High‐field (3 Tesla) MRI of the navicular apparatus of sound horses shows good agreement to histopathology

: Magnetic resonance imaging and the correlation to histopathological findings of the equine palmar foot of lame horses have been described previously, using 0.27 and 1.5 T systems. Compared to these, 3 T systems provide superior spatial resolution and imaging contrast. The aim of our prospective anatomic study was to characterize the imaging anatomy of the navicular region on 3 T MRI in comparison to histopathological findings. We hypothesized that 3 T MRI allows a good visualization of the entire navicular apparatus and reliable measurements of navicular cartilage and cortical bone thickness. Twenty front feet of sound horses were examined using a 3 T MRI system. For histopathological examination, sagittal tissue sections of the navicular bones and adjacent ligaments were prepared. Alterations in magnetic resonance signal were graded for each region and compared to corresponding histological slices. Overall, there was good visualization of the anatomical detail and a very good agreement between MRI and histology for compact bone and spongiosa, good agreement for the fibrocartilage and the distal sesamoidean impar ligament, but only moderate agreement for the hyaline cartilage and the collateral sesamoidean ligament. A comparative measurement of cartilage and cortical bone thickness on magnetic resonance images and histological sections was performed. In MRI, the hyaline cartilage of the articular surface appeared significantly thinner and the fibrocartilage of the flexor surface appeared significantly thicker compared to histology. Findings indicated that MRI at a field strength of 3 T allows reliable depiction of anatomic details of the navicular apparatus. Abstract Magnetic resonance imaging and the correlation to histopathological findings of the equine palmar foot of lame horses have been described previously, using 0.27 and 1.5 T systems. Compared to these, 3 T systems provide superior spatial resolution and imaging contrast. The aim of our prospective anatomic study was to characterize the imaging anatomy of the navicular region on 3 T MRI in comparison to histopathological findings. We hypothesized that 3 T MRI allows a good visualization of the entire navicular apparatus and reliable measurements of navicular cartilage and cortical bone thickness. Twenty front feet of sound horses were examined using a 3 T MRI system. For histopathological examination, sagittal tissue sections of the navicular bones and adjacent ligaments were prepared. Alterations in magnetic resonance signal were graded for each region and compared to corresponding histological slices. Overall, there was good visualization of the anatomical detail and a very good agreement between MRI and histology for compact bone and spongiosa, good agreement for the fibrocartilage and the distal sesamoidean impar ligament, but only moderate agreement for the hyaline cartilage and the collateral sesamoidean liga-ment.Acomparativemeasurementofcartilageandcorticalbonethicknessonmagneticresonance images and histological sections was performed. In MRI, the hyaline cartilage of the articular surface appeared significantly thinner and the fibrocartilage of the flexor surface appeared significantly thicker compared to histology. Findings indicated that MRI at a field strength of 3 T allows reliable depiction of anatomic details of the navicular apparatus.

a good visualization of the entire navicular apparatus and reliable measurements of navicular cartilage and cortical bone thickness. Twenty front feet of sound horses were examined using a 3 T MRI system. For histopathological examination, sagittal tissue sections of the navicular bones and adjacent ligaments were prepared. Alterations in magnetic resonance signal were graded for each region and compared to corresponding histological slices. Overall, there was good visualization of the anatomical detail and a very good agreement between MRI and histology for compact bone and spongiosa, good agreement for the fibrocartilage and the distal sesamoidean impar ligament, but only moderate agreement for the hyaline cartilage and the collateral sesamoidean ligament. A comparative measurement of cartilage and cortical bone thickness on magnetic resonance images and histological sections was performed. In MRI, the hyaline cartilage of the articular surface appeared significantly thinner and the fibrocartilage of the flexor surface appeared significantly thicker compared to histology. Findings indicated that MRI at a field strength of 3 T allows reliable depiction of anatomic details of the navicular apparatus. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

METHODS
The study was a prospective, anatomic design.   Table 1). Image acquisition took approximately 30 min per foot and both front feet were examined.

