Claerhoudt Et Al-2012-Equine Veterinary Journal
Content
bs_bs_banner
Equine Veterinary Journal ISSN 0425-1644
DOI: 10.1111/j.2042-3306.2012.00547.x
Differences in the morphology of distal border synovial
invaginations of the distal sesamoid bone in the horse as evaluated
by computed tomography compared with radiography
S. CLAERHOUDT†, H. J. BERGMAN‡, H. VAN DER VEEN‡, L. DUCHATEAU§, E. V. RAES† and J. H. SAUNDERS*†
†
Veterinary Medical Imaging and Small Animal Orthopaedics, Faculty of Veterinary Medicine, Ghent University, Merelbeke, Belgium
Lingehoeve Diergeneeskunde/VetCT, Lienden, The Netherlands
§
Department of Physiology and Biometry, Faculty of Veterinary Medicine, Ghent University, Merelbeke, Belgium.
‡
*Correspondence email: [email protected]; Received: 04.05.11; Accepted: 17.12.11
Summary
Reasons for performing study: Distal border synovial invaginations of the distal sesamoid bone are radiographically assessed during the selection process
of horses admitted as breeding stallions or in purchase examinations. Nowadays, many moderately or some deeply penetrating proximally enlarged synovial
invaginations are considered as moderate or severe radiographic findings.
Objective: To measure the difference between and agreement of the morphology of distal border synovial invaginations on radiography vs. computed
tomography (CT). It was hypothesised that the morphology of distal border synovial invaginations would be better evaluable on CT compared
with radiography.
Methods: Computed tomography scans and 3 dorsoproximal–palmarodistal oblique (DPr-PaDiO) radiographs were obtained on 50 cadaver forefeet
from 25 Warmblood horses. Computed tomography was assumed to be the gold standard. The number, shape and depth of penetration of distal
border synovial invaginations into the distal sesamoid bone were evaluated with both methods, and the comparison of their measurements was
statistically described.
Results: A statistically significant mean difference for number of distal synovial invaginations between CT and all 3 DPr-PaDiO projections was found and was
approximately equal to 2, meaning that CT permits visualisation of an average of 2 more invaginations than radiography. In none of the cases did radiography
have a higher number observed than CT. A large variation in the difference of measurements for depth of penetration against their mean difference between
CT and the 3 radiographic projections was seen. Radiography underestimated the depth of invaginations, and more so when these were deeper. There was
no statistically significant mean difference found between the techniques for depth. A moderate to good agreement between measurements on CT and the
three DPr-PaDiO projections for shape was seen, in which the D55°Pr-PaDiO projection showed the best agreement. A high specificity (90–99%) and low
sensitivity (65%) for all projections for shape were found.
Conclusions and potential relevance: Radiography differs considerably from CT concerning the morphology of distal navicular border synovial
invaginations. For the evaluation of the number, depth and shape of distal synovial invaginations in the distal sesamoid bone, radiography shows only
partially the morphology seen on CT.
Keywords: horse; distal sesamoid bone; Warmblood horse; synovial invagination; radiography; computed tomography
Abbreviations
CT: Computed tomography
DPr-PaDiO: Dorsoproximal-palmarodistal oblique
DSB: Distal sesamoid bone
Introduction
The terms podotrochleosis, navicular syndrome or navicular disease
denote a chronic, progressive uni- or bilateral forelimb lameness typically
affecting riding horses of middle age [1,2]. Radiographic evaluation plays an
essential role in the diagnosis of navicular disease; however, radiography is
limited to changes of the bony component of the distal sesamoid bone
(DSB, also called the navicular bone).
During the selection process of horses admitted as breeding stallions
or in purchase examinations, the synovial invaginations of the distal
border of the DSB are often graded according to radiographic
classification systems, using a dorso45°–70°proximal-palmarodistal
oblique (D45°-70°Pr-PaDiO) radiographic projection. Although opinion is
divided, some clinicians consider many moderately or some deeply
penetrating rounded or inverted flask-shaped synovial invaginations as
moderate or severe radiographic findings. These changes have been
suggested by some authors to be associated with joint pain and
lameness, and would impair a horse’s future sport career [3,4].
Consequently, abnormal synovial invaginations are responsible for
negative advice in purchase examinations and may therefore have major
financial consequences for horse owners. Although many authors
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
describe these findings related to navicular disease, there is still
discussion about their significance [1,5–8].
Computed tomography (CT) is the most appropriate imaging modality for
detailed imaging of normal bone and detection of bony disorders [9–11].
This imaging modality was assumed by the authors to be the gold standard
for evaluation of the DSB in the present study.
We hypothesised that the morphology of distal border synovial
invaginations of the DSB would be better evaluable on CT than on
radiography. The purpose of the study was to demonstrate this hypothesis
by measuring the difference between and agreement of the morphology of
synovial invaginations of the DSB on radiography vs. CT.
Materials and methods
Material
The material used in this study consisted of 50 forefeet of 25 Warmblood
horses (mean age 7 years). All horses were subjected to euthanasia for
reasons unrelated to this study. All feet were severed at the level of the
fetlock joint immediately after euthanasia. The shoe and loose horn in the
sole, if present, were removed, and the frog was cleaned. The feet were
not selected by any particular criteria, and both forefeet of each horse
were included.
Computed tomography examination
The CT scans were performed with a 4-detector row spiral CT scannera, in
which the feet were placed in the gantry with the longitudinal axis of the
679
Evaluation of distal navicular border synovial invaginations
foot oriented parallel to the CT table and perpendicular to the plane of the
CT gantry. The medial side of the foot was marked. The limbs were scanned
in a distal-to-proximal direction. The output parameters were 120 kV and
250 mA s per slice. The slice thickness was 0.6 mm, pitch of 0.875 cm, 0.3
increment and 1 s rotation time. Transverse CT scans were reconstructed
from the level of the distal aspect to the level of the proximal aspect of the
DSB using a bone window setting (window level, 200–600 H.U.; window
width, 1000–2000 H.U.), 250 mm field of view and 512 ¥ 512 pixel matrix.
The average total time required for scanning of each foot was 46.25 s. From
the transverse images, sagittal and dorsal reconstructions, with a slice
thickness of 0.6 mm, were reformatted using of softwareb.
Radiographic examination
Radiographic examination was performed after the CT examination. The
sulci of the frog were packed with modelling compoundc. Three
dorsoproximal-palmarodistal oblique (DPr-PaDiO) radiographic projections
with different hoof angles (D45°Pr-PaDiO, D55°Pr-PaDiO and D65°Pr-PaDiO)
were performed on all feet. The x-ray beam of the tubed was kept horizontal
and centred 2 cm proximal to the coronary band at the midline of the foot.
The foot was placed on a wooden block with a slope of 45° to the horizontal.
Using wedges (slope 5°), D55°Pr-PaDiO and D65°Pr-PaDiO projections were
made of all feet. The feet were radiographed using 60 kV and 12.5 mA s, a
grid (6:1 ratio, 103 lines/cm) and a 100 cm focus–film distance.
S. Claerhoudt et al.
TABLE 1: Difference between computed tomographic (CT) and 3
radiographic projections for number and depth
Comparison for number
CT vs. D45°Pr-PaDiO
CT vs. D55°Pr-PaDiO
CT vs. D65°Pr-PaDiO
Comparison for depth
CT vs. D45°Pr-PaDiO
CT vs. D55°Pr-PaDiO
CT vs. D65°Pr-PaDiO
Mean
difference
95% Reference
interval
95% Confidence
interval
1.88
2.04
2.20
-1.18 to 4.94
-0.95 to 5.03
-0.66 to 5.06
1.45–2.31
1.62–2.46
1.80–2.60
0.003
0.016
0.012
-0.28 to 0.29
-0.29 to 0.32
-0.28 to 0.30
-0.02 to 0.02
-0.01 to 0.04
-0.01 to 0.03
comparisonwise significance level equal to 0.0125 (Bonferroni
adjustment).
