Background: White matter lesions (WML) increase with age and are associated with stroke, cognitive decline and dementia. They can be visually rated or computationally assessed. Methods: We compared WML Fazekas visual rating scores and volumes, determined using a validated multispectral image-fusion technique, in Magnetic Resonance Imaging from 672 participants of the Lothian Birth Cohort 1936 and sought explanations for subjects in whom the correlation (Spearman’s ρ) between the total Fazekas score (summed deep and periventricular ratings, 0–6) and WML volume did not concur (z-score difference >1). Infarcts were identified separately. Results: The median WML Fazekas score was 2 [inter-quartile range (IQR): 2], median WML volume 7.7 ml (IQR: 13.6 ml) and median infarct volume (n = 95) 0.98 ml. Score and volume were highly correlated (Spearman’s ρ = 0.78, p < 0.001). Infarcts did not alter the correlation. Minor discordance occurred in 94/672 (14%) subjects, most with total Fazekas score of 1 (n = 20, WML volume = 4.5–14.8 ml) or 2 (n = 50, WML volume = 0.1–34.4 ml). The main reasons were: subtle WML identified visually but omitted from the volume; prominent ventricular caps but thin body lining giving a periventricular score of 1/2 but large WML volume, and small deep focal lesions which increase the score disproportionally when beginning to coalesce with little change in WML volume. Conclusions: WML rating scores and volumes provide near-equivalent estimates of WML burden, therefore either can be used depending on research circumstances. Even closer agreement could result from improved computational detection of subtle WML and modified visual ratings to differentiate prominent ventricular caps from thin periventricular linings, and small non-coalescent from early coalescent deep WML.

White matter lesions (WML) or leukoaraiosis, commonly observed in older subjects on magnetic resonance imaging (MRI), are associated with cognitive decline, dementia, increased stroke risk and disability in old age. Accurate measurement of WML burden at presentation and progression over time is of crucial importance for epidemiological studies to determine: associations between WML, cognitive and clinical data; their causes, and the effects of new treatments in randomised trials. Moreover, consistency in the performance of different WML assessment techniques is crucial for meta-analysis of large samples (e.g. of large epidemiological observational studies, or genome-wide association studies), but are inevitably a conglomerate of different study samples. Although differences in study methods and heterogeneity of patient populations may in part explain contradictory results between studies exploring the frequency, clinical significance and risk factors [1,2,3] of WML, part of this diversity may be due to differences in the methods used to evaluate the degree of white matter damage.

There are many visual (qualitative) rating scales for assessing WML [1] and, more recently, several methods to measure WML volume (quantitative) have also been developed. Both qualitative and quantitative approaches have advantages and disadvantages [3,4,5,6,7]. There have been four previous comparisons of visual rating scales and quantitative WML volume measurements (table 1) [8,9,10,11], but their results differ: two of the studies were small, none differentiated focal lesions (e.g. stroke) from the WML volume, identified sources of disagreement or provided suggestions as to how these could be remedied.

Table 1

Previous studies comparing agreement between WMLs and visual rating scales

Previous studies comparing agreement between WMLs and visual rating scales
Previous studies comparing agreement between WMLs and visual rating scales

Limitations of visual rating scales are well recognized [8,9,12], but little attention has yet been given to difficulties encountered with computational WML volume measurement, e.g. sources of bias such as WML load itself, ability to differentiate severity of individual lesions and regional localisation of WML [7,13]. Furthermore, previous studies (table 1) used age-heterogeneous subjects, with various associated pathologies, risk factors, combined assessments obtained from different MRI protocols and performed quantitative measurements using methods that could be susceptible to intra- and inter-observer variation without providing a detailed analysis of sources of disagreement. Possible variations across centres are common, including populations, raters’ expertise and technological resources, making it difficult to evaluate the causes of the variability in previous studies.

In order to investigate factors affecting agreement between WML scales and volumes, we investigated correlations between one of the most widely-used and well-validated visual rating scales (the Fazekas scale [14]) and WML volume measured using a validated image-processing method [7] in a large, well-described cohort [4,15]. We also investigated the differences in how rating scales and volumes measure different types and extents of WML, and the influence of other common features in ageing populations such as cortical and subcortical infarcts.