Image interpretation
The magnetic resonance images of the 20 front feet were reviewed using a diagnostic workstation (easyIMAGE, VetZ, Isernhagen, Germany) in a randomized order by three observers independently (two trained interns and one experienced equine surgeon) using a grading system established for evaluation of the navicular bone in high-field (1.5 T) magnetic resonance images. 8

Tissue harvest
The navicular bones of the front feet and their supporting ligaments were dissected. The samples were fixed in 10% buffered formalin, cut into sagittal slices with a thickness of 3 mm, decalcified for 4 weeks, and subsequently embedded in paraffin wax and sectioned at 3 µm.  Each tissue section was photographed with a camera-microscope combination and measurements of cortical bone thickness and cartilage thickness were taken using an imaging software (cell D, Olympus, Hamburg, Germany). Seven measuring points corresponding to the magnetic resonance images were set at the navicular bone: proximal, central, and distal both on articular surface and the flexor surface, respectively, and at the distal margin ( Figure 1). All measurements were taken perpendicular to the cartilage surface, once with and once without consideration of the tidemark.

Statistics
Selection

Flexor surface
In all images, the flexor surface was clearly separated from the hyperin-

Articular surface
The The cortex at the articular surface appeared normal (grade 0) on MRI in all but two navicular bones (90%), in which a slightly irregular tidemark was identified (grade 1) but no histopathologic changes were present. Thus, agreement in grading was very good ( = .8).

Spongiosa
The spongiosa showed homogenous hypointense signal in fatsuppressed proton density with fat saturation (spectral attenuated inversion recovery)-weighted images and intermediate to hyperintense signal in proton density and T1 weighted sequences without any signal alterations (grade 0). Correspondingly, no histopathological findings were recorded within the spongiosa (grade 0). The agreement between grading in MRI and histology was very good ( = 1).

Proximal margin and collateral sesamoidean ligament
The proximal compact bone was of uniform low signal intensity. In three navicular bones (15%), grade 1 irregularities were evident on MRI and histology. There was very good agreement between MRI and histology ( = 1).

Distal margin and distal sesamoidean impar ligament
There were no pathologic changes at the distal cortex in 16 limbs (80%). There were four navicular bones with grade 1 irregularities in histology; two of them were graded normal on MRI (grade 0). Agreement between MRI and histology for the distal compact bone was good ( = .61).

Hyaline cartilage
In all navicular bones, the hyaline cartilage appeared significantly thinner on MRI compared to histology (see Table 2). After excluding the calcified cartilage (mean thickness 0.3 mm) from the histological measurement, there was no more significant difference between measurements (P = .06).

Fibrocartilage
At the proximal aspect of the flexor surface, the fibrocartilage appeared significantly thinner on MRI compared to histology (P < .001). In contrast, in the central region of the flexor surface, the fibrocartilage was significantly thicker on MRI, compared to histology (P < .001). At the distal measuring point, there was no significant difference between results of MRI and histology (P = .52; see Table 2, measuring point 5-7). The magnetic resonance sequences did affect the measuring of the fibrocartilage. In proton density with fat saturation (spectral attenuated inversion recovery)-weighted images, the fibrocartilage came out significantly thicker compared to proton density or T1-weighted images (P = .007).

Compact bone
There was good agreement between measurements of the compact bone at the articular surface. There was neither a significant difference   In contrast, there was a significant difference between measurements at the palmar cortex. The cortex appeared significantly thicker in MRI, compared to histology (p = 0.02) for the proximal measuring point, and p < 0.001 for the central, as well as at the distal measuring point.