The degree of agreement between CT and the DPr-PaDiO projections for
number and depth (taken as categorical variables) was quantified using the
weighted k statistic. The degree of agreement for shape was quantified
using the unweighted k statistic. The guidelines for strength of agreement
based on the values of k were as follows: <0.20 poor, 0.21–0.40 fair,
0.41–0.60 moderate, 0.61–0.80 good and 0.81–1.00 very good [12].
Image analysis
Two observers, one board-certified radiologist (J.H.S.) and one PhD student
(S.C.), interpreted all images together and a diagnosis was made by
consensus. The radiographic images of a particular foot and hoof angle
were reviewed in a randomised order at the same workstation, on the same
diagnostic imaging screense and using a similar evaluationf, to determine
the number of distal border synovial invaginations. Furthermore, for each
synovial invagination, the depth and shape were determined. Next, the CT
images of a particular foot were reviewed in a random order to determine
the number of distal border synovial invaginations using transverse slices
and dorsal reconstructions, and for each synovial invagination the depth
and shape were determined.
In a second step, the corresponding radiographic and CT images were
considered together. To compare depth and shape assessments on the 2
imaging modalities, only synovial invaginations for which an assessment
was available on both radiography and CT were used (some invaginations
seen with CT were not seen with radiography).
The depth of penetration of the synovial invaginations was assessed on
the dorsal CT and radiographic images. Each synovial invagination was
calculated by an imaging software programb, using the following equation:
depth (R) = A/B, where A is the distance (in centimetres) between the most
distal basis and the proximal top of the synovial invagination and B the
distance (in centimetres) between the distal and proximal flexor borders
of the DSB. Data on depth were classified into the following 3 categories:
1 = R ⱕ 0.33, 2 = 0.33<R ⱕ 0.5 and 3 = R>0.5.
The shape of the synovial invaginations was assessed on dorsal CT and
radiographic images. The shape could be categorised as ‘conical’, ‘linear’,
‘lollipop’ or ‘branched’ (4 categories, further described as shape4), with
1 = normal (conical or linear shaped) and 2 = abnormal (lollipop or branched
shaped; further described as shape2).
Results
The differences and variability between measurements on CT and the
different radiographic projections for number and depth of distal border
synovial invaginations are summarised in Table 1.
Number
The average total number of synovial invaginations was 5.9 ⫾ 1.56 on CT, 4
⫾ 1.85 on the D45°Pr-PaDiO, 3.9 ⫾ 1.71 on the D55°Pr-PaDiO and 3.7 ⫾
1.81 on the D65°Pr-PaDiO projection. In none of the cases did radiography
have a higher number observed than CT (Fig 1). In only 11 of 50 (22%), 7 of
50 (14%) and 6 of 50 feet (12%) were the number of synovial invaginations
counted on CT scans and on the D45°Pr-PaDiO, D55°Pr-PaDiO and
D65°Pr-PaDiO projections, respectively, equal. In 2 of 50 (4%) feet, no
synovial invaginations were detected on the 3 radiographic projections;
however, 3 and 4 invaginations, respectively, were counted on CT. A
statistically significant mean difference between CT and all 3 DPr-PaDiO
projections was found (P<0.001) and was approximately equal to 2,
meaning that CT permits visualisation of an average of 2 more invaginations
Data analysis
To compare the observed number of invaginations and the depth of the
invagination between CT and the DPr-PaDiO projections, Student’s paired
t test was used with foot as block variable for the number and invagination
as block factor for the depth. The results were summarised by the average
difference and corresponding 95% confidence interval (CI) and 95%
reference interval. The 95% reference interval is given by the mean
difference ⫾ 2 s.d., which contains 95% of the actual differences if the
normal distribution assumption holds. Bland-Altman plots are provided
(Figs S1 and S2) for CT vs. the different DPr-PaDiO projections, to
investigate a possible relationship between the difference and the
magnitude of the measurement. A global significance level of 0.05 was
used, but each of the 3 pairwise comparisons was tested at a
680
Fig 1: Dorsally reconstructed computed tomographic (CT) image (a) and 55°
dorsoproximal-palmarodistal oblique (D55°Pr-PaDiO) radiographic image (b) of a distal
sesamoid bone, showing 7 and 4 distal border synovial invaginations, respectively.
Lateral is on the left side and medial on the right side of the image.
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
Evaluation of distal navicular border synovial invaginations
S. Claerhoudt et al.
TABLE 2: Agreement between computed tomographic and 3
radiographic projections for number, depth and shape (*: only
unweighted k was calculated for shape2)
Radiographic projection
Categorical variable
k
Weighted k
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
Number
Number
Number
Depth
Depth
Depth
Shape4
Shape4
Shape4
Shape2
Shape2
Shape2
0.12
0.02
0.01
0.23
0.19
0.23
0.49
0.50
0.42
0.68*
0.7*
0.55*
0.24
0.17
0.15
0.30
0.26
0.29
0.57
0.60
0.49
–
–
–
than radiography. The D45°Pr-PaDiO projection showed the smallest mean
difference. The corresponding 95% reference intervals, as well as the 95%
CIs, were very wide for all 3 comparisons, reflecting a great variation in the
differences (Table 1). The Bland-Altman plots of the differences for number
between the methods against their means are shown in Figure S1.
The weighted k values for all 3 projections were very low, specifically for
the D55°Pr-PaDiO and D65°Pr-PaDiO projections, meaning that a greater
disagreement with CT for number was calculated for these 2 projections
(Table 2).
Depth
The linear measurements of depth of synovial invaginations (measurement
A of the equation) and distance between the distal and proximal flexor
borders (measurement B) increased with radiographic angle in 25 of 50
(50%) DSBs. There were no significant mean differences for depth between
CT and the 3 radiographic projections, and the confidence intervals were
also narrow, meaning that there seems to be little or no bias. The mean
differences of measurements for depth between CT and the 3 radiographic
projections were all positive and small, with the D45°Pr-PaDiO projection
showing the smallest mean difference. By evaluating reference intervals, it
is possible to see how precise the individual estimates are. Consequently,
the reference intervals for all 3 DPr-PaDiO projections showed quite wide,
comparable ranges (Table 1). Ninety-five per cent of differences between
CT and the different DPr-PaDiO projections for depth lie between the limits
-0.28 and 0.32 of the interval (0.28 below or 0.32 above zero-level). Also, the
Bland-Altman plots showed a large variation in the differences against their
means (see Fig S2). It is illustrated that as the mean measurements of CT
and radiography increase (larger equations; deeper invaginations), the
differences (measurements CT minus radiography) also seem to increase,
meaning that the underestimation by radiography was greater in the case
of deeper invaginations (Fig 2).
The k statistics also revealed quite poor agreement between the
techniques (average k value of 0.21 and low weights to disagreements, as
summarised in Table 2).
Fig 2: Dorsally reconstructed CT image (a) and corresponding D65°Pr-PaDiO
radiographic image (b) of a distal sesamoid bone, showing a deeply penetrating
invagination on the CT image (arrow), which is mildly penetrating on the radiographic
image (arrow). Lateral is on the left side and medial on the right side of the image. An
example of a false-negative result is shown in Fig 3.
shape2) and corresponding 95% CIs for all 3 DPr-PaDiO projections for
shape, with CT as gold standard, are summarised in Table 3.
