Ultrasound-guided axial facet joint interventions for chronic spinal pain: A narrative review

ABSTRACT Background Axial facet joint interventions (e.g., medial branch block and radiofrequency ablation, facet joint intra-articular injection) are commonly performed for managing chronic spinal pain. Although traditionally performed with fluoroscopy or computed tomography (CT) guidance, ultrasound-guided techniques have also been developed for these interventions. Aims The aim of this study is to present contemporary ultrasound-guided techniques for facet joint interventions and synthesize data addressing their accuracy, safety, and efficacy. Methods The PubMed, MEDLINE, CINAHL, Embase, and Cochrane Central Register of Controlled Trials databases were systematically searched for studies of ultrasound-guided facet joint interventions with human subjects from November 1, 1992, to November 1, 2022. Additional sources were drawn from reference lists and citations of relevant studies. Results We found 48 studies assessing ultrasound-guided facet joint interventions. Ultrasound guidance for injection of the cervical facet joints and their innervating nerves had favorable accuracy (78%–100%), with lower procedural time compared to fluoroscopy or CT guidance and comparable pain relief. Accuracy with ultrasound-guided lumbar facet joint intra-articular injection (86%–100%) was more reliable than medial branch block (72%–97%); analgesia was comparable to fluoroscopy and CT guidance. In general, these procedures were more challenging for patients with obesity, and deeper structures were more difficult to accurately target (e.g., lower cervical levels, L5 dorsal ramus). Conclusions Ultrasound-guided facet joint interventions continue to evolve. Some technically challenging interventions may be impractical for widespread usage or require further technical refinement. The utility of ultrasound guidance with obesity and abnormal anatomy may be reduced.


Introduction
Facet, or zygapophyseal, joints are a common cause of chronic spinal pain. Among patients with chronic spinal pain, the prevalence of facet joint-mediated pain is estimated to be 36% to 67% in the cervical spine, [1][2][3][4] 42% to 48% in the thoracic spine, 3,5 and 15% to 45% in the lumbar spine. 3,[6][7][8][9][10] Additionally, the prevalence of facet jointmediated pain increases with age. 11 As ubiquitous pain generators, facet joints are among the most common targets for therapeutic and diagnostic interventional techniques (i.e., intra-articular injection, medial branch block, and radiofrequency ablation), which may be used alongside physical therapy, self-management, and pharmacotherapy in the holistic management of chronic spinal pain. 1 Indeed, local anesthetic injection of the nerves innervating the facet joints (i.e., medial branches) is a diagnostic standard for facet joint-mediated spinal pain. 12,13 Facet joint intra-articular injection with local anesthetic also has some diagnostic utility, though it may be a relatively less reliable approach. 14 The addition of a corticosteroid may prolong the analgesic effect from these interventions; however, this is not well supported in routine practice, [15][16][17] particularly given concerns about negative systemic effects of repeated corticosteroid administration. 18 A key benefit of establishing a diagnosis of facet jointmediated spinal pain is that radiofrequency ablation may then be considered as a therapeutic intervention. 12,13 Radiofrequency ablation, which uses thermal energy to coagulate the nerves innervating the offending facet joints, may provide longer duration analgesia than either facet joint intra-articular injection or medial branch block, without need for corticosteroid administration.
Although facet joint interventions have traditionally been performed with fluoroscopic guidance, the growing availability of ultrasound imaging has facilitated the development of new techniques for managing chronic spinal pain, potentially improving accessibility, safety, and effectiveness. 19 Ultrasound guidance permits the avoidance of ionizing radiation exposure associated with traditional fluoroscopic or computed tomography (CT)-guided approaches and also facilitates real-time visualization of soft tissue and neurovascular structures around the site of intervention. The portability of ultrasound allows more interventions to be performed in the clinic setting, which is less resource intensive than the fluoroscopy suite. However, limitations of ultrasound guidance (e.g., image quality) present potential challenges during facet joint intervention.
This narrative review provides a broad overview of ultrasound-guided facet joint interventions. Its first objective is to provide a practical overview of facet joint and medial branch anatomy. The second objective is to present, based on the results of a systematic literature search, current techniques for ultrasound-guided facet joint interventions. This final objective is to synthesize data on the performance, safety, and efficacy of ultrasound-guided facet joint interventions.