Subjects

Participants came from the Lothian Birth Cohort 1936 (LBC1936), which comprises community-dwelling surviving members of the Scottish Mental Survey of 1947 [4,15]. From 700 subjects that underwent brain MRI between 8th November 2007 and 29th June 2010, aged between 71.1 and 74.3 years, 672 (358 men and 314 women) provided relevant sequences to assess WML both quantitatively (T2*-weighted and fluid attenuation inversion recovery, FLAIR) and visually using the Fazekas score (FLAIR, T2- and T1-weighted). Full details of imaging acquisition and analysis are described in Wardlaw et al. [4]. Written informed consent was obtained from all participants under protocols approved by the Lothian (REC 07/MRE00/58) and Scottish Multicentre (MREC/01/0/56) Research Ethics Committees.

MRI Scans

All MRI data were acquired using a 1.5T GE Signa Horizon HDxt clinical scanner (General Electric, Milwaukee, Wisc., USA) operating in research mode and using a self-shielding gradient set with maximum gradient of 33 mT/m, and an 8-channel phased-array head coil. The imaging data included T1-, T2-, T2*-weighted and FLAIR whole-brain scans, with sequence parameters summarised in table 2 and described in Wardlaw et al. [4].

Table 2

Parameters for each structural sequence used in this study

Parameters for each structural sequence used in this study
Parameters for each structural sequence used in this study

Identification of WML

We defined WML as punctate or diffuse areas in the white matter and deep grey matter of the cerebral hemispheres or in the brainstem that were 3 mm or larger in diameter, and hyperintense with respect to normal-appearing white and grey matter on T2-weighted and FLAIR images; some hypointensity on T1-weighted MRI was allowed as long as this was not as hypointense as cerebrospinal fluid. Using the Fazekas visual rating scale [14], WML were scored primarily on the FLAIR images, checking the T2- and T1-weighted images where necessary. One consultant neuroradiologist performed all the ratings after training on a standard dataset. Another consultant neuroradiologist cross-checked a random sample of 20% of ratings, all scans with stroke lesions and any scans where the first rater was uncertain. We used the Fazekas scale as it is one of, if not the most widely used WML visual rating scale and has been in use for over 2 decades. The stroke lesions were assessed visually by the two neuroradiologists and segmented and quantified separately from the WML. WML volumes were measured using MCMxxxVI [7], as described below. All analyses were performed blind to other imaging data and all subject information.

Fazekas Visual Rating Scale

In the Fazekas rating scale [14] periventricular and deep WML are rated separately. Periventricular hyperintensities are scored as: 0 = absence, 1 = ‘caps’ or pencil-thin lining, 2 = smooth ‘halo’ and 3 = irregular periventricular hyperintensities extending into the deep white matter. Deep white matter hyperintensities are scored as: 0 = absence, 1 = punctuate foci, 2 = beginning confluence of foci and 3 = large confluent areas. For statistical comparison with WML volumes, a total Fazekas WML score, ranging from 0 to 6, was obtained by summing the periventricular and deep white matter scores.

White Matter and Infarct Volume Measurement

One trained observer extracted the WML and any cortical or other discrete infarct T2/FLAIR hyperintense lesions using MCMxxxVI (http://sourceforge.net/projects/bric1936/) [7]. MCMxxxVI maps two or more different MRI sequences that display the tissues/lesions of the brain in different signal intensity levels to the red/green/blue colour space. It then reduces the colour levels of the fused image to 32 clusters using minimum variance quantisation. In direct comparisons, MCMxxxVI performed better than thresholding using one MRI sequence and better than other multispectral methods when using two or more sequences [16]. The T2*-weighted and FLAIR sequences were mapped respectively to the red/green colour space [7]. For the automatic generation of the binary mask of the WML, the clusters corresponding to hyperintense lesions were selected by visual inspection, setting the levels of red and green to enclose the selected clusters on the normalised colour space. Then, any incorrect areas or artefacts, etc. were manually removed to produce the ‘total lesion mask’. Stroke lesions were then manually removed from the total lesion mask to obtain the ‘WML mask’. We defined infarcts as cortical or large subcortical areas of hyperintensity on T2 or FLAIR, consistent with cerebromalacia and in a vascular distribution. Areas of tissue loss and replacement by cerebrospinal fluid due to infarcts (including lacunes) were also masked by reference to the contralateral side for symmetry. All stroke lesion masking was guided by a neuroradiologist. Note that we considered small hyperintensities ≥3 mm diameter in the deep grey matter as ‘WMLs’ and included them in the WML mask, but we did not count enlarged perivascular spaces in the WML mask or score. Hyperintense and hypointense areas of cerebromalacea due to old cortical/subcortical infarcts or lacunes were masked out from the total lesion by thresholding the FLAIR sequences using a region-growing algorithm from Analyze 10.0 (http://www.analyzedirect.com/Analyze/) [4] with manual editing. Where stroke lesions were occasionally contiguous with WML, the boundary between the two was determined by evaluation of the WML and underlying anatomy in the contralateral hemisphere and neuroradiological knowledge. To evaluate the intra-observer reliability of the WML volume measurements, the same observer remeasured 14 scans selected to represent the full range of WML load and calculated the non-parametric intra-class correlation coefficient which was very high (0.964; p < 0.01, 2-tailed). Intracranial volume (ICV) and brain tissue volume were also measured using the MCMxxxVI method [4]. Volumes of total lesion, stroke lesion and WML were determined in absolute units (millilitres) and percentage values relative to ICV and brain tissue volume.