Discussion
The morphological features of the distal navicular border synovial
invaginations of the present population were described in a previous study
[13] using CT. The present investigation was carried out to assess the
variability and agreement between the appearance of distal border
synovial invaginations of the DSB on radiographs vs. CT examinations.
Histology is regarded as the gold standard for the diagnosis of tissue
abnormalities [14]. In the present study, no histological examination was
Shape
In only 3 of 11 (27%), 4 of 7 (57%) and 3 of 6 (50%) feet with an equal number of
synovial invaginations on both the CT scans and D45°Pr-PaDiO,
D55°Pr-PaDiO and D65°Pr-PaDiO projections, respectively, was the shape
comparable by both methods. The agreement for shape4 ranged between
0.42 ⱕ k ⱕ 0.50, representing a moderate agreement, whereas the k value
for shape2 was higher (range 0.55 ⱕ k ⱕ 0.71), representing a moderate to
good agreement. Weighted k for shape was higher for the D55°Pr-PaDiO
projection, meaning that, relative to the other projections, a better
agreement with the gold standard for shape was calculated for
this projection.
The k and weighted k values for agreement of number, depth and shape
are presented in Table 2. The sensitivity, specificity (both calculated for
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
Fig 3: Dorsally reconstructed CT image (a) and corresponding D55°Pr-PaDiO
radiographic image (b) of a distal sesamoid bone, showing 2 abnormally shaped
invaginations on the CT image (arrows), which were normally shaped on the
radiographic image (arrows). Lateral is on the left side and medial on the right side of
the image.
681
Evaluation of distal navicular border synovial invaginations
TABLE 3: Sensitivity and specificity for all 3 radiographic projections
for shape with computed tomography as gold standard
Radiographic
projection
Sensitivity
(95% confidence
interval)
Specificity
(95% confidence
interval)
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
0.65 (0.51–0.78)
0.65 (0.51–0.78)
0.63 (0.48–0.76)
0.97 (0.92–0.99)
0.99 (0.95–1)
0.90 (0.84–0.95)
performed; however, owing to the possibility of reconstruction of
multiplanar, high-resolution images without superimposition, CT provides
detailed anatomical information on the synovial invaginations [11].
Therefore, CT was assumed by the authors to be the gold standard.
The total number of synovial invaginations counted on CT was much
higher than with radiography. The differences between CT and the 3
DPr-PaDiO projections for the number of synovial invaginations were all
statistically positive, with the D45°Pr-PaDiO projection showing the
smallest mean difference. In fact, CT permitted visualisation of an average
of 2 more invaginations per navicular bone than with radiography,
indicating a better visibility of invaginations on CT than with radiography.
Similar conclusions were reported earlier [10,11]. The synovial
invaginations that were ill defined or undetectable on radiography were
mostly small and located at the distal sloping borders of the DSB on the CT
images. It has been described that the presence of small synovial
invaginations at the sloping borders on CT may be observed in normal
DSBs, owing to lack of superimposition of surrounding bone and the high
sensitivity of CT for detecting bone in detail [11]. However, the clinical
significance of these subtle CT findings remains questionable.
In approximately 80% of the cases in our study, a DSB with 7 invaginations
on radiography had more than 7 distal border invaginations on CT.
According to the literature, up to 7 radiographically detectable distal
border synovial invaginations are considered normal, and more than 7
significant [15,16], although an overlap between sound and lame horses is
described [5]. However, the clinical relevance of this increased number of
invaginations on CT remains unclear, and further research is required.
The present results show that when the mean measurements for depth
on CT and radiography increase, the differences (measurements CT minus
radiography) also seem to increase, resulting in a larger underestimation by
radiography in case of deeper invaginations. As described above, CT
appeared better in evaluating synovial invaginations than radiography,
owing to the better visibility of subtle changes on CT [9–11]. On CT, most
deep invaginations ended proximally as very tiny, deeply penetrating lines,
which were undetectable on radiography. Therefore, care should be taken
in judging the depth of distal border invaginations on radiographs during
purchase examinations, because deeply penetrating invaginations may be
missed on DPr-PaDiO projections. It is reported that deeply penetrating
distal border invaginations are significant radiographic findings that could
be responsible for future joint pain and lameness [3,4]; however, the clinical
significance of subtle CT findings remains unclear. Further investigation
with clinical association is necessary to determine the importance of these
deeply penetrating, tiny synovial invaginations.
In the present study, the sensitivity and specificity for shape of the distal
navicular border synovial invaginations were calculated on the 3 DPr-PaDiO
projections using CT as the gold standard. For radiography as diagnostic
modality, both a high sensitivity (i.e. high number of correctly identified
abnormal shaped invaginations) and a high specificity (i.e. correctly
identified absence of abnormally shaped invaginations) are desired.
False-positive diagnosis (poor specificity) of abnormally shaped
invaginations can have major consequences, because the literature
describes these findings to be related to navicular disease [1,5–8].
However, our results show an almost 100% specificity for all projections for
shape, meaning that almost no false-positive results were seen. On the
contrary, more false-negative results were present, resulting in a much
lower sensitivity of 65% for shape. In other words, an abnormal invagination
on radiography is effectively abnormal, but a normal one can in fact
be abnormal.
The present results show a variable degree of agreement (k values)
between measurements on CT and the DPr-PaDiO projections for the 3
682
S. Claerhoudt et al.
variables. All data for shape were grouped into 2 categories (2 ¥ 2 table) for
shape2 and 4 categories (4 ¥ 4 table) for shape4, resulting in variable values
of k. For the resulting 2 ¥ 2 table a better average k value of 0.65 was found,
compared with k = 0.47 for the 4 ¥ 4 table. In contrast, data on numbers
were classified into 12 categories (0 being the lowest number of
invaginations found and 11 the highest), logically resulting in very low
k values. In theory, any value of k much below 0.5 will indicate poor
agreement. However, despite these published guidelines [12], no value of k
can be regarded universally as indicating some degree of agreement [17].
In fact, the value of k depends on the number of categories and upon
circumstances, as demonstrated in our results. For multiple categories on
an ordinal scale, weighted k has the advantage that it ‘weights’ the degree
of disagreements. Greater disagreement is penalised more, resulting in
lower weighted k values.
The results of the present study showed that the variability of the actual
differences between CT and radiography is high for number and depth of
penetration of distal navicular border synovial invaginations. This can be
explained by 3 factors. A first factor is the variable orientation of the
synovial invaginations in the DSB. Indeed, a recent study has demonstrated
that the orientation of the distal border synovial invaginations into the DSB
can vary from a straight, dorsoproximal to a palmaroproximal direction
[13]. A second factor is the variable height of the heels. Pearce et al. [18]
demonstrated that the degree of distal interphalangeal joint angulation
increases (increased joint flexion) with heel elevation, and van Dixhoorn
et al. [19] reported that the DSB follows the coffin bone in vitro during distal
interphalangeal joint flexion. Thus, elevation of the heels results in at least
an increased upright motion of the DSB in the sagittal plane. Finally, a third
factor is the superimposition of the DSB over other structures on a
DPr-PaDiO projection, preventing visualisation of the exact point of origin of
the synovial invaginations [10]. The distal contour of the DSB on DPr-PaDiO
projections is visualised as 2 lines, one representing the articular border
and the second the flexor border, with the distal border synovial
invaginations being situated in the groove between these borders [16,20].
