, TI 400) were obtained with a 0.5 Tesla superconductivity system GYROSCAN S5 (Philips). The axial scans were made parallel to the orbitomeatal axis and were 10 mm thick. The control group included seven persons aged 36 to 44 years (5 men and 2 women; mean age: 38.1 years), who exhibited neither neuropathological nor other clinical symptoms. The clinical group consisted of seven adult persons with DS aged 30-45 years who were part of a n inter-
Received November 2, 1990; accepted February 11, 1991.
594
S. WEIS
TABLE 1. Volumetric data (cm3)of different brain structures (comparison between our data derived from MRI with data published previously) This study 7 1,313 632 462 23 16 8 14 149 29 1.571
Sample size Brain Cerebral cortex White matter Ventricles Thalamus Caudate Lentiformis Cerebehm Brainstem Cranial cavitv
Schlenska (1969) 7 1,501 711 505 25 e e e
160 35 nd
Wessely (1970) 31 1,267
Paul (1971) 31 1,267 :79
Zilles (1972) 78 1,198 t55
d
b
lT
b
b
b
b
b d
a b
15"
141 31 nd
d
nd
b
nd nd nd
Eggers et al. (1984) 12 1,184 586 391 nd 17 10 15 nd nd nd
"For data see Paul (1971). bVolume of prosencephalon without cerebral cortex: 516 cm3 by Wessely (1970) and Paul (1971), and 481 cm3 by Zilles (1972). 'Data given only for lateral ventricles. dFor data see Wessely (1970). "Schlenska (1969) reported a value of 67.6 cm3 for caudate, putamen, amygdala, internal capsule, thalamus, and glopbus pallidus. nd: not determined
disciplinary investigation where medical screening tests, as well a s psychological assessments, were included (Weber and Rett, 1991).
Volume of the cranial cavity was set at 100% and volume percentage of each brain structure was calculated:
Cavalieri's Principle
Volume measurements were based on Cavalieri's principle (Cavalieri, 1635; Cruz-Orive, 1987a; Gundersen and Jensen, 1987). The volume (V) of any object may be estimated from randomized and parallel sections separated by a known distance d, by summing up the areas of all cross sections (A,) of the object, and multiplying this sum by d. The formula of Cavalieri's principle is then given by: est V
=
A,
d. i=l
where ccv = volume of cranial cavity; vbs = volume of each brain structure; and plbs = percentage of each brain structure. Furthermore, in order to validate the previous normalization, the volume of different brain structures was related to the volume of the whole brain. Thus the volume of the whole brain was set a t 100% and volume percentage of each brain structure was calculated: 1 0 0 ' vbs = pzbs vbrain
(3)
Normalization Procedure
where vbrain = volume of the whole brain; vbs = volume of each brain structure; and p2bs = percentage of each brain structure. Values of plbs and pzbs were calculated for controls and persons with DS; the existence of differences between the two groups was statistically tested. Coefficient of Error The coefficient of error (CE) is a n indicator of the precision of the estimates made in a n individual brain. The calculation of the CE when using systematic sampling has been developed by Matheron (1971) and elaborated by Gundersen and Jensen (1987). (For details about CE, see Gundersen and Jensen, 1987, as well a s West and Gundersen, 1990). Coefficient of error was calculated for different brain structures by using a set of data based on 128 MRI scans derived from a n individual brain. The CE was calculated for different slice thicknesses (i.e., 1, 2, 5, 10 mm). For statistical analysis, the one-way analysis of variance (ANOVA) was computed with the software package STATGRAPHICS.
A normalization procedure for brain structures was established as follows: the volume of different brain structures was related to the volume of cranial cavity.
Volumetric data of different brain structures from normal control persons obtained by morphometric
The MRI scans were randomized since the first slice hitting the brain fell randomly, followed by systematic sections with the known fixed interval equal to the thickness of the scanning plane. The thickness of the first slice was chosen less than 10 mm and t h a t of the following slices equaled 10 mm. Profile area of the further defined brain structures was measured semiautomatically in each slice with the image analysis system IBAS 2000 (Kontron, Eching, Germany). Profile areas of each brain structure were summed up and multiplied by slice thickness. The volume of the following brain structures was determined: cerebral cortex, white matter, ventricles (including lateral ventricles, third and fourth ventricles), thalamus, caudate nucleus, lentiform nucleus (i.e., putamen and globus pallidus), cerebellum, brainstem. Volume of the whole brain resulted in summing up the volumes of the previously defined structures. In addition, the volume of the inner cranial cavity was determined.
RESULTS
595
MORPHOMETRY AND MRI OF THE HUMAN BRAIN
TABLE 2. Absolute values for volume of different brain structures obtained by morphometry of MRI scans from persons with Down’s syndrome and controlsa Control (n = 7) Brain Cerebral cortex White matter Ventricles (I-IV) Thalamus Caudate Lentiformis Cerebellum Brainstem Cranial cavity
Mean 1,313 632 462 23 16 8 14 149 29 1,571
(SDI (146) (65) (56) (9) (4)
(1) (1) (31) (6) (231)
Down’s syndrome (n = 7) Mean (SDI 1,081 (81) 528 (40) (51) 360 (13) 29 14 (6) 8 (2) 14 (3) 121 (12) 24 (3) 1,443 (99)
P 0.003 0.004 0.004 0.36 0.46 0.96 0.76 0.05 0.07 0.20
“Volume is given in cm3.
evaluation of MRI-scans are given in Table 1. Our volumetric data are comparable with those data obtained from autopsy brains and that have previously been published by Schlenska (19691, Wessely (1970), Paul (19711, Zilles (1972), and Eggers et al. (1984); see Table 1. The morphometric data of the different brain structures for the control group and the investigation group of persons with Down syndrome are shown in Table 2. Volume of the whole brain was significantly smaller in persons with DS as compared to controls. This difference in volume resulted from a reduced volume of cerebral cortex and white matter in DS. The volume of cerebellum in persons with DS was significantly reduced as compared to the control group. The other brain structures did not differ significantly between persons with DS and control subjects. Comparing persons with DS to controls by using ratios (volume of brain structures related either to volume of cranial cavity or to volume of the whole brain) gave different results (Table 3). When using the proposed normalization procedure based on volume of cranial cavity, the ANOVA computations gave approximately the same results a s obtained with absolute values (compare to Table 2). But, relating volume of brain structures to the volume of the whole brain showed results t h a t did not point out obvious differences between both groups (Table 3). The values for the coeffkient of error resulting from calculation when using different slice thicknesses are displayed in Table 4. DISCUSSION Advances in Stereology
The development of stereology took a new direction since the publication of the “disector paper” by Sterio (1984). Before t h a t time, formulas used for calculating numerical density were based on the assumption that cells under investigation were either spheres or ellipsoids, or that the thickness of the histological section was zero. In the study of Sterio (1984), for the first time a method was presented where no assumptions on form were made. However, a specific design for generating sections had to be followed. Stereological methods and theorems used before the year of 1984, i.e., before the introduction of assump-
TABLE 3. Normalization procedures of brain structures: Comparison of ratios between persons with Down’s syndrome and controls Brain structure Cortex White matter Ventricles Caudate Lentiform Thalamus Cerebellum Brainstem Brain
bs”/brain 0.12 0.28 0.08 0.18 0.07 0.88 0.90 0.89
-
bsiccvb 0.01 0.02 0.16 0.59 0.36 0.66 0.06 0.22 0.001
“Volume of brain structure. bVolume of cranial cavity.