Statistical Analysis

All data were checked for normality using the Kolmogorov-Smirnov test after performing Lilliefors significance correction and examining histograms of the distributions. WML absolute volumes and their volume relative to brain tissue volume and ICV were all half-normally distributed in the whole sample and not normally distributed in the subsample of subjects with stroke lesions. Correlation between the total Fazekas score and WML volume were tested using Spearman’s ρ. Subjects in whom the visual ratings and the volumetric measurements did not concur (z-score difference >1, i.e. more than twice the variance of the z-score difference distribution) were examined to identify reasons for disagreement. We tested the sensitivity of both rating and volume measurements to detect differences between participants who had stroke lesions and those who did not with χ2 for Fazekas scale and Mann-Whitney test for WML volumes. To test the relationship between the Fazekas visual rating scale and the volumetric WML measurement, monotonic functions were fitted using linear regression with a correction factor based on the local variance [8].

For the whole sample of 672 subjects (mean: 72.7, SD: 0.7 years; table 3), the median WML volume was 7.70 ml [interquartile range (IQR): 13.35] and the median Fazekas score was 2 (fig. 1).

Table 3

Characteristics of the study sample at the time of scanning

Characteristics of the study sample at the time of scanning
Characteristics of the study sample at the time of scanning
Fig. 1

Histograms showing the distribution of WML volume (×10–3 ml) in the sample for participants with (a) and without (b) radiologically identifiable stroke lesions, and stem plots showing the distribution of WML volumes per total Fazekas score on participants with (c) and without (d) stroke lesions.

Fig. 1

Histograms showing the distribution of WML volume (×10–3 ml) in the sample for participants with (a) and without (b) radiologically identifiable stroke lesions, and stem plots showing the distribution of WML volumes per total Fazekas score on participants with (c) and without (d) stroke lesions.

Close modal

Although all subjects were born in 1936, the scanning took place over 2.5 years and therefore the difference in age (median: 72.75, IQR: 1.2 years) was 0–1,161 days at the time of scanning. Vascular risk factors were common (e.g. 49% had hypertension; table 3).

In the whole sample, Fazekas scores and WML volumes were highly correlated (Spearman’s ρ = 0.78, p < 0.001; fig. 1). The median stroke lesion volume in the 95 subjects with stroke lesions was 0.83 ml (IQR: 2.43) indicating that most of the stroke lesions were small. Including the stroke lesions within the total lesion volume made no difference to the correlation (Spearman’s ρ = 0.78, p < 0.001). Correcting the WML volume for brain volume and ICV did not affect the correlation either (both Spearman’s ρ = 0.78, p < 0.001; table 4).

