and displacement of the magnetic moment from equilibrium. Following the RF pulse, a gradient is used to produce a small, known variation in the magnetic field. Subsequent emission of energy (relaxation), which restores equilibrium, is proportional to the number of excited protons in the tissue volume. Protons may lose energy by dissipation into the surrounding molecular environment (T1 recovery), transfer between protons (T2 decay) or due to inhomogeneities of the magnetic field (T2* decay). Differing proton density and relaxation methods between tissues creates contrast. Multiple repetitions of the RF pulse enable the signal in an entire volume of tissue to be recorded by a receiver coil and, following a complex of mathematical processes, slices of cross‐sectional images are formed. Sagittal, transverse and dorsal planes are acquired as standard. However, MRI is multiplanar and images can be acquired in any slice plane without changing the position of the region of interest. A number of textbooks delve into the physics of MR image generation, and interested readers are referred to these for further information. [1, 137, 138].
Contrast resolution in MRI is high compared to radiography, ultrasonography and CT. Multiple factors contribute to the spatial resolution, including field and gradient strengths, matrix size and slice thickness. The magnetic field strength is measured in Tesla (T). In general, greater field strengths create images with improved contrast and more signal. Both high field (1.0–3.0 T) systems, which require general anaesthesia, and low field (0.27 T) standing MRI (sMRI) systems are available. Though sMRI units are purpose built, some institutions will use these scanners in horses under general anaesthesia. The SNR increases in a nearly linear relation to magnetic field strength [139].
The sequence generated is based on the pattern and timing of acquisition parameters. The main sequences used are spin echo (SE), fast spin echo (FSE) and gradient recalled echo (GRE). Their values pertaining to the specific tissue type is different, and each has a trade‐off in terms of acquisition time, spatial resolution and SNR. SE and FSE have higher contrast resolution than GRE, but this has a higher resolution relative to acquisition time and provides a more robust scan for sMRI if patient motion becomes challenging. Most manufacturers have proprietary sequences, particularly high field scanners intended for human use, which are developed to optimize imaging of a specific tissue type. Users must understand for which tissue or tissues proprietary sequences were developed, or understand with which traditional sequence they are most closely aligned, e.g. fluid‐sensitive sequence with higher anatomic detail.
Fat suppression can be achieved using a short tau inversion recovery (STIR) sequence or fat saturation. The latter is not possible in sMRI units. Fat suppression is essential in a fracture study; once the high signal from fat is eliminated from the image, any remaining hyperintensity pertaining to a possible fracture is clearly discernible.
Image contrast is generated through tissue weighting. T1 weighting (T1W) has high signal and good anatomical detail, but due to the increased shades of grey the contrast is reduced. T2 weighting (T2W) has lower signal than T1W or proton density weighting (PDW) but greater contrast resolution between normal and abnormal tissue. T2* weighting (T2*W) is susceptible to magnetic field inhomogeneities and ferrous materials, but since it is created using a GRE sequence it is rapidly acquired with thinner slices. It is also fluid sensitive and creates phase cancellation artefact that is helpful for ascertaining the presence of intra‐osseous fluid accumulation. PDW signal intensity and contrast are connected to the mobile population of protons within the tissue. They have good resolution and tissue contrast and can delineate between articular cartilage and synovial fluid.
Each sequence gives different information. The signal intensity of tissue on a number of sequences needs to be ascertained in order to characterize a lesion. In assessing human fractures, a T1W SE or PDW SE is utilized for the anatomic detail it affords and a fluid‐sensitive sequence, such as a STIR or fat‐suppressed T2W SE sequence, for emphasizing contrast differences between normal and abnormal tissues. In sMRI of horses, a T1W 3D or GRE, depending upon the area, and a fluid‐sensitive sequence (ideally both STIR and T2*W) are principally employed using the same rationale.
Technical Considerations
Appreciation of artefacts is necessary in order to avoid interpretation errors. Absence of patient motion is important. Many fracture evaluations will employ sMRI, but it is necessary for horses to be sufficiently comfortable to stand square without resting pain. Immobility is essential to avoid phase mismapping and loss of image quality. The team involved in patient handling, sedation and acquisition have a substantial bearing on end image quality.
Phase cancellation or chemical shift artefact is the result of the differing precessional frequencies of protons in water and fat, caused by hydrogen in water being arranged with oxygen and hydrogen in fat being arranged with carbon. When they are in phase their signals add together, and when they are out of phase their signals cancel out. This results in a dark line at the interface of fat and water which is extremely useful in highlighting the presence of intra‐osseous fluid accumulation on T2*W GRE sequences.
Susceptibility artefact is produced by agents that disrupt the local magnetic field due to their ability to become magnetized, e.g. ferromagnetic materials or blood degradation products. This results in dephasing at the agent's interface resulting in signal loss or void and is most prominent on gradient echo sequences as the gradient reversal is unable to compensate for the phase difference. Implants also cause distortion of the magnetic field and can complicate interpretation.
Within each voxel, the signals received are averaged creating the potential for volume averaging artefacts. Increased slice thickness and the poorer resolution of sMRI exacerbate this process [140]. A common example occurs in the metacarpal/metatarsal condyles where the curvature and thin articular cartilage can be susceptible to volume averaging artefacts.
Clinical Indications
The decision to use MRI in the equine fracture patient is multifactorial, but prior regionalization of the injury is a prerequisite. Lesion location, patient comfort level and the type of system available are all determinants. In the absence of definitive radiographic findings, the commonality of fracture location in horses in training (carpus, fetlock and pastern) means that sMRI can provide a safe method to determine the presence, suspicion or absence of features supportive of a fracture (Figure 5.12). MRI has also proved beneficial in sports horses for fractures when there are discrete clinical findings, but radiographs have been negative [141] or following localization with diagnostic analgesia, again with negative radiographic and ultrasonographic findings (Figure 5.13). In addition to assisting in diagnosis, MRI also gives an insight into the health of subchondral bone [142]. When considering the bone stress injury continuum, a BML depicting stress reaction at a predilection site for an exercise‐related fracture can represent prodromal damage [88, 143]. Following the bone’s normal pathogenetic response, a discernible fracture line may, in time, become evident [144] and demonstrate a lesion that requires surgical intervention. MRI under general anaesthesia is not usually indicated in suspected equine fractures.
Limitations
The principal limitations in equine fracture detection are lesion location, acquisition time, motion artefact and low SNRs associated with STIR sequences in sMRI. The low signal intensity of normal compact bone complicates the detection of subtle non‐displaced cortical fractures [145]. This is particularly important if secondary signs of fracture such as intra‐osseous fluid accumulation are not identified. In addition, the low signal intensity of compact bone, tendon and ligament can make avulsed bone fragments difficult to identify [146]. In general, identification of any small osseous or osteochondral fragment can be difficult if the fragment is near to compact bone or intact collagen. The requirement for multiple coil placement for the evaluation of long fractures in sMRI has both time and sedation implications [145].