tion-free methods, should be named “model based” stereology (MBS). Methods published since 1984 can be regrouped under the term “design based” stereology (DBS). DBS comprise methods like the disector (Sterio, 1984), the unbiased brick (Howard et al., 1985), the ortrip (Mattfeldt et al., 19851, the point-sampled intercepts (Gundersen and Jensen, 1985), the fractionator (Gundersen, 19861, the vertical sections (Baddeley et al., 1986), the selector (Cruz-Orive, 1987b), the nucleator (Gundersen, 19881, and the orientator (Mattfeldt, in press). Cavalieri’s method for volume estimation belongs to DBS; other methods for calculating the volume of a structure like Simpson’s rule as described by Aherne (1982) became obsolete. In the introductory section, it was noted that the first step in a morphometric study should be the assessment of the size of different brain structures as well a s their changes in size at the macroscopic level. Subsequently, morphometric evaluations should be performed a t light and electronmicroscopic levels. Such a n approach, which uses data ranging from gross anatomy to ultrastructure, can be termed “vertical approach.” In contrast, the “horizontal approach” analyzes the structural elements and their changes a t one specific level, i.e., changes of neurons, astroglia, and oligodendrocytes a t the light microscopic level. Combination of the vertical and horizontal approaches of investigation will lead to a complete picture of the quantitative structural changes occurring in the brain.
596
S. WEIS
TABLE 4. Values of CE (%I when using different
slice thicknesses
Section thickness Cortex Frontal cortex Parietal cortex Temporal cortex Occipital cortex White matter Caudate ncl Putamen Globus pallidum Thalamus Ventricle
1 mm 0.0 0.1 1.2 0.4
0.6 0.0 0.9 0.8 1.9 1.3 0.4
2 mm 0.1 0.4 2.8 0.9 1.2 0.1 2.5 3.0 5.6 2.8 1.0
5 mm 0.3 1.2 6.7 2.3 3.4 0.6 9.5 8.4 15.9 10.7 5.1
10 mm 1.5 3.2 22.2 5.9 8.1 2.7 -
-
-
14.3
Normalization Procedures and the “Reference Trap”
A normalization of brain volume to body height has recently been proposed by Schapiro et al. (1986). They relied on the arguments that: (1)brain size is proportional to body height, and (2) persons with DS have a smaller stature. Schapiro et al. (1986) performed the normalization of brain volume in the following way: the absolute volume was divided by the subjects height, then multiplied by 170 cm (the average height of men from their laboratory). However, the relationship between brain size and body height remains obscure, although many times evoked. Rothig and Schaarschmidt (1977) analyzed 3,406 autopsy brains but could not find any significant correlation between brain weight and body height. As a matter of fact, both authors calculated a correlation coefficient of r = 0.28 for males and r = 0.30 for females (Rothig and Schaarschmidt (1977). Similar values for correlation coefficients are found in the work of Zilles (1972). Some authors express their data in relative values or as ratios (Jernigan et al., 1990; Schapiro et al., 1987; Yamaura et al., 1980; Yerby et al., 1985). Interpretation of such results must be done very carefully. It should be kept in mind that, when using ratios, no significant changes of the structures of interest can be detected when the structure of interest as well as the reference structure show equivalent changes, i.e., when changes of both structures are correlated. This problem, called “reference-trap,” is discussed in detail by Braendgaard and Gundersen (1986). As shown in Table 3, the ratio of the volume of brain structure/volume of brain masked differences that were seen when using absolute values. It could be shown that the variable “brain volume” is correlated with the volumes of the other brain structures and hence is no suitable variable for a normalization procedure in this context. For the normalization procedure, an independent reference variable was therefore introduced. In the sample, it could be pointed out that volume of cranial cavity showed no difference between the DS group and the control group. Thus when using volume of cranial cavity as the normalization variable to other volumes of brain structures, it was demonstrated that the normalized values showed similar differences between the two groups, as when using absolute volume values. It is of interest to point out that the typical brachycephalic skull form of persons with DS, which was also obvious in the subjects of the present investi-
gation group, had no influence on the volume of cranial cavity as compared to the controls (see Table 2). Coefficient of Error and “Patient’s Compliance”
The coefficient of error is a very good indicator of the precision of a measurement (Gundersen and Jensen, 1987; West and Gundersen, 1990). The CE should have a value of less than 5% when accurate measurements are required. The general rule to be followed when CE > 5% is to either increase the number of points counted or to use a higher number of slices having a lower thickness. However, it happens sometime that these idealistic rules cannot be followed in practice. Older MRI devices need a long imaging time. Furthermore, the thinner the slices, the longer the time required for imaging. In contrast, compliance of mentally retarded persons is quite reduced. Therefore, a compromise has to be found between good quality of imaging and good accuracy of measurements for an imaging session not to strain the patient. Morphometric Procedures Applied to lmaging Techniques
Quantification of CT scans was mostly performed to assess changes of brain structures occurring with normal and pathological aging (e.g., Gyldensted, 1977; Massman et al., 1986; Meese et al., 1980; Schwartz et al., 1985; Takeda and Matsuzawa, 1985; Yamaura et al., 1980; Yerby et al., 1985). Linear, i.e., one-dimensional, measurements were used in the majority of the papers. Specific parameters related to the ventricular system were defined, namely, the greatest distance between the anterior horns, the distance between caudate nuclei, the distance between choroid plexuses, the distance between the lateral ventricles at the level of the sella media, and the widths of third and fourth ventricle were measured. Quantitative parameters of the skull vault were determined to be the greatest external diameter of the frontal bone at the level of the anterior horns, and the greatest internal and external diameter between the temporal bones. A tentative quantification of the external cerebrospinal fluid (CSF) spaces was given by measuring the width of the anterior portion of the interhemispheric fissure, the width of the insular cisterns, the number of visible sulci, and the maximum width of sulci in a standardized slice of the upper brain convexity. Basic rules of stochastic geometry were not followed. The sectioning planes of the so-called standardized slices are mostly not identical among subjects. Thus reliability and hence meaningfulness of results based on one-dimensional parameters is limited. Weis et al. (1989) could demonstrate that the incorrect and naive application of one-dimensional measurements led to speculations about a morphological substrate of cerebral asymmetries. Preference has to be given to higher order dimensional parameters whenever a reliable evaluation is possible. In gross anatomy, the reliable parameter to be determined is the three-dimensional parameter volume. Pakkenberg et al. (1989) recently made an unbiased estimation of total ventricular volume of the brain in 10 hydrocephalic children and adults obtained from CT scans. They were able to show that the linear ratio of