Table 4

Numerical results of the correlation between quantitative measurements of WML and Fazekas scores

Numerical results of the correlation between quantitative measurements of WML and Fazekas scores
Numerical results of the correlation between quantitative measurements of WML and Fazekas scores

Separate analysis of the subsample of 95 subjects with stroke lesions showed an almost identical correlation between Fazekas score and WML volume when compared with the 577 subjects with no stroke lesions (Spearman’s ρ = 0.768 and 0.775, respectively, both p < 0.001), even though participants with stroke lesions had higher WML scores and larger volumes than those without stroke lesions. For example, 28.4% of participants with a stroke lesion had a Fazekas score of 4 or 5 versus only 12.3% of those without; only 4.21% of those with stroke lesions had Fazekas total score of 1 versus 16.63% of those without (full details, including separate presentation of periventricular and deep WML scores; see fig. 2). The relationship between Fazekas and WML volume was slightly better described by a quadratic (R2 = 0.65) than by a linear (R2 = 0.61) model or by a LOESS (locally weighted scatterplot smoothing) estimator with Gaussian or Cauchy kernels.

Fig. 2

Proportion of participants with/without stroke lesions with Fazekas (periventricular and deep) scores 0–3. More than 50% of participants with stroke lesions have ‘smooth halo’ (35.8%) and irregular periventricular hyperintensities extending into the deep white matter (15.8%). In participants with stroke lesions, the proportion of deep white matter focal lesions that are beginning to coalesce and large confluent areas is approximately double the proportion in participants without stroke lesions (31.6 and 5.3% vs. 16.1 and 2.6%, respectively).

Fig. 2

Proportion of participants with/without stroke lesions with Fazekas (periventricular and deep) scores 0–3. More than 50% of participants with stroke lesions have ‘smooth halo’ (35.8%) and irregular periventricular hyperintensities extending into the deep white matter (15.8%). In participants with stroke lesions, the proportion of deep white matter focal lesions that are beginning to coalesce and large confluent areas is approximately double the proportion in participants without stroke lesions (31.6 and 5.3% vs. 16.1 and 2.6%, respectively).

Close modal

Despite the high correlation, in 94 subjects (14% of the whole sample), the z-scores of WML volume and Fazekas ratings differed by more than 1 (fig. 3). The proportion of subjects where visual ratings and volumetric measurements did not concur (z-score difference >1), with and without stroke lesions, was similar (12.6 and 14.2%, respectively). Most of the 94 non-concurrent subjects had low total Fazekas scores of either 1 (n = 20, WML volume 0–14.8 ml) or 2 (n = 50, WML volume 0–34.37 ml; table 5).

Table 5

Subjects where visual ratings and volumetric measurements did not concur (z-score difference >1), tabulated per total Fazekas score

Subjects where visual ratings and volumetric measurements did not concur (z-score difference >1), tabulated per total Fazekas score
Subjects where visual ratings and volumetric measurements did not concur (z-score difference >1), tabulated per total Fazekas score
Fig. 3

Box-plot of the difference between the z-scores of the volumetric measurements and the z-scores of the Fazekas ratings. The descriptive statistics of the distribution are: mean value: –0.0047, median: 0.034, variance: 0.50 and SD: 0.71.

Fig. 3

Box-plot of the difference between the z-scores of the volumetric measurements and the z-scores of the Fazekas ratings. The descriptive statistics of the distribution are: mean value: –0.0047, median: 0.034, variance: 0.50 and SD: 0.71.

Close modal

The main reasons for disagreements were: subtle WML seen visually but not detected by the WML volume method (fig. 4a); prominent periventricular caps with thin ventricular body lining producing a Fazekas periventricular score of 1 or 2 but a large WML volume, and small deep white matter focal lesions which score 1 without coalescence and 2 with coalescence, although the actual volume may differ little at this boundary between those scoring 1 and 2 and the range is wide (<0.5 to >12 ml; fig. 4b).

Fig. 4

a Two examples showing the flaws of both qualitative and quantitative methods: in the upper row a subject with a total Fazekas score of 1 has big caps, thin lining around the ventricles and one tiny focal lesion considered too trivial to rate; while in the second row a subject rated with total Fazekas score of 3 has significant regions of ‘dirty’ white matter and subtle focal hyperintensities not recognised by the WML segmentation algorithm. b Examples of two subjects with total Fazekas scores of 2, but large difference in the WML volume.

Fig. 4

a Two examples showing the flaws of both qualitative and quantitative methods: in the upper row a subject with a total Fazekas score of 1 has big caps, thin lining around the ventricles and one tiny focal lesion considered too trivial to rate; while in the second row a subject rated with total Fazekas score of 3 has significant regions of ‘dirty’ white matter and subtle focal hyperintensities not recognised by the WML segmentation algorithm. b Examples of two subjects with total Fazekas scores of 2, but large difference in the WML volume.

