Reductions in distances between dorsal spinous processes on radiographs are used as criteria for the diagnosis of impingement of the thoracic dorsal spinous processes in horses but are potentially altered by spine motion and different head and neck positions.
To determine the influence of head and neck positions on intervertebral distances between dorsal spinous processes on radiographs of thoracic spines of clinically sound horses.
Lateral–lateral radiographs were obtained from 23 horses in 3 head and neck positions. The width of the thoracic dorsal spinous processes and intervertebral distances between adjacent thoracic dorsal spinous processes were measured at points perpendicular to a tangent between the dorsal spinous processes and the caudal extremity of the thoracic vertebrae.
A low head and neck position increased intervertebral distances between adjacent thoracic dorsal spinous processes from the 8th to 15th dorsal spinous processes whereas a high head and neck position had the opposite effect (P<0.05). Overall, intervertebral distances between adjacent thoracic dorsal spinous processes decreased from cranial to caudal in intermediate head and neck positions (P<0.01). The 12th thoracic dorsal spinous process was readily identifiable due to its significant difference to the narrower cranial and broader caudal dorsal spinous process (P<0.05).
The head and neck position influences the distances between the dorsal spinous processes of the vertebrae of equine thoracic spine on radiography.
The measuring system reported here offers potential to improve and standardise radiographic evaluation of thoracic dorsal spinous processes.
Impingement of the dorsal thoracic dorsal spinous processes is a common cause of back pain in horses  and is diagnosed by a narrowing of the distance between the dorsal spinous processes of the thoracic vertebrae on radiographs [2-5]. However, the intervertebral distances can be altered by spine motion, whereby flexion of the thoracic vertebral column increases and extension decreases the space between the dorsal spinous processes . Changes in the horse's head and neck position can also alter these intervertebral distances . A high head and neck position leads to extension of the cranial thoracic spine, flexion of the caudal thoracic spine and decreases overall dorsoventral movement of the dorsal spinous processes [8, 9]. Conversely, a low head and neck position results in flexion of the cranial thoracic spine, extension of the caudal thoracic spine and increases overall dorsoventral movement of the dorsal spinous processes [8, 9]. The latter effect can also be caused by sedation frequently used to facilitate radiographic examinations .
Because impingement of the thoracic dorsal spinous processes is a common reason for insurance claims [11, 12], radiographic examination of the dorsal spinous processes is often included in prepurchase examinations [13-15]. Interpretation of these radiographs can be challenging due to the poor correlation between clinical signs and radiographic abnormalities [5, 16] and the lack of a uniform system for determining the intervertebral distances between thoracic dorsal spinous processes [3, 4]. Most commonly, the subjectively narrowest intervertebral distance or the intervertebral distance 2 cm ventral to the dorsal edge of the dorsal spinous processes between adjacent dorsal spinous processes are measured. The aims of this study were to determine the influence of head and neck position on the intervertebral distances between thoracic dorsal spinous processes and to develop a reliable measuring system for standardised evaluation of radiographs of the dorsal spinous processes.
Materials and methods
Twenty-three horses (aged 3–22 years; mean 11 years, 11 Warmbloods, 8 Haflinger, 3 ponies and one Heavy Warmblood; 11 mares, 10 geldings and 2 stallions) were included in this study. Only 3 of the horses were regularly ridden; the others were 13 pasture horses, 5 stud mares and 2 pleasure horses.
All horses underwent a clinical examination of the back, including inspection and palpation of the back muscles and evaluation of the dorsoventral and lateral mobility of the back. Neurological examinations and lameness evaluations were performed to rule out other abnormalities. Only horses lacking clinical signs of back problems were included in this study.
After sedation with romifidine hydrochloride (Sedivet1, 0.06 mg/kg bwt i.v.) and butorphanol tartrate (Alvegesic2, 0.01 mg/kg bwt i.v.) the horses were radiographed in the following 3 head and neck positions (Fig 1):
Intermediate: mouth at the level of the shoulder joint,
Low: mouth at the level of the carpal joint,
High: mouth at the level of the withers.