MORPHOMETRY AND MRI OF THE HUMAN BRAIN
597
Conclusion ventricular skull width, called Evan’s ratio, is a measure of doubtful value. In the present investigation, morphometric data With the advent of computer facilities, CT and MRI were derived from measurements of MRI scans of norscans could be analyzed by automatic image analysis mal controls. Quantitative MRI data were comparable (Caviness et al., 1989; Creasy e t al., 1986; DeLeo et al., to those values obtained from analyses of autopsy 1985; Filipek et al., 1989; Jernigan et al., 1990; Scha- brains. The comparison of MRI data with those data piro et al., 1987, 1989; Schwartz e t al., 1985). These derived from persons with DS showed differences in analytical procedures use the differences in grey level brain volume, cerebral cortex, white matter, and cereof pixels to make differentiation and delineation of bellum. In addition, measurements of caudate nucleus, brain structures. The ventricles and CSF spaces can lentiform nucleus, and thalamus were reported. No quite easily be resolved on CT scans. However, i t is not changes in volume of these structures were found bepossible to generate the exact boundaries of cerebral tween subjects with DS and controls. It is of further cortex as well as subcortical nuclei on CT scans. Thus interest to see that there existed no difference in volresults of morphometric analysis based on CT scans ume of cranial cavity between persons with DS and give only limited information. MRI allows a more exact controls. delineation between grey and white matter as well as Despite their finely developed ability to interpret between subcortical nuclei. Recently, semiautomatic morphological patterns by eye and brain, there are image analysis procedures have been developed for some things morphologists cannot assess by these morphometric analysis of MRI brain scans by Filipek means, mainly quantitative differences. Volume inet al. (19891, Caviness et al. (19891, and Jernigan et al. creases or decreases must be on the order of 30-50% to (1990). be recognizable (Reith and Mayhew, 1988). Cavalieri’s principle for volume estimation of strucComparison of Data With Other DS Studies tures is a very powerful stereological tool and when Absolute values of different brain regions derived applied to well-delineated brain structures obtained from CT or MRI scans can be compared with those val- from MRI scans or autopsy slices, provides exact and ues obtained from autopsy brains. Thus the reliability unbiased data. Combination of both techniques (morof the measurements performed on CT and MRI scans phometry and MRI) is the most suitable method to incan be evaluated. Major discrepancies in volume esti- vestigate changes in the central nervous system in a mation are found between the CT investigations of living population for cross-sectional as well as longituSchapiro e t al. (19891, Schwartz et al. (19851, Creasy e t dinal studies. al. (1986), and Eggers et al. (1984) based on autopsy brains and our own investigation, which was perLITERATURE CITED formed on MRI scans. Schapiro et al. (19891, Schwartz Aherne, W.A., and M.S. Dunhill 1982 Morphometry. Arnold, London. et al. (1985), and Creasy et al. (1986) used only seven Baddeley, A.J., H.J.G. Gundersen, and L.M. Cruz-Orive 1986 EstimaCT scans for volume estimation, which explains the tion of surface area from vertical sections. J . Micrsoc., 142.259276. discrepancies seen. It is clear that by this procedure a n additional source of bias is introduced. Out of the in- Braendgaard, H., and H.J.G. Gundersen 1986 The impact of recent stereological advances on quantitative studies of the nervous sysvestigations of Filipek et al. (1989) and Caviness et al. tem. J . Neurosci. Methods, 18t39-78. (1989) and our own investigation, it could be seen that Cavalieri B 1635 and 1966 Geometria degli Indivisibili. Unione Tipografico-Editrice Torinese, Bologna and Torino. the values derived from MRI scans were very close to those values obtained from analysis of autopsy brains Caviness, V.S., P.A. Filipek, and D.N. Kennedy 1989 Magnetic resonance technology in human brain science: Blueprint for a pro(Eggers et al., 1984). gram based upon morphometry. Brain Dev., 1Itl-13. A few quantitative investigations using CT scans Cooper, D.N., and C. Hall 1988 Down’s syndrome and the molecular biology of chromosome 21. Prog. Neurobiol., 3Ot507-530. were performed on persons with DS by Schapiro et al. (1986, 1987, 1989) and LeMay and Alvarez (1990). Creasey, H., M. Schwartz, H. Frederickson, J.V. Haxby, and S.I. Rapoport 1986 Quantitative computed tomography in dementia of the Schapiro et al. (1986, 1987, 1989) showed t h a t healthy Alzheimer type. Neurology, 36:1563-1568. adults with DS had smaller brains and smaller intra- Cruz-Orive, L.M. 1987a Stereology: Recent solutions to old problems and a glimpse into the future. Acta Stereol., 6(Suppl. 3):3-18. cranial volumes than controls. After normalization, the volumes of CSF, ventricles, and brain parenchyma did Cruz-Orive, L.M. 1987b Particle number can be estimated using a disector of unknown thickness: The selector. J. Microsc., 145:121not differ in DS as compared to controls. Furthermore, 142. Schapiro et al. (1987) showed t h a t persons with DS and DeLeo, J.M., M. Schwartz, H. Creasey, N. Cutler, and S.I. Rapoport 1985 Computer-assisted categorization of brain computerized tocontrols showed similar significant age-related incremography pixels into cerebrospinal fluid, white matter, and gray ments on volumes of CSF and ventricles. Schapiro et al. matter. Computers Biomed. Res., 18.79-88. (1989) demonstrated t h a t CSF and third ventricular Drayer, B.P. 1988 Imaging of the brain. I. Normal findings. Radiolvolumes were increased, whereas gray and white matogy, 166:785-796. ter volume were decreased in persons with DS older Eggers, R., H.Haug, and D. Fischer 1984 Preliminary report on macroscopic age changes in the human prosencephalon. A. stereologic than 45 years and showing signs of dementia as cominvestigation. J. Hirnforsch., 25.129-139. pared to younger DS adults. Schapiro e t al. (1989) con- Filipek, P.A., D.N. Kennedy, V.S. Caviness, S.L. Rossnick, T.A. cluded that brain atrophy must be present to accomSpraggins, and P.M. Starewicz 1989 Magnetic resonance imagpany dementia in older persons with DS, despite the ingbased brain morphometry: Development and application to normal subjects. Ann. Neurol., 25:61-67. presence of Alzheimer’s neuropathology in all older subjects with DS. Furthermore, in persons with DS Gundersen, H.J.G. 1986 Stereology of arbitrary particles. A review of unbiased number and size estimators and the presentation of over 50 years old, a significant widening of the temposome new ones, in memory of William R. Thomas. J. Microsc., ral horns was recently demonstrated by LeMay and 143:3-45. Gundersen, H.J.G. 1988 The nucleator. J. Microsc., I51:3-21. Alvarez (1990).