DISCUSSION
The results of this study supported the hypothesis that 3 T MRI allows good visualization of anatomic details in the equine navicular apparatus. Poor agreement between 1.5 T MRI and histological grading has been reported for the dorsal and proximal border of the navicular bone. 7 In the present study, the agreement in grading at the articular surface was better than postulated for 1.5 T MRI (moderate agreement for the hyaline cartilage; very good agreement for the dorsal compact bone). This can be explained by the fact that a higher field strength results in a lower signal-to-noise ratio, increased spatial resolution, and decreased susceptibility to artifacts. 9 Several horses showed a focal hyperintense signal in the fibrocartilage layer due to pooling of synovial fluid next to the sagittal ridgeobvious in proton density and proton density with fat saturation (spectral attenuated inversion recovery)-weighted sequences. This finding was likely due to the mid-ridge synovial fossa (fossa nudata),whichis considered a normal anatomical variation. 13,20 Alterations at the proximal border of the navicular bone have been over interpreted on 1.5 T magnetic resonance images compared to histology. 7 The agreement between MRI grading and histological grading at the proximal border was very good ( = 1) in our study, but mild pathological changes were evident.
The axial collateral sesamoidean ligament was formerly described as structure of homogeneous low signal intensity on spoiled gradient echo and gradient echo sequences, interspersed with uniformly distributed areas of higher signal on gradient echo. 8 In contrast, we described a generally increased signal at the insertion in sagittal slices ( Figure 5) in our group of horses without lameness. Probable explanation is the presence of mostly cross-sectioned fiber bundles in this region, compared to the rest of the ligament, as this was obvious also on histology. This might explain the increased signal intensity in all sequences in the absence of any pathological findings. Based on these results, we recommend modifying the evaluation criteria for the axial collateral sesamoidean ligament, and to consider a slightly increased signal as a physiological variation without clinical significance.
Pathologies of the distal sesamoidean impar ligament have been rated difficult to detect due to the heterogeneous appearance of the ligament. 2,7,12 In the current study, blood vessels and synovial invaginations often led to focal, clearly demarcated signal increase within the distal sesamoidean impar ligament and should not be misinterpreted as pathological findings. Two navicular bones had a distal border fragment, which appeared as an oval-shaped area of low signal intensity within the distal sesamoidean impar ligament. Agreement between MRI and histological grading for the distal sesamoidean impar ligament was good ( = .61) and therefore better than the postulated fair to moderate agreement described previously for 1.5 T images. 7 Our study supports the previously described finding of distal border fragments in nonlame horses. 19 Another focus of this study was the correlation between measurements of bone and cartilage thickness compared to histological measurements. There is clinical interest in identifying cartilage loss at an early stage, because degeneration of the fibrocartilage is an early sign of navicular disease. 4,[11][12][13] Cartilage measurement at the flexor surface in this study showed good correlation in the distal half, only. This is probably related to the optimal 90 • angulation of the sagittal magnetic resonance slice to the cartilage surface in the distal part. The fibrocartilage of the flexor surface is the thinnest cartilage layer in the limb, and there is no adjacent layer of synovial fluid to provide contrast due to the tight apposition between the cartilage of the flexor surface and the deep digital flexor tendon. Saline injection into the navicular bursa (magnetic resonance bursography) improved the accuracy of MRI evaluation of palmar fibrocartilage using a 1.5 T scanner. 13 The hyaline cartilage of the articular surface turned out to be signif-  15 Probably the less curved surface of the carpal cartilage, the greater cartilage thickness, or the use of different magnetic resonance sequences compared to our study did affect the results.
The flexor cortex appeared thicker on magnetic resonance images compared to histology. Chemical shift artifacts lead to potential problems for cortical bone measurements but is limited by increasing the receive bandwidth at high-field strength. 22,23 The tendency to overestimate the bone thickness on magnetic resonance images should be kept in mind for interpreting pathological findings at the palmar border.
A limitation of the study was the relatively small number of horses without history of lameness related to the foot. Additionally, comparison between MRI and histopathology including the measurement was performed only on sagittal slice direction. The acquisition time per foot was 40 min, which is quite long and may not be practicable for every case under general anesthesia in particular when both front feet have to be examined. There was good visibility of the flexor surface on proton density and proton density with fat saturation (spectral attenuated inversion recovery)-weighted sequences but not on T1 weighted images. Therefore, it could be possible to exclude the T1 sequences in two different orientations to help save anesthesia time under clinical conditions.
In conclusion, the current study demonstrated a good visualization of anatomical details of the equine navicular apparatus using 3 T MRI under clinical conditions. Mild to severe MRI lesions were described, but the only severe lesions were distal border fragments that have been previously reported to be of no clinical significance. Care should be taken not to overinterpret mild lesions of the navicular apparatus using 3 T MRI. The agreement between MRI and histological grading was higher than previously postulated for lower field strength MRI, as we expected. The improved depiction of anatomic details in the navicular apparatus with 3 T MRI may help in understanding the spectrum of injuries that contribute to palmar foot pain and may improve future assessments of therapeutic options.