A consequence of the variable orientation of the invaginations into the
bone and of the large individual variation of heel height is that the depth of
the distal navicular border synovial invaginations relative to the horizontal
x-ray beam varies individually when the front of the hoof wall is angled
forward at approximately 45, 55 and 65°. In fact, when elevating the heels,
the depth of dorso- and palmaroproximal oriented synovial invaginations
on DPr-PaDiO radiographic projections respectively shortens and enlarges
relative to the degree of heel elevation. The opposite effects are obtained
by lowering the heels.
A potential limitation of this study could be the absence of a
palmaroproximal-palmarodistal oblique projection, which permits
evaluation of the DSB without superimposition. However, this projection
only allows evaluation of the number and width of the distal border
invaginations [5]. Therefore, it was not included in our study.
In conclusion, the results of the present study indicate that radiography
differs significantly from CT concerning the morphology of distal navicular
border synovial invaginations. CT is a much more sensitive method for the
detection of these invaginations; therefore, regardless of the exact clinical
meaning of the synovial invaginations, the criteria as used for radiography
can absolutely not be used unchanged in case of CT examination.
Prospective epidemiological studies are necessary to assess the clinical
significance of CT-detected abnormalities in this area.
Authors’ declaration of interests
No conflicts of interest have been declared.
Source of funding
This study was financed by the ‘Bijzonder Onderzoeksfonds’, Ghent
University.
Acknowledgements
The authors would like to thank Kim Claus and Marnix Verdonck for
technical assistance. We thank Katrien Vanderperren for constructive
criticism of the manuscript.
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
S. Claerhoudt et al.
Manufacturers’ addresses
a
Mx8000, Philips Medical Systems, AE Eindhoven, The Netherlands.
Osirix Image processing Software, Geneva, Switzerland.
c
Playdoh®: Rainbow Crafts, Cincinnati, Ohio, USA.
d
Mobilux, X-ray Equipment Verachtert, Antwerpen, Belgium.
e
Totoku monochrome LCD display, Lewisville, Texas, USA.
f
Microsoft Excel, Microsoft Corp., Redmond, Washington, USA.
b
References
1. Wright, I.M. (1993) A study of 118 cases of navicular disease: radiological
features. Equine Vet. J. 25, 493-500.
2. Dyson, S., Murray, R., Blunden, T. and Schramme, M. (2006) Current concepts
of navicular disease. Equine Vet. Educ. 18, 45-56.
3. Robert, C., Valette, J.P. and Denoix, J.M. (2006) Correlation between routine
radiographic findings and early racing career in French Trotters. Equine Vet. J.,
Suppl. 36, 473-478.
4. Dik, K.J. (2007) Diagnostische beeldvorming. In: De Veterinaire Keuring Van Het
Paard, 3rd edn., Eds: M.M. Sloet van Oldruitenborgh-Oosterbaan, A.
Barneveld, A.J. Van Den Belt, B.V. Libre, Leeuwarden. pp 85-112.
5. Rose, R.J., Taylor, B.J. and Steel, J.D. (1978) Navicular disease in the horse: an
analysis of seventy cases and assessment of a special radiographic view.
J. Equine Med. Surg. 2, 492-497.
6. MacGregor, C.M. (1986) Radiographic assessment of navicular bones, based
on changes in the distal nutrient foramina. Equine Vet. J. 18, 203-206.
7. Hertsch, B. and Dammer, H. (1988) The blood supply of normal and diseased
navicular bones. Vet. Rad. 29, 276-281.
8. Kaser-Hotz, B. and Ueltschi, G. (1992) Radiographic appearance of the
navicular bone in sound horses. Vet. Radiol. Ultrasound 33, 9-17.
9. Tietje, S. (1995) Computed tomography of the navicular bone region in the
horse: a comparison with radiographic documentation. Pferdeheilkunde 11,
51-61.
10. Widmer, W.R., Buckwalter, K.A., Fessler, J.F., Hill, M.A., VanSickle, D.C. and
Ivancevich, S. (2000) Use of radiography, computed tomography and magnetic
resonance imaging for evaluation of navicular syndrome in the horse. Vet.
Radiol. Ultrasound 41, 108-116.
11. Groth, A.M., May, S.A., Weaver, M.P. and Weller, R. (2009) Intra- and
interobserver agreement in the interpretation of navicular bones on
radiographs and computed tomography scans. Equine Vet. J. 41, 124-129.
12. Landis, J.R. and Koch, G.G. (1977) The measurement of observer agreement for
categorical data. Biometrics 33, 159-174.
13. Claerhoudt, S., Bergman, E.H.J., van der Veen, H., Vanderperren, K., Raes, E.V.
and Saunders, J.H. (2011) Computed tomographic morphology of the synovial
invaginations of the distal sesamoid bone of the horse. Anat. Histol. Embryol.
40, 55-60.
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
Evaluation of distal navicular border synovial invaginations
14. Murray, R.C., Blunden, T.S., Schramme, M.C. and Dyson, S.J. (2006) How does
magnetic resonance imaging represent histologic findings in the equine digit?
Vet. Radiol. Ultrasound 47, 17-31.
15. Colles, C.M. (1983) Interpreting radiographs 1: the foot. Equine Vet. J. 15,
297-303.
16. Rijkenhuizen, A.B.M., Németh, F., Dik, K.J. and Goedegebuure, S.A. (1989)
Development of the navicular bone in foetal and young horses, including the
arterial supply. Equine Vet. J. 21, 405-412.
17. Altman, D.G. (1991) Some common problems in medical research. In: Practical
Statistics for Medical Research, Ed: D.G. Altman, Chapman & Hall, London. pp
396-439.
18. Pearce, S.G., Boure, L.P., Bolger, A., Thomason, J.J. and Dobson, H. (2004)
Effect of heel elevation on forelimb conformation in horses. Aust. Vet. J. 82,
558-562.
19. van Dixhoorn, I.D.E., Meershoek, L.S., Huiskes, R. and Schamhardt, H.C. (2002)
A discription of the motion of the navicular bone during in vitro vertical loading
of the equine forelimb. Equine Vet. J. 34, 594-597.
20. Dik, K.J., van den Belt, A.J.M., Enzerink, E. and van Weeren, P.R. (2001) The
radiographic development of the distal and proximal double contours of the
equine navicular bone on dorsoproximal-palmarodistal oblique (upright pedal)
radiographs, from age 1 to 11 months. Equine Vet. J. 33, 70-74.
Supporting Information
Additional Supporting Information may be found in the online version of
this article:
Fig S1: Bland–Altman plots: differences of counted numbers (readings) on
computed tomographic and radiographic images (A: D45°Pr-PaDiO; B:
D55°Pr-PaDiO and C: D65°Pr-PaDiO) against their means. The numerical
code represents the number of distal sesamoid bones (total sum of 50) with
the same difference in number against mean. Dashed line represents
zero-level (no difference), upper line = mean + 2s.d., middle line = mean,
lower line = mean – 2s.d. s.d. = standard deviation; 95%RI = 95% reference
interval; 95%CI = 95% confidence interval.
Fig S2: Bland–Altman plots: differences of the equations of depth
(readings) on computed tomographic and radiographic images (A:
D45°Pr-PaDiO; B: D55°Pr-PaDiO and C: D65°Pr-PaDiO) against their means.
Each individual plot (O) represents a synovial invagination (total of 295
plots). Dashed line represents zero-level (no difference), upper line = mean
+ 2s.d., middle line = mean, lower line = mean – 2s.d. s.d. = standard
deviation; 95%RI = 95% reference interval; 95%CI = 95% confidence interval.
Please note: Wiley-Blackwell are not responsible for the content or
functionality of any supporting materials supplied by the authors. Any
queries (other than missing material) should be directed to the
corresponding author for the article.