598
S.WEIS
Gundersen, H.J.G., and E.B. Jensen 1985 Stereological estimation of the volume-weighted mean volume of arbitrary particles observed on random sections. J . Microsc., 138t127-142. Gundersen, H.J.G., and E.B. Jensen 1987 The efficiency of systematic sampling in sterology and its prediction. J . Microsc., 147t229263. Gyldensted, C. 1977 Measurements of the normal ventricular system and hemispheric sulci of 100 adults with computed tomography. Neuroradiology, 14:183-192. Hofman, M.A. 1989 On the evolution and geometry of the brain in mammals. Prog. Neurobiol., 32r137-158. Howard, V., S. Reid, A. Baddeley, and A. Boyd 1985 Unbiased estimation of particle density in the tandem scanning reflected light microscope. J . Microsc., 138r203-212. Jernigan, T.L., G.A. Press, and J.R. Hesselink 1990 Methods for measuring brain morphologic features on magnetic resonance images validation and normal aging. Arch. Neurol., 47r27-32. Klekamp, J., A. Riedel, C. Harper, and H.J. Kretschmann 1987 A quantitative study of Australian aboriginal and Caucasian brains. J . Anat., 150:191-210. LeMay, M., and N. Alvarez 1990 The relationship between enlargement of the temporal horns of the lateral ventricles and the dementia in aging patients with Down syndrome. Neuroradiology, 32:104-107. Loesch, D., and C.A.B. Smith 1976 Discriminant diagnosis of 21trisomy mosaicism. J . Mental Defic. Res., 20t207-219. Massman, P.J., E.D. Bigler, C.M. Cullum, and R.I. Naugle 1986 The relationship between cortical atrophy and ventricular volume. Intern. J. Neurosci., 30:87-99. Matheron, G. 1971 The theory of regionalized variables and its application. Les cahiers du centre de morphologie mathematique de Fontainebleau, No. 5, Ecole nationale superieure des Mines. Mattfeldt, T., H.J. Mobius, and G. Mall 1985 Orthogonal triplet probes: a n efficient method for unbiased estimation of length and surface of objects with unknown orientation in space. J. Microsc., 139r279-289. Mattfeldt, T. Estimation of surface and length with the orientator. J . Microsc. (in press). Meese, W., W. Kluge, T. Grumme, and W. Hopfenmuller 1980 CT evaluation of the CSF spaces of healthy persons. Neuroradiology, 19: 131-136. Pakkenberg, B., J. Boesen, M. Albeck, and F. Gjerris 1989 Unbiased and efficient estimation of total ventricular volume of the brain obtained from CT-scans by stereological method. Neuroradiology, 31r413-417. Paul, F. (1971) Biometrische Analyse der Frischvolumina der Groahirnrinde und des Prosencephalon von 31 menschlichen adulten Gehirnen. Z. Anat. Entwick1.-Gesch., 133:325-368. Reith, A., and T.M. Mayhew 1988 Sterology and Morphometry in Electron Microscopy. Hemisphere, New York. Rothig, W., and W. Schaarschmidt 1977 Lineare Zusammenhange
zwischen Korperlange und Hirnmasse. Gegenbaurs Morph. Jahrb. Leipzig, 123t208-213. Schapiro, M.B., J.V. Haxby, C.L. Grady, and S.I. Rapoport 1986 Cerebral glucose utilization, quantitative tomography, and cognitive function in adult Down syndrome. In: The Neurobiology of Down Syndrome. C.J. Epstein, ed. Raven Press, New York. Schapiro, M.B., H. Creasey, M. Schwartz, J.V. Haxby, B. White, A. Moore, and S.I. Rapoport 1987 Quantitative CT analysis of brain morphometry in adult Down’s syndrome a t different ages. Neurology, 37:1424-1427. Schapiro, M.B., J.S. Luxenberg, J.A. Kaye, J.V. Haxby, R.P. Friedland, and S.I. Rapoport 1989 Serial quantitative CT analysis of brain morphometrics in adult Down’s syndrome at different ages. Neurology, 39r1349-1353. Schlenska, G. 1969 Messungen der Oberflache und der Volumenanteile des Gehirnes menschlicher Erwachsener mit neuen Methoden. Z. Anat. Entwick1.-Gesch., 128:47-59. Schwartz, M., H. Creasey, C.L. Grady, M.J. DeLeo, H.A. Frederickson, N.R. Cutler, and S.I. Rapoport 1985 Computed tomographic analysis of brain morphometrics in 30 healthy men, aged 21 to 81 years. Ann. Neurol., 17r146-157. Scott, B.S., L.E. Becker, and T.L. Petit 1983 Neurobiology of Down’s syndrome. Prog. Neurobiol., 21 :199-237. Sterio, D.C. 1984 The unbiased estimation of number and sizes of arbitrary particles using the disector. J . Microsc., 134t127-136. Takeda, S., and T. Matsuzawa 1985 Age-related brain atrophy: a study with computed tomography. J. Gerontol., 40t159-163. Wahlund, L.O., I. Agartz, 0. Almquist, H. Basun, L. Forssell, J . Saaf, and L. Wetterberg 1990 The brain in healthy aged individuals: MR Imaging. Radiology, 174:675-679. Weber, G., and A. Rett 1991 Down Syndrom in Erwachsenenalter. Verlag Hans Huber, Bern. Weibel, E.R. 1979 Stereological Methods, Vol. 1. Academic Press, London. Weis, S., H. Haug, B. Holoubek, and H. b i n 1989 The cerebral dominances: quantitative morphology of the human cerebral cortex. Intern. J . Neurosci., 47r165-168. Wessely, W. 1970 Biometrische Analyse der Frischvolumina des Rhombencephalon des Cerebellum und der Ventrikel von 31 adulten menschlichen Gehirnen. J. Hirnforsch., 12:ll-28. West, M., and H.J.G. Gundersen 1990 Unbiased stereological estimation of the number of neurons in the human hippocampus. J . Comp. Neurol., 296:l-22. Yamaura, H., M. Ito, K. Kubota, and T. Matsuzawa 1980 Brain atrophy during aging: A quantitative study with computed tomography. J . Gerontol., 35r492-498. Yerby, M.S., J.W. Sundsten, E.B. Larson, S.A. Wu, and S.M. Sumi 1985 A new method of measuring brain atrophy? The effects of aging in its application for diagnosing dementia. Neurology, 35: 1316-1320. Zilles, K. 1972 Biometrische Analyse der Frischvolumina verschiedener prosencephaler Hirnregionen von 78 menschlichen, adulten Gehirnen. Gegenbaurs Morph. Jahrb., 2:234-273.
Morphometry and Magnetic Resonance lmaging of the Human Brain in Normal Controls and Down’s Syndrome SERGE WEIS Institute of Neuropathology, Ludwig-Maximilians University of Munich, Thalkirchnerstrasse 36, 0-8000 Munich, Germany
ABSTRACT
A new powerful stereological tool for exact quantification of brain structures on magnetic resonance imaging (MRI) scans was used. Applying Cavalieri’s principle, unbiased estimation of volume can be obtained. The method was applied to estimate the volume of different brain structures from normal controls. Data were used for comparison with data obtained by analyzing the brains of persons with Down’s syndrome. A normalization procedure based on volume of cranial cavity is introduced and its advantages discussed, as is the coefficient of error a s a n indicator for the precision of the measurement.
Morphometry gives reliable, objective, and reproducible data that allow one to easily compare results derived from brains of normal control persons as well as from patients with different clinical conditions. Stereology, a morphometric discipline, is a body of mathematical methods relating three-dimensional parameters defining the structure to two-dimensional measurements obtainable on sections of the structure (Weibel, 1979). By these means, structural changes of the brain occurring under various conditions can easily be assessed. The first step in a morphometric study is to assess at the macroscopic level the size of different brain structures as well as their changes in size. However, morphometric investigations in gross anatomy of human brains were not frequent during the last decades and examined tissue derived from autopsy. The analyses served mainly for correlations between the volume of brain structures and whole brain volume (Klekamp, 1987; Paul, 1971; Schlenska, 1969; Wessely 1970; Zilles, 1972), although in rare instances reports focused on how age changes affect different brain regions (Eggers et al., 1984; Paul, 1971; Zilles, 1972). Many more papers were published dealing with quantitative comparative changes during brain evolution in various species. Furthermore, allometric analyses of different brain structures were reported in these studies (for review see Hofman, 1989). The general impression that gross anatomy had become a superfluous research field cannot be denied. However, with the advent of new imaging techniques in neuroradiology, i.e., axial computerized tomography (CT) and nuclear magnetic resonance imaging (MRI), anatomists should direct their interest toward these new facilities. These methods allow one to analyze the brains of living human beings at a high power of resolution. Investigations of autopsy brains may be affected by possible postmortem problems like preterminal events, postmortem delay time, fixation and weighing, and emptying of ventricles. All of the mentioned problems can be avoided when using CT or MRI scans. However, CT scans are quite inadequate for running morphometric evaluations since resolution and delineation of dif0 1991 WILEY-LISS,
INC.