Methods
To provide a comprehensive review of this topic, the authors systematically searched the PubMed, MEDLINE, CINAHL, Embase, and Cochrane Central Register of Controlled Trials databases from November 1, 1992, to November 1, 2022. The search strategy was constructed with the assistance of a medical information specialist. The following MeSH (Medical Subject Headings) were used: "zygapophyseal joint" OR "facet joint" OR "vertebrae," joined by the Boolean operator "AND" to the MeSH "interventional ultrasound" OR "ultrasonography" OR "injections" OR "pain management" OR "fluoroscopy" as well as "chronic pain" OR "back pain" OR "neck pain." For our detailed search terms, refer to Supplement 1.
Inclusion criteria included studies of ultrasoundguided facet joint interventions (i.e., intra-articular injection, medial branch block, or radiofrequency ablation) involving human subjects (i.e., randomized controlled trials, case series, observational studies, anatomical studies, and cadaveric studies). Abstracts, book chapters, editorials, and reviews were excluded, but their reference lists were examined for relevant primary literature. There was no exclusion based on language. Article screening was performed by M.W., and M.R. was consulted in the event of ambiguity. Additional sources were drawn from the reference lists and citations of relevant studies.
The methodologic quality of included studies was also assessed. Case reports were evaluated using the CARE (CAse REport) checklist. 20 Randomized controlled trials were assessed using the Cochrane Risk of Bias version 2 tool. 21 Nonrandomized, prospective observational studies were assessed using the Risk of Bias in Non-Randomized Studies of Interventions tool, and retrospective observational studies were appraised using the Newcastle-Ottawa Scale. 22 Cadaveric studies were appraised using the QUality Appraisal for Cadaveric Studies scale. 23 Anatomic (i.e., radiographic) studies were assessed with the Anatomical Quality Assessment tool. 24 This review was prepared in accordance with the Scale for the Assessment of Narrative Review Articles. 25

Results
The outcome of the literature search is summarized in Figure 1 and Table 1. We identified 48 studies that met our inclusion criteria, including 1741 human subjects in total, and over 3947 spinal levels were targeted. Among these subjects, there were 94 healthy volunteers, 1601 patients, and 46 cadavers. Nine of these studies were cadaveric studies, 2 were anatomic studies of living subjects, 3 were case reports, 16 were prospective observational studies, 7 were retrospective observational studies, 9 were randomized controlled trials, and 2 had mixed designs. 26,27 The risk of bias assessment for the included studies is summarized in Table 2. In general, most of the included studies were at moderate risk of bias (For further details, refer to Supplemental Data).
Data from the included studies are organized according to spinal region (i.e., cervical, thoracic, and lumbar), with subsections describing the performance of facet joint intra-articular injections and medial branch blocks, followed by a synthesis of the available evidence. The limitations of ultrasound guidance for performing spinal facet interventions are also reviewed, and practical recommendations are provided.

Facet Joint Anatomy
Facet joints are paired joints formed by the superior articular process of one vertebra and the inferior articular process of the level above. 29 The articular facets of these joints are covered with hyaline cartilage and enclosed in a synovial capsule, with the total volume of each facet joint being approximately 1 mL. Facet joints provide axial stability and define the spine's range of motion at each region. Facet joint degenerative changes may involve bony hypertrophy, loss of cartilage and synovial fluid, and associated inflammation, which may all drive spinal pain. 30 However, incidental and asymptomatic facet joint degeneration is also common, 31 and    joint capsular disruption may also cause pain in the absence of obvious radiographic changes. 32 Facet joint sensory innervation is provided by the medial branches, which are terminal divisions of each nerve root's dorsal ramus ( Figure 2). [33][34][35][36] In general, each facet joint receives dual innervation, from the same level and also the level above (e.g., the C3-C4 facet joint is innervated by the C3 and C4 medial branches, and the L2-L3 facet joint is innervated by the L1 and L2 medial branches); therefore, both contributing medial branches must be targeted to block sensation for a given facet joint. The localization of the medial branches varies depending on the region of the spine. 29,37,38 Most cervical medial branches are found on the lateral waist of their respective vertebrae's articular pillars. However, the superficial medial branch of C3 (third occipital nerve; TON) crosses the lateral surface of the C2-C3 joint, which it innervates, and the C7 medial branch is found at the junction of the C7 superior articular process and transverse process. Medial branches arising from T1 to T4 and T9 to T10 cross the superolateral margins of the transverse process below and then continue inferomedially. From T5 to T8, the medial branches are suspended just superior to the transverse process, in the intertransverse space. The T11 and T12 medial branches follow a course similar to that of the lumbar medial branches, which reliably pass over the junction of the transverse process and superior articular process of the level below. The L5-S1 facet joint is unique in that its innervation is thought to arise from the L5 dorsal ramus itself, rather than a discrete medial branch. 29

Ultrasound-Guided Interventions
The most common procedures performed on facet joints include intra-articular injections, blocks of the nerves innervating the joints, and their denervation using radiofrequency ablation. Cryoablation and pulsed radiofrequency ablation of these structures are also described, albeit rarely. 39,40 Though facet joint intra-articular injection has clinically been used for both diagnostic and therapeutic purposes, the gold standard for diagnosing facet joint pain is medial branch block, typically using local anesthetic volumes of 0.5 mL or less. 12,13 Usually, a high-frequency linear ultrasound transducer is used for superficial structures (i.e., cervical or thoracic facet joints), whereas a low-frequency curvilinear transducer is better suited for deeper targets (i.e., lumbar facet joints).