683
Equine Veterinary Journal ISSN 0425-1644
DOI: 10.1111/j.2042-3306.2012.00547.x
Differences in the morphology of distal border synovial
invaginations of the distal sesamoid bone in the horse as evaluated
by computed tomography compared with radiography
S. CLAERHOUDT†, H. J. BERGMAN‡, H. VAN DER VEEN‡, L. DUCHATEAU§, E. V. RAES† and J. H. SAUNDERS*†
†
Veterinary Medical Imaging and Small Animal Orthopaedics, Faculty of Veterinary Medicine, Ghent University, Merelbeke, Belgium
Lingehoeve Diergeneeskunde/VetCT, Lienden, The Netherlands
§
Department of Physiology and Biometry, Faculty of Veterinary Medicine, Ghent University, Merelbeke, Belgium.
‡
*Correspondence email: [email protected]; Received: 04.05.11; Accepted: 17.12.11
Summary
Reasons for performing study: Distal border synovial invaginations of the distal sesamoid bone are radiographically assessed during the selection process
of horses admitted as breeding stallions or in purchase examinations. Nowadays, many moderately or some deeply penetrating proximally enlarged synovial
invaginations are considered as moderate or severe radiographic findings.
Objective: To measure the difference between and agreement of the morphology of distal border synovial invaginations on radiography vs. computed
tomography (CT). It was hypothesised that the morphology of distal border synovial invaginations would be better evaluable on CT compared
with radiography.
Methods: Computed tomography scans and 3 dorsoproximal–palmarodistal oblique (DPr-PaDiO) radiographs were obtained on 50 cadaver forefeet
from 25 Warmblood horses. Computed tomography was assumed to be the gold standard. The number, shape and depth of penetration of distal
border synovial invaginations into the distal sesamoid bone were evaluated with both methods, and the comparison of their measurements was
statistically described.
Results: A statistically significant mean difference for number of distal synovial invaginations between CT and all 3 DPr-PaDiO projections was found and was
approximately equal to 2, meaning that CT permits visualisation of an average of 2 more invaginations than radiography. In none of the cases did radiography
have a higher number observed than CT. A large variation in the difference of measurements for depth of penetration against their mean difference between
CT and the 3 radiographic projections was seen. Radiography underestimated the depth of invaginations, and more so when these were deeper. There was
no statistically significant mean difference found between the techniques for depth. A moderate to good agreement between measurements on CT and the
three DPr-PaDiO projections for shape was seen, in which the D55°Pr-PaDiO projection showed the best agreement. A high specificity (90–99%) and low
sensitivity (65%) for all projections for shape were found.
Conclusions and potential relevance: Radiography differs considerably from CT concerning the morphology of distal navicular border synovial
invaginations. For the evaluation of the number, depth and shape of distal synovial invaginations in the distal sesamoid bone, radiography shows only
partially the morphology seen on CT.
Keywords: horse; distal sesamoid bone; Warmblood horse; synovial invagination; radiography; computed tomography
Abbreviations
CT: Computed tomography
DPr-PaDiO: Dorsoproximal-palmarodistal oblique
DSB: Distal sesamoid bone
Introduction
The terms podotrochleosis, navicular syndrome or navicular disease
denote a chronic, progressive uni- or bilateral forelimb lameness typically
affecting riding horses of middle age [1,2]. Radiographic evaluation plays an
essential role in the diagnosis of navicular disease; however, radiography is
limited to changes of the bony component of the distal sesamoid bone
(DSB, also called the navicular bone).
During the selection process of horses admitted as breeding stallions
or in purchase examinations, the synovial invaginations of the distal
border of the DSB are often graded according to radiographic
classification systems, using a dorso45°–70°proximal-palmarodistal
oblique (D45°-70°Pr-PaDiO) radiographic projection. Although opinion is
divided, some clinicians consider many moderately or some deeply
penetrating rounded or inverted flask-shaped synovial invaginations as
moderate or severe radiographic findings. These changes have been
suggested by some authors to be associated with joint pain and
lameness, and would impair a horse’s future sport career [3,4].
Consequently, abnormal synovial invaginations are responsible for
negative advice in purchase examinations and may therefore have major
financial consequences for horse owners. Although many authors
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
describe these findings related to navicular disease, there is still
discussion about their significance [1,5–8].
Computed tomography (CT) is the most appropriate imaging modality for
detailed imaging of normal bone and detection of bony disorders [9–11].
This imaging modality was assumed by the authors to be the gold standard
for evaluation of the DSB in the present study.
We hypothesised that the morphology of distal border synovial
invaginations of the DSB would be better evaluable on CT than on
radiography. The purpose of the study was to demonstrate this hypothesis
by measuring the difference between and agreement of the morphology of
synovial invaginations of the DSB on radiography vs. CT.
Materials and methods
Material
The material used in this study consisted of 50 forefeet of 25 Warmblood
horses (mean age 7 years). All horses were subjected to euthanasia for
reasons unrelated to this study. All feet were severed at the level of the
fetlock joint immediately after euthanasia. The shoe and loose horn in the
sole, if present, were removed, and the frog was cleaned. The feet were
not selected by any particular criteria, and both forefeet of each horse
were included.
Computed tomography examination
The CT scans were performed with a 4-detector row spiral CT scannera, in
which the feet were placed in the gantry with the longitudinal axis of the
679
Evaluation of distal navicular border synovial invaginations
foot oriented parallel to the CT table and perpendicular to the plane of the
CT gantry. The medial side of the foot was marked. The limbs were scanned
in a distal-to-proximal direction. The output parameters were 120 kV and
250 mA s per slice. The slice thickness was 0.6 mm, pitch of 0.875 cm, 0.3
increment and 1 s rotation time. Transverse CT scans were reconstructed
from the level of the distal aspect to the level of the proximal aspect of the
DSB using a bone window setting (window level, 200–600 H.U.; window
width, 1000–2000 H.U.), 250 mm field of view and 512 ¥ 512 pixel matrix.
The average total time required for scanning of each foot was 46.25 s. From
the transverse images, sagittal and dorsal reconstructions, with a slice
thickness of 0.6 mm, were reformatted using of softwareb.
Radiographic examination
Radiographic examination was performed after the CT examination. The
sulci of the frog were packed with modelling compoundc. Three
dorsoproximal-palmarodistal oblique (DPr-PaDiO) radiographic projections
with different hoof angles (D45°Pr-PaDiO, D55°Pr-PaDiO and D65°Pr-PaDiO)
were performed on all feet. The x-ray beam of the tubed was kept horizontal
and centred 2 cm proximal to the coronary band at the midline of the foot.
The foot was placed on a wooden block with a slope of 45° to the horizontal.
Using wedges (slope 5°), D55°Pr-PaDiO and D65°Pr-PaDiO projections were
made of all feet. The feet were radiographed using 60 kV and 12.5 mA s, a
grid (6:1 ratio, 103 lines/cm) and a 100 cm focus–film distance.
S. Claerhoudt et al.
TABLE 1: Difference between computed tomographic (CT) and 3
radiographic projections for number and depth
Comparison for number
CT vs. D45°Pr-PaDiO
CT vs. D55°Pr-PaDiO
CT vs. D65°Pr-PaDiO
Comparison for depth
CT vs. D45°Pr-PaDiO
CT vs. D55°Pr-PaDiO
CT vs. D65°Pr-PaDiO
Mean
difference
95% Reference
interval
95% Confidence
interval
1.88
2.04
2.20
-1.18 to 4.94
-0.95 to 5.03
-0.66 to 5.06
1.45–2.31
1.62–2.46
1.80–2.60
0.003
0.016
0.012
-0.28 to 0.29
-0.29 to 0.32
-0.28 to 0.30
-0.02 to 0.02
-0.01 to 0.04
-0.01 to 0.03
comparisonwise significance level equal to 0.0125 (Bonferroni
adjustment).