ferent brain structures are limited. In contrast, a higher resolution and proper delineation of grey and white matter as well as of brain nuclei are provided by MRI (Drayer, 1988; Wahlund et al., 1990). In this study, Down’s syndrome (DS) was chosen a s a clinical entity. DS is a major condition of mental retardation resulting from meiotic nondisjunction of chromosome 21, or from translocation of parts of this chromosome, or from mosaicism (Cooper and Hall, 1988; Loesch and Smith, 1976). Morphometric data based on stereological methods are lacking so far. The only attempt to quantify brains of living persons with DS was based on CT (Schapiro e t al., 1987, 1989). Descriptions based on brain autopsies showed that neuropathological features of DS are decreased brain weight, reduced size of frontal lobes, brainstem, and cerebellum, and shortened fronto-orbital diameter (Scott et al., 1983). In the present study, data are presented that were obtained by quantitative analyses of the human brain using MRI. Cavalieri’s principle, which is a powerful stereological tool for estimating volume of brain structures, was applied. Furthermore, problems related to normalization of brain volume to brain height were considered and a normalization procedure based on cranial cavity volume is proposed. MATERIALS AND METHODS Sample
Inversion recovery MRI scans (TR IR 1800, TE 80, TI 400) were obtained with a 0.5 Tesla superconductivity system GYROSCAN S5 (Philips). The axial scans were made parallel to the orbitomeatal axis and were 10 mm thick. The control group included seven persons aged 36 to 44 years (5 men and 2 women; mean age: 38.1 years), who exhibited neither neuropathological nor other clinical symptoms. The clinical group consisted of seven adult persons with DS aged 30-45 years who were part of a n inter-
Received November 2, 1990; accepted February 11, 1991.
594
S. WEIS
TABLE 1. Volumetric data (cm3)of different brain structures (comparison between our data derived from MRI with data published previously) This study 7 1,313 632 462 23 16 8 14 149 29 1.571
Sample size Brain Cerebral cortex White matter Ventricles Thalamus Caudate Lentiformis Cerebehm Brainstem Cranial cavitv
Schlenska (1969) 7 1,501 711 505 25 e e e
160 35 nd
Wessely (1970) 31 1,267
Paul (1971) 31 1,267 :79
Zilles (1972) 78 1,198 t55
d
b
lT
b
b
b
b
b d
a b
15"
141 31 nd
d
nd
b
nd nd nd
Eggers et al. (1984) 12 1,184 586 391 nd 17 10 15 nd nd nd
"For data see Paul (1971). bVolume of prosencephalon without cerebral cortex: 516 cm3 by Wessely (1970) and Paul (1971), and 481 cm3 by Zilles (1972). 'Data given only for lateral ventricles. dFor data see Wessely (1970). "Schlenska (1969) reported a value of 67.6 cm3 for caudate, putamen, amygdala, internal capsule, thalamus, and glopbus pallidus. nd: not determined
disciplinary investigation where medical screening tests, as well a s psychological assessments, were included (Weber and Rett, 1991).
Volume of the cranial cavity was set at 100% and volume percentage of each brain structure was calculated:
Cavalieri's Principle
Volume measurements were based on Cavalieri's principle (Cavalieri, 1635; Cruz-Orive, 1987a; Gundersen and Jensen, 1987). The volume (V) of any object may be estimated from randomized and parallel sections separated by a known distance d, by summing up the areas of all cross sections (A,) of the object, and multiplying this sum by d. The formula of Cavalieri's principle is then given by: est V
=
A,
d. i=l
where ccv = volume of cranial cavity; vbs = volume of each brain structure; and plbs = percentage of each brain structure. Furthermore, in order to validate the previous normalization, the volume of different brain structures was related to the volume of the whole brain. Thus the volume of the whole brain was set a t 100% and volume percentage of each brain structure was calculated: 1 0 0 ' vbs = pzbs vbrain
(3)
Normalization Procedure
where vbrain = volume of the whole brain; vbs = volume of each brain structure; and p2bs = percentage of each brain structure. Values of plbs and pzbs were calculated for controls and persons with DS; the existence of differences between the two groups was statistically tested. Coefficient of Error The coefficient of error (CE) is a n indicator of the precision of the estimates made in a n individual brain. The calculation of the CE when using systematic sampling has been developed by Matheron (1971) and elaborated by Gundersen and Jensen (1987). (For details about CE, see Gundersen and Jensen, 1987, as well a s West and Gundersen, 1990). Coefficient of error was calculated for different brain structures by using a set of data based on 128 MRI scans derived from a n individual brain. The CE was calculated for different slice thicknesses (i.e., 1, 2, 5, 10 mm). For statistical analysis, the one-way analysis of variance (ANOVA) was computed with the software package STATGRAPHICS.
A normalization procedure for brain structures was established as follows: the volume of different brain structures was related to the volume of cranial cavity.
Volumetric data of different brain structures from normal control persons obtained by morphometric
The MRI scans were randomized since the first slice hitting the brain fell randomly, followed by systematic sections with the known fixed interval equal to the thickness of the scanning plane. The thickness of the first slice was chosen less than 10 mm and t h a t of the following slices equaled 10 mm. Profile area of the further defined brain structures was measured semiautomatically in each slice with the image analysis system IBAS 2000 (Kontron, Eching, Germany). Profile areas of each brain structure were summed up and multiplied by slice thickness. The volume of the following brain structures was determined: cerebral cortex, white matter, ventricles (including lateral ventricles, third and fourth ventricles), thalamus, caudate nucleus, lentiform nucleus (i.e., putamen and globus pallidus), cerebellum, brainstem. Volume of the whole brain resulted in summing up the volumes of the previously defined structures. In addition, the volume of the inner cranial cavity was determined.