Close modal

Both the median Fazekas visual score and WML volumes in the 95 patients with stroke lesions were more than double those of the 577 subjects without stroke lesions: median WML volume 15.30 ml (IQR: 11.6) and median Fazekas score 3 in those with stroke lesions, versus 7.17 ml (IQR: 22.16) and median Fazekas score 2 in those without stroke lesions. Both the difference in the Fazekas scores (χ2 = 30.76, p < 0.001) and the difference in WML volume measurements (Mann-Whitney U = 18.29, p < 0.001) between subjects with and without stroke lesions were significant. In the participants with stroke lesions, the volume of the stroke lesions did not correlate with Fazekas ratings (Spearman’s ρ = 0.14, p = 0.18) or with WML volume (Spearman’s ρ = 0.19, p = 0.07; table 6).

Table 6

Correlations of the stroke lesion volume (n = 95)

Correlations of the stroke lesion volume (n = 95)
Correlations of the stroke lesion volume (n = 95)

In this large group of older community-dwelling individuals with a narrow age range and most with some vascular risk factors, we found that visual and volumetric methods of WML assessment are highly correlated. This indicates that both methods similarly agree on the rank order of WML burden and that the choice of WML quantification method in future studies could be determined by the particular characteristics of the population or the circumstances of the data collection, deliberately taking advantage of the strengths of each technique. To our knowledge, this is the first study of more than 100 older subjects with a wide range of WML (from none to severe; table 3) that explores in detail the relationship between a well-known and commonly applied WML visual rating scale and volumetric measurements within a context of limited measurement variability. The reduced age difference among the individuals of this large cohort represents a valuable methodological strength by reducing one source of variance and thus helping to unmask the relative validity of volumetric assessment of WML versus visual assessment. We also accounted for any effect that radiologically detected cortical or subcortical infarcts might have on the observed WML correlation. Minor discordance (14% of cases) occurred where the WML volume measurements missed subtle WML, where prominent ventricular caps (but thin periventricular body lesions) inflated the WML volume (fig. 4a) and at the interface between non-confluent and early confluent lesions where the WML volume might be similar but the Fazekas score could alternate between 1 and 2. This suggests that improvements in WML volume measurement should focus on detecting subtle WML and that rating scores might benefit from greater granularity to rate thick versus thin caps independently of ventricle body lesions and non-confluent versus early confluent lesions.

WML scores and volumes have been compared (table 1) in smaller samples (≤100 subjects) [9,11] and in more than 500 subjects only from the Leukoaraiosis and Dissability (LADIS) Study [8,17]. However LADIS enrolled a large population of subjects, all with WML including many with a high WML load, some with mild cognitive deficits and some with disturbances of gait and balance. The results obtained in [8,9,17] may not be directly applicable to ‘healthier’ older populations. Kapeller et al. [10] assessed the relationship between Fazekas visual rating scale and volumes using the non-parametric Kendall W test and obtained lower correlation than that shown in this study (W = 0.57, p < 0.001) and the inter-rater agreement for Fazekas scale, although good at baseline (average, ĸ = 0.707, SD = 0.05), was poor at follow-up (ĸ = 0.297, SD = 0.05), which may have influenced their results. Van Straaten et al. [8] found large variability in WML volume in patients with high visual scores, but our results do not mirror this finding. For high visual rating scores, namely Fazekas total scores of 5 or 6, visual ratings and volumetric measurements concurred in all subjects (z-score difference >1; table 3). This might be due to the relatively small number of individuals from our cohort that fell into these categories, 4 and 2%, respectively, although the absolute number was respectable at 41. The most common scores in our sample were 1, 2 and 3; therefore, the disagreement between the volumetric measurement and Fazekas scale might not be directly related to the lesion load but more closely aligned to another associated feature, for example to increased influence of stroke lesions which could be more common at higher WML loads and which were not excluded from the volume measurement (as far as we can tell) in the study of van Straaten et al. [8]. In other respects our sample is consistent with the pattern of agreement seen by others, for example the relationship between Fazekas and WML volume was best described by a quadratic model, mirroring the result obtained by van Straaten et al. [8] on participants of the LADIS Study (R2 = 0.62).