The horses stood squarely on all 4 feet to prevent lateroflexion or rotation of the spine. In each of the 3 different positions, 2 digital, lateral–lateral radiographs (Fujifilm FCR 5000)3 of the spine from the 7th thoracic to the first lumbar vertebrae with an overlap of approximately 4 cm were obtained using a Siemens Polydorus 100 X-ray tube4. Radiographs in any one horse were either all left lateral to right lateral or all right lateral to left lateral. The exposure values were 80–87.5 kV and 160–180 mAs depending on the horse's size. The cassettes (Fujifilm Cassette type C, 35 × 43 cm)3 were held as closely as possible to the horse to minimise distortion of the images by magnification. To minimise scattered radiation reaching the imaging plate, a Bucky Grid (r12, N40, Pb)4 and a custom-made device4 positioned on the horse's back were used (Fig 2).
The Bucky grid was vertically aligned to ensure that the arrangement of cassettes remained constant and the cranial border of the cassette was always vertical. Additionally, the tube head and Bucky grid were interlinked to provide parallel adjustment.
The custom-made device consisted of 2 rows of lead pads arranged perpendicular to each other fixed with foam pads. The row close to the horse's back was parallel to it, the other row was centred and vertical to the horse's back. Through their special arrangement, the lead pads minimised the scattered radiation reaching the imaging plate above the thoracic dorsal spinous processes.
To obtain radiographs of the thoracic dorsal spinous processes and vertebral bodies with same exposure values, an aluminium filter5 was used to progressively attenuate the primary beam from ventral to dorsal compensating for variation of tissue thickness. The x-ray beam was centred on the articular process joints, focus object distance was 80 cm and focus film distance was 110 cm.
Isolated spines were extracted and post mortem radiographs obtained from 6 of the Haflingers subjected to euthanasia for unrelated reasons. Pre- and post mortem radiographs were compared subjectively to verify the number of each thoracic dorsal spinous process. In the other horses, the first lumbar vertebra was used as a reference point.
A digital image analysis program (curaSmartClient)6 was used to evaluate the digital radiographs. Due to poor contrast in regions of interest, radiographs were processed with an image filter for local adaptive contrast improvement according to Niblack , which was implemented as a C++ dynamic link library using Microsoft Visual Studio 20087 and integrated into the existing radiology software CuraSmartClient6. Local threshold of an image window with size b around pixel position i,j was calculated using Niblacks formula T(i,j) = m(i,j)-k*s(i,j), where k (scaling factor for s.d.) was set at 0.2 by default and b defined empirically as 55 pixels. This filter width was chosen in order to gain enough pixel values for reliable calculation of local mean and local standard deviation from the greyscale values of the enclosed image pixels. Object boundaries were most noticeable at these values (Fig 3).
A tangent reference line, identically orientated to each vertebra, was drawn at the vertebra from the caudal point of the caudal extremity to the caudal point of the thoracic dorsal spinous processes (Fig 3). The width of the thoracic dorsal spinous processes and the intervertebral distance, i.e. the distance from each dorsal spinous process to the caudal adjacent processs were measured. These measurements were made perpendicularly to the tangent, both starting at the point of intersection between the tangent and the caudal point of the thoracic dorsal spinous processes (Fig 3). The width and intervertebral distance measurements at the low and high positions were divided by the corresponding measurements made at the intermediate position to provide relative widths and intervertebral distances.
Only thoracic dorsal spinous processes that were separate from caudal adjacent dorsal spinous processes and that could be evaluated on radiographs of each different head and neck positions, were included in this study. The 18th vertebrae was excluded because the vertebral body was obscured by soft tissue.
All measurements were repeated 3 times by the same observer (D.B.) at least one week apart. The earlier measurements were not referred to when the measurements were repeated.
Microsoft Excel7 and SPSS188 were used for statistical analysis. To verify accuracy of the measurements, the coefficient of variation and combined means were calculated for repeated measurements and the combined means were used for descriptive statistics. The Shapiro–Wilk test was used to analyse data normality. Analysis of the influence of head and neck position on absolute and relative widths of the dorsal spinous processes and relative intervertebral distances between adjacent dorsal spinous processes were performed using Friedman's ANOVA and Wilcoxon signed-rank test with post hoc Bonferroni adjustment. The minimum level of significance was set at P<0.05.
The coefficient of variation for the measurements made 3 times one week apart was 2–6%. Absolute and relative widths of thoracic dorsal spinous processes of the 7th to 15th and 17th thoracic vertebrae were not significantly different between the 3 different head and neck positions (Fig 4). There were no significant differences in the absolute width of 16th thoracic dorsal spinous processes but the relative width of the 16th thoracic dorsal spinous process was significantly different. Relative intervertebral distances between the 8th to 14th thoracic dorsal spinous processes were significantly wider in the low head and neck position than in the other 2 positions (P<0.05; Fig 5). The relative intervertebral distances between the 15th and 16th and 16th and 17th thoracic dorsal spinous processes were significantly wider in the low position than in the high position. In the high position, the relative intervertebral distances between the 9th, 10th, 11th, 13th and 17th thoracic dorsal spinous processes and their respective caudal adjacent processes were significantly narrower than in the intermediate position (P<0.05).