The degree of agreement between CT and the DPr-PaDiO projections for
number and depth (taken as categorical variables) was quantified using the
weighted k statistic. The degree of agreement for shape was quantified
using the unweighted k statistic. The guidelines for strength of agreement
based on the values of k were as follows: <0.20 poor, 0.21–0.40 fair,
0.41–0.60 moderate, 0.61–0.80 good and 0.81–1.00 very good [12].
Image analysis
Two observers, one board-certified radiologist (J.H.S.) and one PhD student
(S.C.), interpreted all images together and a diagnosis was made by
consensus. The radiographic images of a particular foot and hoof angle
were reviewed in a randomised order at the same workstation, on the same
diagnostic imaging screense and using a similar evaluationf, to determine
the number of distal border synovial invaginations. Furthermore, for each
synovial invagination, the depth and shape were determined. Next, the CT
images of a particular foot were reviewed in a random order to determine
the number of distal border synovial invaginations using transverse slices
and dorsal reconstructions, and for each synovial invagination the depth
and shape were determined.
In a second step, the corresponding radiographic and CT images were
considered together. To compare depth and shape assessments on the 2
imaging modalities, only synovial invaginations for which an assessment
was available on both radiography and CT were used (some invaginations
seen with CT were not seen with radiography).
The depth of penetration of the synovial invaginations was assessed on
the dorsal CT and radiographic images. Each synovial invagination was
calculated by an imaging software programb, using the following equation:
depth (R) = A/B, where A is the distance (in centimetres) between the most
distal basis and the proximal top of the synovial invagination and B the
distance (in centimetres) between the distal and proximal flexor borders
of the DSB. Data on depth were classified into the following 3 categories:
1 = R ⱕ 0.33, 2 = 0.33<R ⱕ 0.5 and 3 = R>0.5.
The shape of the synovial invaginations was assessed on dorsal CT and
radiographic images. The shape could be categorised as ‘conical’, ‘linear’,
‘lollipop’ or ‘branched’ (4 categories, further described as shape4), with
1 = normal (conical or linear shaped) and 2 = abnormal (lollipop or branched
shaped; further described as shape2).
Results
The differences and variability between measurements on CT and the
different radiographic projections for number and depth of distal border
synovial invaginations are summarised in Table 1.
Number
The average total number of synovial invaginations was 5.9 ⫾ 1.56 on CT, 4
⫾ 1.85 on the D45°Pr-PaDiO, 3.9 ⫾ 1.71 on the D55°Pr-PaDiO and 3.7 ⫾
1.81 on the D65°Pr-PaDiO projection. In none of the cases did radiography
have a higher number observed than CT (Fig 1). In only 11 of 50 (22%), 7 of
50 (14%) and 6 of 50 feet (12%) were the number of synovial invaginations
counted on CT scans and on the D45°Pr-PaDiO, D55°Pr-PaDiO and
D65°Pr-PaDiO projections, respectively, equal. In 2 of 50 (4%) feet, no
synovial invaginations were detected on the 3 radiographic projections;
however, 3 and 4 invaginations, respectively, were counted on CT. A
statistically significant mean difference between CT and all 3 DPr-PaDiO
projections was found (P<0.001) and was approximately equal to 2,
meaning that CT permits visualisation of an average of 2 more invaginations
Data analysis
To compare the observed number of invaginations and the depth of the
invagination between CT and the DPr-PaDiO projections, Student’s paired
t test was used with foot as block variable for the number and invagination
as block factor for the depth. The results were summarised by the average
difference and corresponding 95% confidence interval (CI) and 95%
reference interval. The 95% reference interval is given by the mean
difference ⫾ 2 s.d., which contains 95% of the actual differences if the
normal distribution assumption holds. Bland-Altman plots are provided
(Figs S1 and S2) for CT vs. the different DPr-PaDiO projections, to
investigate a possible relationship between the difference and the
magnitude of the measurement. A global significance level of 0.05 was
used, but each of the 3 pairwise comparisons was tested at a
680
Fig 1: Dorsally reconstructed computed tomographic (CT) image (a) and 55°
dorsoproximal-palmarodistal oblique (D55°Pr-PaDiO) radiographic image (b) of a distal
sesamoid bone, showing 7 and 4 distal border synovial invaginations, respectively.
Lateral is on the left side and medial on the right side of the image.
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
Evaluation of distal navicular border synovial invaginations
S. Claerhoudt et al.
TABLE 2: Agreement between computed tomographic and 3
radiographic projections for number, depth and shape (*: only
unweighted k was calculated for shape2)
Radiographic projection
Categorical variable
k
Weighted k
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
Number
Number
Number
Depth
Depth
Depth
Shape4
Shape4
Shape4
Shape2
Shape2
Shape2
0.12
0.02
0.01
0.23
0.19
0.23
0.49
0.50
0.42
0.68*
0.7*
0.55*
0.24
0.17
0.15
0.30
0.26
0.29
0.57
0.60
0.49
–
–
–
than radiography. The D45°Pr-PaDiO projection showed the smallest mean
difference. The corresponding 95% reference intervals, as well as the 95%
CIs, were very wide for all 3 comparisons, reflecting a great variation in the
differences (Table 1). The Bland-Altman plots of the differences for number
between the methods against their means are shown in Figure S1.
The weighted k values for all 3 projections were very low, specifically for
the D55°Pr-PaDiO and D65°Pr-PaDiO projections, meaning that a greater
disagreement with CT for number was calculated for these 2 projections
(Table 2).
Depth
The linear measurements of depth of synovial invaginations (measurement
A of the equation) and distance between the distal and proximal flexor
borders (measurement B) increased with radiographic angle in 25 of 50
(50%) DSBs. There were no significant mean differences for depth between
CT and the 3 radiographic projections, and the confidence intervals were
also narrow, meaning that there seems to be little or no bias. The mean
differences of measurements for depth between CT and the 3 radiographic
projections were all positive and small, with the D45°Pr-PaDiO projection
showing the smallest mean difference. By evaluating reference intervals, it
is possible to see how precise the individual estimates are. Consequently,
the reference intervals for all 3 DPr-PaDiO projections showed quite wide,
comparable ranges (Table 1). Ninety-five per cent of differences between
CT and the different DPr-PaDiO projections for depth lie between the limits
-0.28 and 0.32 of the interval (0.28 below or 0.32 above zero-level). Also, the
Bland-Altman plots showed a large variation in the differences against their
means (see Fig S2). It is illustrated that as the mean measurements of CT
and radiography increase (larger equations; deeper invaginations), the
differences (measurements CT minus radiography) also seem to increase,
meaning that the underestimation by radiography was greater in the case
of deeper invaginations (Fig 2).
The k statistics also revealed quite poor agreement between the
techniques (average k value of 0.21 and low weights to disagreements, as
summarised in Table 2).
Fig 2: Dorsally reconstructed CT image (a) and corresponding D65°Pr-PaDiO
radiographic image (b) of a distal sesamoid bone, showing a deeply penetrating
invagination on the CT image (arrow), which is mildly penetrating on the radiographic
image (arrow). Lateral is on the left side and medial on the right side of the image. An
example of a false-negative result is shown in Fig 3.
shape2) and corresponding 95% CIs for all 3 DPr-PaDiO projections for
shape, with CT as gold standard, are summarised in Table 3.