RESULTS
595
MORPHOMETRY AND MRI OF THE HUMAN BRAIN
TABLE 2. Absolute values for volume of different brain structures obtained by morphometry of MRI scans from persons with Down’s syndrome and controlsa Control (n = 7) Brain Cerebral cortex White matter Ventricles (I-IV) Thalamus Caudate Lentiformis Cerebellum Brainstem Cranial cavity
Mean 1,313 632 462 23 16 8 14 149 29 1,571
(SDI (146) (65) (56) (9) (4)
(1) (1) (31) (6) (231)
Down’s syndrome (n = 7) Mean (SDI 1,081 (81) 528 (40) (51) 360 (13) 29 14 (6) 8 (2) 14 (3) 121 (12) 24 (3) 1,443 (99)
P 0.003 0.004 0.004 0.36 0.46 0.96 0.76 0.05 0.07 0.20
“Volume is given in cm3.
evaluation of MRI-scans are given in Table 1. Our volumetric data are comparable with those data obtained from autopsy brains and that have previously been published by Schlenska (19691, Wessely (1970), Paul (19711, Zilles (1972), and Eggers et al. (1984); see Table 1. The morphometric data of the different brain structures for the control group and the investigation group of persons with Down syndrome are shown in Table 2. Volume of the whole brain was significantly smaller in persons with DS as compared to controls. This difference in volume resulted from a reduced volume of cerebral cortex and white matter in DS. The volume of cerebellum in persons with DS was significantly reduced as compared to the control group. The other brain structures did not differ significantly between persons with DS and control subjects. Comparing persons with DS to controls by using ratios (volume of brain structures related either to volume of cranial cavity or to volume of the whole brain) gave different results (Table 3). When using the proposed normalization procedure based on volume of cranial cavity, the ANOVA computations gave approximately the same results a s obtained with absolute values (compare to Table 2). But, relating volume of brain structures to the volume of the whole brain showed results t h a t did not point out obvious differences between both groups (Table 3). The values for the coeffkient of error resulting from calculation when using different slice thicknesses are displayed in Table 4. DISCUSSION Advances in Stereology
The development of stereology took a new direction since the publication of the “disector paper” by Sterio (1984). Before t h a t time, formulas used for calculating numerical density were based on the assumption that cells under investigation were either spheres or ellipsoids, or that the thickness of the histological section was zero. In the study of Sterio (1984), for the first time a method was presented where no assumptions on form were made. However, a specific design for generating sections had to be followed. Stereological methods and theorems used before the year of 1984, i.e., before the introduction of assump-
TABLE 3. Normalization procedures of brain structures: Comparison of ratios between persons with Down’s syndrome and controls Brain structure Cortex White matter Ventricles Caudate Lentiform Thalamus Cerebellum Brainstem Brain
bs”/brain 0.12 0.28 0.08 0.18 0.07 0.88 0.90 0.89
-
bsiccvb 0.01 0.02 0.16 0.59 0.36 0.66 0.06 0.22 0.001
“Volume of brain structure. bVolume of cranial cavity.
tion-free methods, should be named “model based” stereology (MBS). Methods published since 1984 can be regrouped under the term “design based” stereology (DBS). DBS comprise methods like the disector (Sterio, 1984), the unbiased brick (Howard et al., 1985), the ortrip (Mattfeldt et al., 19851, the point-sampled intercepts (Gundersen and Jensen, 1985), the fractionator (Gundersen, 19861, the vertical sections (Baddeley et al., 1986), the selector (Cruz-Orive, 1987b), the nucleator (Gundersen, 19881, and the orientator (Mattfeldt, in press). Cavalieri’s method for volume estimation belongs to DBS; other methods for calculating the volume of a structure like Simpson’s rule as described by Aherne (1982) became obsolete. In the introductory section, it was noted that the first step in a morphometric study should be the assessment of the size of different brain structures as well a s their changes in size at the macroscopic level. Subsequently, morphometric evaluations should be performed a t light and electronmicroscopic levels. Such a n approach, which uses data ranging from gross anatomy to ultrastructure, can be termed “vertical approach.” In contrast, the “horizontal approach” analyzes the structural elements and their changes a t one specific level, i.e., changes of neurons, astroglia, and oligodendrocytes a t the light microscopic level. Combination of the vertical and horizontal approaches of investigation will lead to a complete picture of the quantitative structural changes occurring in the brain.
596
S. WEIS
TABLE 4. Values of CE (%I when using different
slice thicknesses
Section thickness Cortex Frontal cortex Parietal cortex Temporal cortex Occipital cortex White matter Caudate ncl Putamen Globus pallidum Thalamus Ventricle
1 mm 0.0 0.1 1.2 0.4
0.6 0.0 0.9 0.8 1.9 1.3 0.4
2 mm 0.1 0.4 2.8 0.9 1.2 0.1 2.5 3.0 5.6 2.8 1.0
5 mm 0.3 1.2 6.7 2.3 3.4 0.6 9.5 8.4 15.9 10.7 5.1
10 mm 1.5 3.2 22.2 5.9 8.1 2.7 -
-
-
14.3
Normalization Procedures and the “Reference Trap”
A normalization of brain volume to body height has recently been proposed by Schapiro et al. (1986). They relied on the arguments that: (1)brain size is proportional to body height, and (2) persons with DS have a smaller stature. Schapiro et al. (1986) performed the normalization of brain volume in the following way: the absolute volume was divided by the subjects height, then multiplied by 170 cm (the average height of men from their laboratory). However, the relationship between brain size and body height remains obscure, although many times evoked. Rothig and Schaarschmidt (1977) analyzed 3,406 autopsy brains but could not find any significant correlation between brain weight and body height. As a matter of fact, both authors calculated a correlation coefficient of r = 0.28 for males and r = 0.30 for females (Rothig and Schaarschmidt (1977). Similar values for correlation coefficients are found in the work of Zilles (1972). Some authors express their data in relative values or as ratios (Jernigan et al., 1990; Schapiro et al., 1987; Yamaura et al., 1980; Yerby et al., 1985). Interpretation of such results must be done very carefully. It should be kept in mind that, when using ratios, no significant changes of the structures of interest can be detected when the structure of interest as well as the reference structure show equivalent changes, i.e., when changes of both structures are correlated. This problem, called “reference-trap,” is discussed in detail by Braendgaard and Gundersen (1986). As shown in Table 3, the ratio of the volume of brain structure/volume of brain masked differences that were seen when using absolute values. It could be shown that the variable “brain volume” is correlated with the volumes of the other brain structures and hence is no suitable variable for a normalization procedure in this context. For the normalization procedure, an independent reference variable was therefore introduced. In the sample, it could be pointed out that volume of cranial cavity showed no difference between the DS group and the control group. Thus when using volume of cranial cavity as the normalization variable to other volumes of brain structures, it was demonstrated that the normalized values showed similar differences between the two groups, as when using absolute volume values. It is of interest to point out that the typical brachycephalic skull form of persons with DS, which was also obvious in the subjects of the present investi-
gation group, had no influence on the volume of cranial cavity as compared to the controls (see Table 2). Coefficient of Error and “Patient’s Compliance”
The coefficient of error is a very good indicator of the precision of a measurement (Gundersen and Jensen, 1987; West and Gundersen, 1990). The CE should have a value of less than 5% when accurate measurements are required. The general rule to be followed when CE > 5% is to either increase the number of points counted or to use a higher number of slices having a lower thickness. However, it happens sometime that these idealistic rules cannot be followed in practice. Older MRI devices need a long imaging time. Furthermore, the thinner the slices, the longer the time required for imaging. In contrast, compliance of mentally retarded persons is quite reduced. Therefore, a compromise has to be found between good quality of imaging and good accuracy of measurements for an imaging session not to strain the patient. Morphometric Procedures Applied to lmaging Techniques