The absolute width of the 12th thoracic dorsal spinous process was significantly greater than that of the 7th to 10th thoracic dorsal spinous process and significantly less than that of the 14th to 15th thoracic dorsal spinous processes (P<0.05) (Fig 6). Absolute intervertebral distances between the dorsal spinous processes decreased from cranial to caudal in the intermediate position (P<0.01).
Subjective comparison of pre- and post mortem radiographs in the subset of 6 horses confirmed that the transition between narrow cranial and wide caudal dorsal spinous processes occurred at the 12th thoracic vertebrae in all cases.
This study investigated the influence of different head and neck positions on intervertebral distances between thoracic dorsal spinous processes measured from lateral–lateral radiographs. A high head and neck position resulted in a decrease in the intervertebral distances to the caudal adjacent spinous processes from the 8th to 15th thoracic dorsal spinous processes. Conversely, lowering of the head increased the intervertebral distances between these spinous processes.
Lowering of head and neck is an effect of sedation that is commonly used to facilitate radiographic examination of horses . In standing horses, flexion of the neck leads to flexion of the thoracic spine ; conversely, lifting of the head and neck most probably causes extension of the thoracic spine. This is deduced from the fact, that extension of the thoracic spine has been associated with lifting of the head and neck . Our study confirms this deduction as the distances between adjacent thoracic dorsal spinous processes were lowered by the high head and neck position. Flexion of the thoracic spine alters the positions of the thoracic vertebrae causing intervertebral distances between the spinous processes to become wider; extension induces the contrary. The region that undergoes most dorsoventral movement is the cranial saddle region between the 10th and 14th thoracic vertebrae [6, 7, 20, 21], which corresponds very well with the results of this study.
Several studies have indicated that raising head and neck position leads to flexion of the caudal thoracic spine [8, 9]. This region was excluded from the present study because the measuring system used in the study was not reliable for the caudal spinous processes as they were not clearly defined due to their superimposition with the diaphragm and soft tissues. In contrast to previous studies, in the current study, the head and neck were only low but not flexed [8, 9]. Previous researchers have used rollers or side reins to manipulate head and neck position but this was not possible here because of the unavoidable superimposition with the spinous processes on the radiographs that such equipment produces. Additionally, horses stood squarely on all 4 feet to limit changes in the intervertebral distances only to different head and neck positions. According to the bow-and-string-concept, the thoracic spine is flexed by retraction of forelimbs and protraction of hindlimbs. Protraction of forelimbs and retraction of hindlimbs result in extension of the thoracic spine . Consequently, intervertebral distances between the spinous processes might increase if radiographs were obtained in positions in which horses flex their neck, retract their forelimbs and protract their hindlimbs. Due to its influence on spine movement, the head and neck positions should always be considered when taking radiographs of the thoracic spine in horses.
There is little information on the reference points used to determine the intervertebral distance between the thoracic dorsal spinous processes. One common approach is to identify and measure the subjectively narrowest intervertebral distance between spinous processes. However, by evaluating the same radiographs taken at each of the head and neck positions on 3 separate occasions, we have found that measurements derived by that method were not repeatable (data not shown). In other studies, intervertebral distances were measured 1.5 cm ventral to the proximal edge of the thoracic dorsal spinous processes in foals  and 2 cm in adult horses , respectively.
This study included both Warmbloods and ponies with high variation in lengths of the thoracic dorsal spinous processes and it was not possible to identify an optimal measurement point based on the intervertebral distance from the proximal edge of the thoracic dorsal spinous processes. Therefore, we developed a new measuring system using 2 reference points that were static to each other to allow measurements to be made at the same point on each radiograph and have provided evidence for the repeatability of this method. Magnification of anatomical structures varies among horses of different shapes and sizes. To account for this, the widths of the thoracic dorsal spinous processes and the intervertebral distance between vertebrae were corrected for radiographic magnification and the horse's size by being expressed as a ratio of each measurement at the high and low head and neck positions to the corresponding measurement from the intermediate position.