The strengths of the present study are the analysis of the distribution and characteristics of the WML burden in an ageing population with a narrow age range (SD = 0.7 years), a fifth of whom had a radiologically apparent infarct, and also our examination of the correlations between qualitative and quantitative WML load measures separately for participants with radiologically identifiable stroke lesions from those without. The quantitative method we used to calculate WML volumes had very high intra- as well inter-observer reliability and performed better than thresholding on one sequence than other multispectral methods [7,16]. This fact, together with the double-revised visual rating, reduces the influence that the variability in the quality of the assessment can produce in the measured correlations, and may explain the higher values we obtained compared to those reported in other studies [8,11].

The study also has limitations. In our large cohort of ageing individuals there were few subjects with a high visual rating score (6.25% with Fazekas periventricular score of 3 and 2.98% with Fazekas deep score of 3), so our results are not conclusive on how the qualitative and quantitative methods compare at the severe end of the WML spectrum. The quantitative analysis did not consider the division of the WML into periventricular and deep types. However, to our knowledge few studies have used semi-automated methods to segment periventricular from deep WML. Those that did used different criteria to dichotomise the WML volumes, all of which are experimental. Wen and Sachdev [18], for example, considered all discrete lesions as deep WML, assumed a 3D shape of a non-connected hyperintense cluster as a sphere, and classified each cluster as punctuate (<3 mm, 3–10 voxels), focal (3–12 mm, 11–115 voxels) and large (>12 mm, over 115 voxels). DeCarli et al. [19] used non-linear warping to determine the exact distance between each WML voxel and the ventricular ependymal surface and created a periventricular-deep division on the basis of a 1-cm distance from the ventricular system. Godin et al. [20] calculated the Euclidean distance of each WML to the ventricular system and labelled as periventricular those within a distance of 1 mm, but did not provide any validation of their approach or further details. None of these assumptions considered how to delimit the coalescent lesions that extend from the periventricular surface into the deep white matter, while these two categorical distinctions may be somewhat arbitrary. The standardisation of the definition of the boundaries between periventricular and deep WML based on anatomical definitions will be necessary to determine etiological or functional correlates of WML distribution before validated computational segmentation techniques capable of addressing these issues can be developed.

What are the implications for future research? In a population-based setting, when evaluating risk factors for WML, it is acceptable to combine studies that have used volumetric measurements of WML with studies that used Fazekas visual rating scale, provided that appropriate statistics are used. In our healthy ageing cohort, radiologically visible stroke lesions were associated with a greater WML load – with quantitative analysis the infarcts have to be manually separated from the WML volume. In contrast, during visual rating the observer ‘automatically’ ignores the stroke lesions, so the Fazekas visual rating scale is not contaminated by stroke lesions. Therefore, qualitative ratings should be used in studies of WML in patients with stroke, or in subjects at risk of developing stroke during follow-up, whether or not quantitative WML volume measurement is also performed, and if volumes are measured, then the stroke lesions have to be manually separated to avoid undue bias on WML volume measurements. Methods are needed to improve computational WML quantification, for example adaptive filtering, to measure subtle WML areas. Consideration should also be given to modifying WML scores to differentiate ventricular caps from other periventricular lesions, and small but early coalescent from larger deep WML.

We thank the LBC1936 participants, Catherine Murray for work on recruitment and the nurses, radiographers and other staff at the Brain Research Imaging Centre (http://www.sbirc.ed.ac.uk/): a SINAPSE (Scottish Imaging Network, A Platform for Scientific Excellence) collaboration Centre.

This work was funded by Age UK and the UK Medical Research Council as part of the project ‘Disconnected Mind: LBC1936’ (http://www.disconnectedmind.ed.ac.uk/) (including the Sidney De Haan Award for Vascular Dementia), the Centre for Cognitive Ageing and Cognitive Epidemiology (http://www.ccace.ed.ac.uk/; G0700704/84698) and Row Fogo Charitable Trust, with additional funding from the Biotechnology and Biological Sciences Research Council, the Engineering and Physical Sciences Research Council and the Economic and Social Research Council.

The authors have no conflicts of interest.