Discussion
The morphological features of the distal navicular border synovial
invaginations of the present population were described in a previous study
[13] using CT. The present investigation was carried out to assess the
variability and agreement between the appearance of distal border
synovial invaginations of the DSB on radiographs vs. CT examinations.
Histology is regarded as the gold standard for the diagnosis of tissue
abnormalities [14]. In the present study, no histological examination was
Shape
In only 3 of 11 (27%), 4 of 7 (57%) and 3 of 6 (50%) feet with an equal number of
synovial invaginations on both the CT scans and D45°Pr-PaDiO,
D55°Pr-PaDiO and D65°Pr-PaDiO projections, respectively, was the shape
comparable by both methods. The agreement for shape4 ranged between
0.42 ⱕ k ⱕ 0.50, representing a moderate agreement, whereas the k value
for shape2 was higher (range 0.55 ⱕ k ⱕ 0.71), representing a moderate to
good agreement. Weighted k for shape was higher for the D55°Pr-PaDiO
projection, meaning that, relative to the other projections, a better
agreement with the gold standard for shape was calculated for
this projection.
The k and weighted k values for agreement of number, depth and shape
are presented in Table 2. The sensitivity, specificity (both calculated for
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
Fig 3: Dorsally reconstructed CT image (a) and corresponding D55°Pr-PaDiO
radiographic image (b) of a distal sesamoid bone, showing 2 abnormally shaped
invaginations on the CT image (arrows), which were normally shaped on the
radiographic image (arrows). Lateral is on the left side and medial on the right side of
the image.
681
Evaluation of distal navicular border synovial invaginations
TABLE 3: Sensitivity and specificity for all 3 radiographic projections
for shape with computed tomography as gold standard
Radiographic
projection
Sensitivity
(95% confidence
interval)
Specificity
(95% confidence
interval)
D45°Pr-PaDiO
D55°Pr-PaDiO
D65°Pr-PaDiO
0.65 (0.51–0.78)
0.65 (0.51–0.78)
0.63 (0.48–0.76)
0.97 (0.92–0.99)
0.99 (0.95–1)
0.90 (0.84–0.95)
performed; however, owing to the possibility of reconstruction of
multiplanar, high-resolution images without superimposition, CT provides
detailed anatomical information on the synovial invaginations [11].
Therefore, CT was assumed by the authors to be the gold standard.
The total number of synovial invaginations counted on CT was much
higher than with radiography. The differences between CT and the 3
DPr-PaDiO projections for the number of synovial invaginations were all
statistically positive, with the D45°Pr-PaDiO projection showing the
smallest mean difference. In fact, CT permitted visualisation of an average
of 2 more invaginations per navicular bone than with radiography,
indicating a better visibility of invaginations on CT than with radiography.
Similar conclusions were reported earlier [10,11]. The synovial
invaginations that were ill defined or undetectable on radiography were
mostly small and located at the distal sloping borders of the DSB on the CT
images. It has been described that the presence of small synovial
invaginations at the sloping borders on CT may be observed in normal
DSBs, owing to lack of superimposition of surrounding bone and the high
sensitivity of CT for detecting bone in detail [11]. However, the clinical
significance of these subtle CT findings remains questionable.
In approximately 80% of the cases in our study, a DSB with 7 invaginations
on radiography had more than 7 distal border invaginations on CT.
According to the literature, up to 7 radiographically detectable distal
border synovial invaginations are considered normal, and more than 7
significant [15,16], although an overlap between sound and lame horses is
described [5]. However, the clinical relevance of this increased number of
invaginations on CT remains unclear, and further research is required.
The present results show that when the mean measurements for depth
on CT and radiography increase, the differences (measurements CT minus
radiography) also seem to increase, resulting in a larger underestimation by
radiography in case of deeper invaginations. As described above, CT
appeared better in evaluating synovial invaginations than radiography,
owing to the better visibility of subtle changes on CT [9–11]. On CT, most
deep invaginations ended proximally as very tiny, deeply penetrating lines,
which were undetectable on radiography. Therefore, care should be taken
in judging the depth of distal border invaginations on radiographs during
purchase examinations, because deeply penetrating invaginations may be
missed on DPr-PaDiO projections. It is reported that deeply penetrating
distal border invaginations are significant radiographic findings that could
be responsible for future joint pain and lameness [3,4]; however, the clinical
significance of subtle CT findings remains unclear. Further investigation
with clinical association is necessary to determine the importance of these
deeply penetrating, tiny synovial invaginations.
In the present study, the sensitivity and specificity for shape of the distal
navicular border synovial invaginations were calculated on the 3 DPr-PaDiO
projections using CT as the gold standard. For radiography as diagnostic
modality, both a high sensitivity (i.e. high number of correctly identified
abnormal shaped invaginations) and a high specificity (i.e. correctly
identified absence of abnormally shaped invaginations) are desired.
False-positive diagnosis (poor specificity) of abnormally shaped
invaginations can have major consequences, because the literature
describes these findings to be related to navicular disease [1,5–8].
However, our results show an almost 100% specificity for all projections for
shape, meaning that almost no false-positive results were seen. On the
contrary, more false-negative results were present, resulting in a much
lower sensitivity of 65% for shape. In other words, an abnormal invagination
on radiography is effectively abnormal, but a normal one can in fact
be abnormal.
The present results show a variable degree of agreement (k values)
between measurements on CT and the DPr-PaDiO projections for the 3
682
S. Claerhoudt et al.
variables. All data for shape were grouped into 2 categories (2 ¥ 2 table) for
shape2 and 4 categories (4 ¥ 4 table) for shape4, resulting in variable values
of k. For the resulting 2 ¥ 2 table a better average k value of 0.65 was found,
compared with k = 0.47 for the 4 ¥ 4 table. In contrast, data on numbers
were classified into 12 categories (0 being the lowest number of
invaginations found and 11 the highest), logically resulting in very low
k values. In theory, any value of k much below 0.5 will indicate poor
agreement. However, despite these published guidelines [12], no value of k
can be regarded universally as indicating some degree of agreement [17].
In fact, the value of k depends on the number of categories and upon
circumstances, as demonstrated in our results. For multiple categories on
an ordinal scale, weighted k has the advantage that it ‘weights’ the degree
of disagreements. Greater disagreement is penalised more, resulting in
lower weighted k values.
The results of the present study showed that the variability of the actual
differences between CT and radiography is high for number and depth of
penetration of distal navicular border synovial invaginations. This can be
explained by 3 factors. A first factor is the variable orientation of the
synovial invaginations in the DSB. Indeed, a recent study has demonstrated
that the orientation of the distal border synovial invaginations into the DSB
can vary from a straight, dorsoproximal to a palmaroproximal direction
[13]. A second factor is the variable height of the heels. Pearce et al. [18]
demonstrated that the degree of distal interphalangeal joint angulation
increases (increased joint flexion) with heel elevation, and van Dixhoorn
et al. [19] reported that the DSB follows the coffin bone in vitro during distal
interphalangeal joint flexion. Thus, elevation of the heels results in at least
an increased upright motion of the DSB in the sagittal plane. Finally, a third
factor is the superimposition of the DSB over other structures on a
DPr-PaDiO projection, preventing visualisation of the exact point of origin of
the synovial invaginations [10]. The distal contour of the DSB on DPr-PaDiO
projections is visualised as 2 lines, one representing the articular border
and the second the flexor border, with the distal border synovial
invaginations being situated in the groove between these borders [16,20].