As the measurements obtained for the widths of the 7th to 15th thoracic dorsal spinous processes were not significantly different, these values were used to verify comparability of the radiographs and ensure that significant differences in intervertebral distances could be associated with changes in the head and neck positions. One of the limitations of the system reported here related to the requirement that the radiographs include the dorsal spinous processes as well as the vertebral bodies. The vertebral bodies of the caudal thoracic vertebrae are often not clearly defined due to their superimposition with the diaphragm and soft tissues. This is reflected in the significant differences identified for the relative widths of the 16th thoracic dorsal spinous processes between the low and high head and neck positions.
Radiologically detectable pathological changes have been described in several studies examining horses without back problems [3, 5, 16]. Thus, in prepurchase examinations and in horses without clinical signs of back pain, it is difficult to determine the clinical significance of changes identified on radiographs. Radiographs of the equine back are often taken as a part of the prepurchase examination to rule out impingement of the thoracic dorsal spinous processes [13-15] and can lead to common insurance claims [11, 12]. Therefore, it is crucial to standardise the horse's position and the measurement system in order to guarantee a uniform interpretation of these radiographs. According to the German ‘Roentgenleitfaden’ (x-ray guide) , intervertebral distances under 8 mm are a marginal variance from the norm. Other authors define only intervertebral distances under 4 mm as indicative of pathological changes [2, 3, 5, 23], or only impingement of the dorsal spinous processes by themselves . In the present study, we determined that altering the horse's head and neck position affected the intervertebral distances between the thoracic dorsal spinous processes and therefore might influence the interpretation of radiographs of this area.
The most common site for impingement of the thoracic dorsal spinous processes is the saddle region (i.e. 12th to 17th thoracic vertebrae) [1, 5, 24-26] and repeated overextension of the vertebrae is considered to be one factor in the development of impingement of the thoracic dorsal spinous processes. Athough some authors suggest that the rider's weight is the source for the over extension [24, 25], impingement of the thoracic dorsal spinous processes has been identified in horses that had never been ridden . In the present study, intervertebral distances between the dorsal spinous processes in the saddle region were smaller than those in the cranial area, which could explain the more frequent changes identified in the saddle region. In horses with back pain and in older horses, dorsoventral movement decreases [28, 29], which can result in a smaller change in the intervertebral distances between the dorsal spinous processes when using different head and neck positions. Although the average age of the horses in this study was high compared to horses investigated in other studies, we detected significant differences between different head and neck positions.
Numerical identification of dorsal spinous processes on radiographs is not easy. The diaphragm and anticlinal thoracic dorsal spinous process can serve as reference points but the diaphragm moves during respiration and the anticlinal thoracic dorsal spinous process has been described as either the 15th or 16th thoracic dorsal spinous processes [6, 16]. Comparison of premortem with post mortem radiographs of isolated spines in this study has confirmed that the 12th thoracic dorsal spinous processes marked the change between the narrow cranial and wide caudal spinous processes and therefore be can be used for numerical identification of thoracic dorsal spinous processes. Thus, radiographs should include the 11th to 13th dorsal spinous processes so that other dorsal spinous processes can be referenced from the 12th.
In conclusion, this study demonstrates the influence of head and neck positions on the intervertebral distances between the thoracic dorsal spinous processes in horses lacking clinical evidence of back pain. Because a low head and neck position results in an increase in these intervertebral distances and a high head and neck position has the opposite effect, veterinarians should take care to standardise head and neck position for radiographic examination of thoracic dorsal spinous processes in horses. Radiographs should include the 11th to 13th spinous processes in order to use the 12th thoracic dorsal spinous process as a reference point to identify the others. The measuring system described in this study proved to be reliable and easy to apply and therefore offers potential to improve standardised evaluation of such radiographs.
Authors' declaration of interests
No competing interests have been declared.
Sources of funding
This study was not funded by any external body.
The authors thank the Large Animal Clinic for Surgery of the Faculty of Veterinary Medicine of the University of Leipzig for providing equipment and facilities for the study.
Dagmar Berner: contributed to study design, data collection, study execution, data analysis and interpretation and preparation of the manuscript. Karsten Winter: contributed to study execution, data analysis and interpretation and preparation of the manuscript. Walter Brehm: contributed to study design and preparation of the manuscript. Kerstin Gerlach: contributed to study design, data analysis and interpretation and preparation of the manuscript.
Boehringer Ingelheim Vetmedica GmbH, Ingelheim am Rhein, Germany.