Quantification of CT scans was mostly performed to assess changes of brain structures occurring with normal and pathological aging (e.g., Gyldensted, 1977; Massman et al., 1986; Meese et al., 1980; Schwartz et al., 1985; Takeda and Matsuzawa, 1985; Yamaura et al., 1980; Yerby et al., 1985). Linear, i.e., one-dimensional, measurements were used in the majority of the papers. Specific parameters related to the ventricular system were defined, namely, the greatest distance between the anterior horns, the distance between caudate nuclei, the distance between choroid plexuses, the distance between the lateral ventricles at the level of the sella media, and the widths of third and fourth ventricle were measured. Quantitative parameters of the skull vault were determined to be the greatest external diameter of the frontal bone at the level of the anterior horns, and the greatest internal and external diameter between the temporal bones. A tentative quantification of the external cerebrospinal fluid (CSF) spaces was given by measuring the width of the anterior portion of the interhemispheric fissure, the width of the insular cisterns, the number of visible sulci, and the maximum width of sulci in a standardized slice of the upper brain convexity. Basic rules of stochastic geometry were not followed. The sectioning planes of the so-called standardized slices are mostly not identical among subjects. Thus reliability and hence meaningfulness of results based on one-dimensional parameters is limited. Weis et al. (1989) could demonstrate that the incorrect and naive application of one-dimensional measurements led to speculations about a morphological substrate of cerebral asymmetries. Preference has to be given to higher order dimensional parameters whenever a reliable evaluation is possible. In gross anatomy, the reliable parameter to be determined is the three-dimensional parameter volume. Pakkenberg et al. (1989) recently made an unbiased estimation of total ventricular volume of the brain in 10 hydrocephalic children and adults obtained from CT scans. They were able to show that the linear ratio of
MORPHOMETRY AND MRI OF THE HUMAN BRAIN
597
Conclusion ventricular skull width, called Evan’s ratio, is a measure of doubtful value. In the present investigation, morphometric data With the advent of computer facilities, CT and MRI were derived from measurements of MRI scans of norscans could be analyzed by automatic image analysis mal controls. Quantitative MRI data were comparable (Caviness et al., 1989; Creasy e t al., 1986; DeLeo et al., to those values obtained from analyses of autopsy 1985; Filipek et al., 1989; Jernigan et al., 1990; Scha- brains. The comparison of MRI data with those data piro et al., 1987, 1989; Schwartz e t al., 1985). These derived from persons with DS showed differences in analytical procedures use the differences in grey level brain volume, cerebral cortex, white matter, and cereof pixels to make differentiation and delineation of bellum. In addition, measurements of caudate nucleus, brain structures. The ventricles and CSF spaces can lentiform nucleus, and thalamus were reported. No quite easily be resolved on CT scans. However, i t is not changes in volume of these structures were found bepossible to generate the exact boundaries of cerebral tween subjects with DS and controls. It is of further cortex as well as subcortical nuclei on CT scans. Thus interest to see that there existed no difference in volresults of morphometric analysis based on CT scans ume of cranial cavity between persons with DS and give only limited information. MRI allows a more exact controls. delineation between grey and white matter as well as Despite their finely developed ability to interpret between subcortical nuclei. Recently, semiautomatic morphological patterns by eye and brain, there are image analysis procedures have been developed for some things morphologists cannot assess by these morphometric analysis of MRI brain scans by Filipek means, mainly quantitative differences. Volume inet al. (19891, Caviness et al. (19891, and Jernigan et al. creases or decreases must be on the order of 30-50% to (1990). be recognizable (Reith and Mayhew, 1988). Cavalieri’s principle for volume estimation of strucComparison of Data With Other DS Studies tures is a very powerful stereological tool and when Absolute values of different brain regions derived applied to well-delineated brain structures obtained from CT or MRI scans can be compared with those val- from MRI scans or autopsy slices, provides exact and ues obtained from autopsy brains. Thus the reliability unbiased data. Combination of both techniques (morof the measurements performed on CT and MRI scans phometry and MRI) is the most suitable method to incan be evaluated. Major discrepancies in volume esti- vestigate changes in the central nervous system in a mation are found between the CT investigations of living population for cross-sectional as well as longituSchapiro e t al. (19891, Schwartz et al. (19851, Creasy e t dinal studies. al. (1986), and Eggers et al. (1984) based on autopsy brains and our own investigation, which was perLITERATURE CITED formed on MRI scans. Schapiro et al. (19891, Schwartz Aherne, W.A., and M.S. Dunhill 1982 Morphometry. Arnold, London. et al. (1985), and Creasy et al. (1986) used only seven Baddeley, A.J., H.J.G. Gundersen, and L.M. Cruz-Orive 1986 EstimaCT scans for volume estimation, which explains the tion of surface area from vertical sections. J . Micrsoc., 142.259276. discrepancies seen. It is clear that by this procedure a n additional source of bias is introduced. Out of the in- Braendgaard, H., and H.J.G. Gundersen 1986 The impact of recent stereological advances on quantitative studies of the nervous sysvestigations of Filipek et al. (1989) and Caviness et al. tem. J . Neurosci. Methods, 18t39-78. (1989) and our own investigation, it could be seen that Cavalieri B 1635 and 1966 Geometria degli Indivisibili. Unione Tipografico-Editrice Torinese, Bologna and Torino. the values derived from MRI scans were very close to those values obtained from analysis of autopsy brains Caviness, V.S., P.A. Filipek, and D.N. Kennedy 1989 Magnetic resonance technology in human brain science: Blueprint for a pro(Eggers et al., 1984). gram based upon morphometry. Brain Dev., 1Itl-13. A few quantitative investigations using CT scans Cooper, D.N., and C. Hall 1988 Down’s syndrome and the molecular biology of chromosome 21. Prog. Neurobiol., 3Ot507-530. were performed on persons with DS by Schapiro et al. (1986, 1987, 1989) and LeMay and Alvarez (1990). Creasey, H., M. Schwartz, H. Frederickson, J.V. Haxby, and S.I. Rapoport 1986 Quantitative computed tomography in dementia of the Schapiro et al. (1986, 1987, 1989) showed t h a t healthy Alzheimer type. Neurology, 36:1563-1568. adults with DS had smaller brains and smaller intra- Cruz-Orive, L.M. 1987a Stereology: Recent solutions to old problems and a glimpse into the future. Acta Stereol., 6(Suppl. 3):3-18. cranial volumes than controls. After normalization, the volumes of CSF, ventricles, and brain parenchyma did Cruz-Orive, L.M. 1987b Particle number can be estimated using a disector of unknown thickness: The selector. J. Microsc., 145:121not differ in DS as compared to controls. Furthermore, 142. Schapiro et al. (1987) showed t h a t persons with DS and DeLeo, J.M., M. Schwartz, H. Creasey, N. Cutler, and S.I. Rapoport 1985 Computer-assisted categorization of brain computerized tocontrols showed similar significant age-related incremography pixels into cerebrospinal fluid, white matter, and gray ments on volumes of CSF and ventricles. Schapiro et al. matter. Computers Biomed. Res., 18.79-88. (1989) demonstrated t h a t CSF and third ventricular Drayer, B.P. 1988 Imaging of the brain. I. Normal findings. Radiolvolumes were increased, whereas gray and white matogy, 166:785-796. ter volume were decreased in persons with DS older Eggers, R., H.Haug, and D. Fischer 1984 Preliminary report on macroscopic age changes in the human prosencephalon. A. stereologic than 45 years and showing signs of dementia as cominvestigation. J. Hirnforsch., 25.129-139. pared to younger DS adults. Schapiro e t al. (1989) con- Filipek, P.A., D.N. Kennedy, V.S. Caviness, S.L. Rossnick, T.A. cluded that brain atrophy must be present to accomSpraggins, and P.M. Starewicz 1989 Magnetic resonance imagpany dementia in older persons with DS, despite the ingbased brain morphometry: Development and application to normal subjects. Ann. Neurol., 25:61-67. presence of Alzheimer’s neuropathology in all older subjects with DS. Furthermore, in persons with DS Gundersen, H.J.G. 1986 Stereology of arbitrary particles. A review of unbiased number and size estimators and the presentation of over 50 years old, a significant widening of the temposome new ones, in memory of William R. Thomas. J. Microsc., ral horns was recently demonstrated by LeMay and 143:3-45. Gundersen, H.J.G. 1988 The nucleator. J. Microsc., I51:3-21. Alvarez (1990).