1.
Pantoni L, Garcia J: The significance of cerebral white matter abnormalities 100 years after Binswanger’s report: a review. Stroke 1995;26:1293–1301.
2.
Schmidt R, Fazekas F, Kapeller P, et al: White matter changes and cognitive decline in non-demented subjects; in Leys D, Pasquier F, Scheltens P (eds): Stroke and Alzheimer’s Disease. The Hague, Academic Graphics, 1998, vol 9, pp 174–182.
3.
Fazekas F, Barkhof F, Wahlund LO, et al: CT and MRI rating of white matter lesions. Cerebrovasc Dis 2002;13:31–36.
4.
Wardlaw JM, Bastin ME, Valdés Hernández MC, et al: Brain ageing, cognition in youth and old age, and vascular disease in the Lothian Birth Cohort 1936: rationale, design and methodology of the imaging protocol. Int J Stroke 2011;6:547–549.
5.
Mäntylä R, Erkinjuntti T, Salonen O, et al: Variable agreement between visual rating scales for white matter hyperintensities on MRI: comparison of 13 rating scales in a poststroke cohort. Stroke 1997;28:1614–1623.
6.
Guillermaud R, Brady M: A dual approach for minimising false lesion classifications on magnetic resonance images. Magn Reson Med 1997;37:94–102.
7.
Valdés Hernández MC, Ferguson KJ, Chappell FM, et al: New multispectral MRI data fusion technique for white matter lesion segmentation: method and comparison with thresholding in FLAIR images. Eur Radiol 2010;20:1684–1691.
8.
van Straaten ECW, Fazekas F, Rostrup E, et al: Impact of white matter hyperintensities scoring method on correlations with clinical data – the LADIS study. Stroke 2006;3:836–840.
9.
Gouw AA, van der Flier WM, van Straaten ECW, et al: Reliability and sensitivity of visual scales versus volumetry for evaluating white matter hyperintensity progression. Cerebrovasc Dis 2008;25:247–253.
10.
Kapeller P, Barber R, Vermeulen RJ, et al: Visual rating of age-related white matter changes on magnetic resonance imaging: scale comparison, interrater agreement, and correlations with quantitative measurements. Stroke 2003;34:441–445.
11.
van den Heuvel DMJ, ten Dam VH, de Craen AJM, et al: Measuring longitudinal white matter changes: comparison of a visual rating scale with a volumetric measurement. AJNR Am J Neuroradiol 2006;27:875–878.
12.
Enzinger C, Fazekas F, Ropele S, et al: Progression of cerebral white matter lesions – clinical and radiological considerations. J Neurol Sci 2007;257:5–10.
13.
Wardlaw JM, Ferguson KJ, Graham C: White matter hyperintensities and rating scales-observer reliability varies with lesion load. J Neurol 2004;251:584–590.
14.
Fazekas F, Niederkorn K, Schmidt R, et al: White matter signal abnormalities in normal individuals: correlation with carotid ultrasonography, cerebral blood flow measurements, and cerebrovascular risk factors. Stroke 1988;19:1285–1288.
15.
Deary IJ, Gow AJ, Taylor MD, et al: The Lothian Birth Cohort 1936: a study to examine influences on cognitive ageing from age 11 to age 70 and beyond. BMC Geriatr 2007;7:28.
16.
Valdés Hernández MC, Gallacher PJ, Bastin ME, et al: Automatic segmentation of brain white matter and white matter lesions in normal aging: comparison of five multispectral techniques. Magn Reson Imaging 2012;30:222–229.
17.
Gouw AA, van der Flier WM, van Straaten ECW, et al: Simple versus complex assessment of white matter hyperintensities in relation to physical performance and cognition: the LADIS study. J Neurol 2006;253:1189–1196.
18.
Wen W, Sachdev P: The topography of white matter hyperintensities on brain MRI in healthy 60- to 64-year-old individuals. Neuroimage 2004;22:144–154.
19.
DeCarli C, Fletcher E, Ramey V, et al: Anatomical mapping of white matter hyperintensities (WMH): exploring the relationships between periventricular WMH, deep WMH, and total WMH burden. Stroke 2005;36:50–55.
20.
Godin O, Tzourio C, Maillard P, et al: Antihypertensive treatment and change in blood pressure are associated with the progression of white matter lesion volumes: the Three City (3C)-Dijon Magnetic Resonance Imaging Study. Circulation 2011;123:266–273.
Copyright / Drug Dosage / Disclaimer
Copyright: All rights reserved. No part of this publication may be translated into other languages, reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, microcopying, or by any information storage and retrieval system, without permission in writing from the publisher.
Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug.
Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.