A consequence of the variable orientation of the invaginations into the
bone and of the large individual variation of heel height is that the depth of
the distal navicular border synovial invaginations relative to the horizontal
x-ray beam varies individually when the front of the hoof wall is angled
forward at approximately 45, 55 and 65°. In fact, when elevating the heels,
the depth of dorso- and palmaroproximal oriented synovial invaginations
on DPr-PaDiO radiographic projections respectively shortens and enlarges
relative to the degree of heel elevation. The opposite effects are obtained
by lowering the heels.
A potential limitation of this study could be the absence of a
palmaroproximal-palmarodistal oblique projection, which permits
evaluation of the DSB without superimposition. However, this projection
only allows evaluation of the number and width of the distal border
invaginations [5]. Therefore, it was not included in our study.
In conclusion, the results of the present study indicate that radiography
differs significantly from CT concerning the morphology of distal navicular
border synovial invaginations. CT is a much more sensitive method for the
detection of these invaginations; therefore, regardless of the exact clinical
meaning of the synovial invaginations, the criteria as used for radiography
can absolutely not be used unchanged in case of CT examination.
Prospective epidemiological studies are necessary to assess the clinical
significance of CT-detected abnormalities in this area.
Authors’ declaration of interests
No conflicts of interest have been declared.
Source of funding
This study was financed by the ‘Bijzonder Onderzoeksfonds’, Ghent
University.
Acknowledgements
The authors would like to thank Kim Claus and Marnix Verdonck for
technical assistance. We thank Katrien Vanderperren for constructive
criticism of the manuscript.
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
S. Claerhoudt et al.
Manufacturers’ addresses
a
Mx8000, Philips Medical Systems, AE Eindhoven, The Netherlands.
Osirix Image processing Software, Geneva, Switzerland.
c
Playdoh®: Rainbow Crafts, Cincinnati, Ohio, USA.
d
Mobilux, X-ray Equipment Verachtert, Antwerpen, Belgium.
e
Totoku monochrome LCD display, Lewisville, Texas, USA.
f
Microsoft Excel, Microsoft Corp., Redmond, Washington, USA.
b
References
1. Wright, I.M. (1993) A study of 118 cases of navicular disease: radiological
features. Equine Vet. J. 25, 493-500.
2. Dyson, S., Murray, R., Blunden, T. and Schramme, M. (2006) Current concepts
of navicular disease. Equine Vet. Educ. 18, 45-56.
3. Robert, C., Valette, J.P. and Denoix, J.M. (2006) Correlation between routine
radiographic findings and early racing career in French Trotters. Equine Vet. J.,
Suppl. 36, 473-478.
4. Dik, K.J. (2007) Diagnostische beeldvorming. In: De Veterinaire Keuring Van Het
Paard, 3rd edn., Eds: M.M. Sloet van Oldruitenborgh-Oosterbaan, A.
Barneveld, A.J. Van Den Belt, B.V. Libre, Leeuwarden. pp 85-112.
5. Rose, R.J., Taylor, B.J. and Steel, J.D. (1978) Navicular disease in the horse: an
analysis of seventy cases and assessment of a special radiographic view.
J. Equine Med. Surg. 2, 492-497.
6. MacGregor, C.M. (1986) Radiographic assessment of navicular bones, based
on changes in the distal nutrient foramina. Equine Vet. J. 18, 203-206.
7. Hertsch, B. and Dammer, H. (1988) The blood supply of normal and diseased
navicular bones. Vet. Rad. 29, 276-281.
8. Kaser-Hotz, B. and Ueltschi, G. (1992) Radiographic appearance of the
navicular bone in sound horses. Vet. Radiol. Ultrasound 33, 9-17.
9. Tietje, S. (1995) Computed tomography of the navicular bone region in the
horse: a comparison with radiographic documentation. Pferdeheilkunde 11,
51-61.
10. Widmer, W.R., Buckwalter, K.A., Fessler, J.F., Hill, M.A., VanSickle, D.C. and
Ivancevich, S. (2000) Use of radiography, computed tomography and magnetic
resonance imaging for evaluation of navicular syndrome in the horse. Vet.
Radiol. Ultrasound 41, 108-116.
11. Groth, A.M., May, S.A., Weaver, M.P. and Weller, R. (2009) Intra- and
interobserver agreement in the interpretation of navicular bones on
radiographs and computed tomography scans. Equine Vet. J. 41, 124-129.
12. Landis, J.R. and Koch, G.G. (1977) The measurement of observer agreement for
categorical data. Biometrics 33, 159-174.
13. Claerhoudt, S., Bergman, E.H.J., van der Veen, H., Vanderperren, K., Raes, E.V.
and Saunders, J.H. (2011) Computed tomographic morphology of the synovial
invaginations of the distal sesamoid bone of the horse. Anat. Histol. Embryol.
40, 55-60.
Equine Veterinary Journal 44 (2012) 679–683 © 2012 EVJ Ltd
Evaluation of distal navicular border synovial invaginations
14. Murray, R.C., Blunden, T.S., Schramme, M.C. and Dyson, S.J. (2006) How does
magnetic resonance imaging represent histologic findings in the equine digit?
Vet. Radiol. Ultrasound 47, 17-31.
15. Colles, C.M. (1983) Interpreting radiographs 1: the foot. Equine Vet. J. 15,
297-303.
16. Rijkenhuizen, A.B.M., Németh, F., Dik, K.J. and Goedegebuure, S.A. (1989)
Development of the navicular bone in foetal and young horses, including the
arterial supply. Equine Vet. J. 21, 405-412.
17. Altman, D.G. (1991) Some common problems in medical research. In: Practical
Statistics for Medical Research, Ed: D.G. Altman, Chapman & Hall, London. pp
396-439.
18. Pearce, S.G., Boure, L.P., Bolger, A., Thomason, J.J. and Dobson, H. (2004)
Effect of heel elevation on forelimb conformation in horses. Aust. Vet. J. 82,
558-562.
19. van Dixhoorn, I.D.E., Meershoek, L.S., Huiskes, R. and Schamhardt, H.C. (2002)
A discription of the motion of the navicular bone during in vitro vertical loading
of the equine forelimb. Equine Vet. J. 34, 594-597.
20. Dik, K.J., van den Belt, A.J.M., Enzerink, E. and van Weeren, P.R. (2001) The
radiographic development of the distal and proximal double contours of the
equine navicular bone on dorsoproximal-palmarodistal oblique (upright pedal)
radiographs, from age 1 to 11 months. Equine Vet. J. 33, 70-74.
Supporting Information
Additional Supporting Information may be found in the online version of
this article:
Fig S1: Bland–Altman plots: differences of counted numbers (readings) on
computed tomographic and radiographic images (A: D45°Pr-PaDiO; B:
D55°Pr-PaDiO and C: D65°Pr-PaDiO) against their means. The numerical
code represents the number of distal sesamoid bones (total sum of 50) with
the same difference in number against mean. Dashed line represents
zero-level (no difference), upper line = mean + 2s.d., middle line = mean,
lower line = mean – 2s.d. s.d. = standard deviation; 95%RI = 95% reference
interval; 95%CI = 95% confidence interval.
Fig S2: Bland–Altman plots: differences of the equations of depth
(readings) on computed tomographic and radiographic images (A:
D45°Pr-PaDiO; B: D55°Pr-PaDiO and C: D65°Pr-PaDiO) against their means.
Each individual plot (O) represents a synovial invagination (total of 295
plots). Dashed line represents zero-level (no difference), upper line = mean
+ 2s.d., middle line = mean, lower line = mean – 2s.d. s.d. = standard
deviation; 95%RI = 95% reference interval; 95%CI = 95% confidence interval.
Please note: Wiley-Blackwell are not responsible for the content or
functionality of any supporting materials supplied by the authors. Any
queries (other than missing material) should be directed to the
corresponding author for the article.
683