598
S.WEIS
Gundersen, H.J.G., and E.B. Jensen 1985 Stereological estimation of the volume-weighted mean volume of arbitrary particles observed on random sections. J . Microsc., 138t127-142. Gundersen, H.J.G., and E.B. Jensen 1987 The efficiency of systematic sampling in sterology and its prediction. J . Microsc., 147t229263. Gyldensted, C. 1977 Measurements of the normal ventricular system and hemispheric sulci of 100 adults with computed tomography. Neuroradiology, 14:183-192. Hofman, M.A. 1989 On the evolution and geometry of the brain in mammals. Prog. Neurobiol., 32r137-158. Howard, V., S. Reid, A. Baddeley, and A. Boyd 1985 Unbiased estimation of particle density in the tandem scanning reflected light microscope. J . Microsc., 138r203-212. Jernigan, T.L., G.A. Press, and J.R. Hesselink 1990 Methods for measuring brain morphologic features on magnetic resonance images validation and normal aging. Arch. Neurol., 47r27-32. Klekamp, J., A. Riedel, C. Harper, and H.J. Kretschmann 1987 A quantitative study of Australian aboriginal and Caucasian brains. J . Anat., 150:191-210. LeMay, M., and N. Alvarez 1990 The relationship between enlargement of the temporal horns of the lateral ventricles and the dementia in aging patients with Down syndrome. Neuroradiology, 32:104-107. Loesch, D., and C.A.B. Smith 1976 Discriminant diagnosis of 21trisomy mosaicism. J . Mental Defic. Res., 20t207-219. Massman, P.J., E.D. Bigler, C.M. Cullum, and R.I. Naugle 1986 The relationship between cortical atrophy and ventricular volume. Intern. J. Neurosci., 30:87-99. Matheron, G. 1971 The theory of regionalized variables and its application. Les cahiers du centre de morphologie mathematique de Fontainebleau, No. 5, Ecole nationale superieure des Mines. Mattfeldt, T., H.J. Mobius, and G. Mall 1985 Orthogonal triplet probes: a n efficient method for unbiased estimation of length and surface of objects with unknown orientation in space. J. Microsc., 139r279-289. Mattfeldt, T. Estimation of surface and length with the orientator. J . Microsc. (in press). Meese, W., W. Kluge, T. Grumme, and W. Hopfenmuller 1980 CT evaluation of the CSF spaces of healthy persons. Neuroradiology, 19: 131-136. Pakkenberg, B., J. Boesen, M. Albeck, and F. Gjerris 1989 Unbiased and efficient estimation of total ventricular volume of the brain obtained from CT-scans by stereological method. Neuroradiology, 31r413-417. Paul, F. (1971) Biometrische Analyse der Frischvolumina der Groahirnrinde und des Prosencephalon von 31 menschlichen adulten Gehirnen. Z. Anat. Entwick1.-Gesch., 133:325-368. Reith, A., and T.M. Mayhew 1988 Sterology and Morphometry in Electron Microscopy. Hemisphere, New York. Rothig, W., and W. Schaarschmidt 1977 Lineare Zusammenhange
zwischen Korperlange und Hirnmasse. Gegenbaurs Morph. Jahrb. Leipzig, 123t208-213. Schapiro, M.B., J.V. Haxby, C.L. Grady, and S.I. Rapoport 1986 Cerebral glucose utilization, quantitative tomography, and cognitive function in adult Down syndrome. In: The Neurobiology of Down Syndrome. C.J. Epstein, ed. Raven Press, New York. Schapiro, M.B., H. Creasey, M. Schwartz, J.V. Haxby, B. White, A. Moore, and S.I. Rapoport 1987 Quantitative CT analysis of brain morphometry in adult Down’s syndrome a t different ages. Neurology, 37:1424-1427. Schapiro, M.B., J.S. Luxenberg, J.A. Kaye, J.V. Haxby, R.P. Friedland, and S.I. Rapoport 1989 Serial quantitative CT analysis of brain morphometrics in adult Down’s syndrome at different ages. Neurology, 39r1349-1353. Schlenska, G. 1969 Messungen der Oberflache und der Volumenanteile des Gehirnes menschlicher Erwachsener mit neuen Methoden. Z. Anat. Entwick1.-Gesch., 128:47-59. Schwartz, M., H. Creasey, C.L. Grady, M.J. DeLeo, H.A. Frederickson, N.R. Cutler, and S.I. Rapoport 1985 Computed tomographic analysis of brain morphometrics in 30 healthy men, aged 21 to 81 years. Ann. Neurol., 17r146-157. Scott, B.S., L.E. Becker, and T.L. Petit 1983 Neurobiology of Down’s syndrome. Prog. Neurobiol., 21 :199-237. Sterio, D.C. 1984 The unbiased estimation of number and sizes of arbitrary particles using the disector. J . Microsc., 134t127-136. Takeda, S., and T. Matsuzawa 1985 Age-related brain atrophy: a study with computed tomography. J. Gerontol., 40t159-163. Wahlund, L.O., I. Agartz, 0. Almquist, H. Basun, L. Forssell, J . Saaf, and L. Wetterberg 1990 The brain in healthy aged individuals: MR Imaging. Radiology, 174:675-679. Weber, G., and A. Rett 1991 Down Syndrom in Erwachsenenalter. Verlag Hans Huber, Bern. Weibel, E.R. 1979 Stereological Methods, Vol. 1. Academic Press, London. Weis, S., H. Haug, B. Holoubek, and H. b i n 1989 The cerebral dominances: quantitative morphology of the human cerebral cortex. Intern. J . Neurosci., 47r165-168. Wessely, W. 1970 Biometrische Analyse der Frischvolumina des Rhombencephalon des Cerebellum und der Ventrikel von 31 adulten menschlichen Gehirnen. J. Hirnforsch., 12:ll-28. West, M., and H.J.G. Gundersen 1990 Unbiased stereological estimation of the number of neurons in the human hippocampus. J . Comp. Neurol., 296:l-22. Yamaura, H., M. Ito, K. Kubota, and T. Matsuzawa 1980 Brain atrophy during aging: A quantitative study with computed tomography. J . Gerontol., 35r492-498. Yerby, M.S., J.W. Sundsten, E.B. Larson, S.A. Wu, and S.M. Sumi 1985 A new method of measuring brain atrophy? The effects of aging in its application for diagnosing dementia. Neurology, 35: 1316-1320. Zilles, K. 1972 Biometrische Analyse der Frischvolumina verschiedener prosencephaler Hirnregionen von 78 menschlichen, adulten Gehirnen. Gegenbaurs Morph. Jahrb., 2:234-273.