Cervical Facet Joint Intra-Articular Injection
Ultrasound-guided cervical facet joint intra-articular injection was initially described by Galiano and colleagues with the patient lying in decubitus ( Figure  3a). 41 With the transducer oriented coronally on the lateral neck, the facet column appears as a characteristic wavy hyperechoic line ("sawtooth pattern"; Figure 3b), and the articular pillars appear as troughs and the facet joints as peaks.
Several characteristic structures can be used to help identify the target level. The C2 inferior articular process has a distinctive drop-off before the C1 transverse process appears slightly superior to it (Figure 3c). The C2-C3 facet joint is also inferior and slightly anterior to the mastoid process, and anterosuperior to that is the vertebral artery. The levels may also be identified with the transducer placed transversely in midline to view the spinous processes. C1 is immediately caudal to the occiput and has, at most, a rudimentary spinous process, whereas C2 has a prominent bifid spinous process. Inferiorly, the vertebral levels can be verified by viewing the transverse processes of C5, C6, and C7 (Figure 3d-f). The C7 transverse process has a rudimentary anterior tubercle, the C6 transverse process has a very prominent anterior tubercle, and the C5 transverse process has more equally sized tubercles.
With the target facet joint visualized, the transducer is typically rotated transversely to allow in-plane needle placement with a posterior-to-anterior trajectory (Figure 3a,g). Alternatively, a posterior approach may be considered with the patient positioned prone ( Figure  3h,i), which allows bilateral injections to be performed without repositioning the patient. At the target level, an in-plane inferior-to-superior needle trajectory is used.

Cervical Medial Branch Block
Eichenberger et al. introduced the ultrasound-guided technique for blocking the TON. 42 The patient is positioned lateral decubitus. The lateral aspect of the neck is first scanned at the level of the mastoid, with the transducer in a coronal plane. The TON itself may be visible at the level of the C2-C3 joint, which is localized as described above. However, to perform the block, TON visualization is not strictly necessary. After identifying the C2-C3 joint space, the transducer is rotated transversely, and the needle is inserted in-plane from a posterolateral entry point until periosteum is contacted. If desired, the needle tip position may be confirmed before injection by rotating the transducer coronally again to demonstrate the needle tip by the TON and C2-C3 facet.
From C3 to C6, the medial branches are targeted at the center of their respective articular pillars, seen with the transducer oriented coronally on the lateral neck (Figure 3b). The medial branches may themselves be visible on the articular pillars. At the desired level, the  transducer is rotated transversely and the needle is advanced in-plane from a posterolateral entry point until the articular pillar is contacted (Figure 3g). Note that the cervical articular processes are located posterior to the posterior tubercules. The lower levels of C5 and C6 are technically challenging due to increased target depth in the base of the neck; therefore, it may be especially helpful to dynamically scan the anatomy and verify needle tip placement with biplanar imaging.
For TON and C3 to C6 medial branch blocks alike, an alternative approach has also been described, where the target is identified from the coronal transducer orientation and injected with an out-of-plane approach, without rotating the transducer to a transverse position for an in-plane trajectory.
The C7 medial branch requires a different technique for intervention due to the unique anatomy of the C7 vertebra. 34 Ultrasound transducer positioning may also be impeded by the clavicle. 43 The C7 and T1 transverse processes are first located by scanning with the transducer transversely oriented at the lower part of the lateral neck. The target is the C7 superior articular process, immediately posterior to the C7 transverse process (Figure 3j). Alternatively, if the C7 superior articular process is not apparent, the part of the C7 transverse process immediately caudal to the C6-C7 facet joint is targeted. For either target, an in-plane needle trajectory is used with a posterolateral insertion site. Needle tip placement may be confirmed by scanning with the transducer in a coronal orientation.
For facet joint interventions at the cervical level, care must be taken to maintain strict control of the needle tip during manipulation. The cervical nerve roots are in close proximity to the cervical facet joints, as are numerous important vascular structures (i.e., vertebral artery, superficial and deep cervical arteries, inferior thyroid artery). Thorough scanning of the local sonoanatomy may be helpful for procedural planning given a high incidence of blood vessels overlying structures of interest for cervical facet joint interventions. 44

Evidence for Cervical Facet Interventions
Seventeen of the included studies were focused on cervical facet joint interventions, of which five were observational studies and five were randomized controlled trials (Table 1). Ten studies examined interventions targeting the nerves innervating the facet joints (i.e., TON, medial branches), and five studies assessed facet joint intra-articular injection. Notably, there were two studies of ultrasound-guided cervical medial branch radiofrequency ablation, one in cadavers and another in patients.
Performance-Related Outcomes. The accuracy of ultrasound-guided cervical facet joint intra-articular injection ranges from 78% to 100%, 41,45-47 with a potentially higher failure rate at the more challenging lower levels. 46 One randomized trial compared ultrasound-guided facet joint intra-articular injection with CT guidance in 40 patients with facet-mediated midto low-cervical spine pain and reported that ultrasound guidance had superior accuracy on first attempt (100% versus 35%) and had a shorter procedural time (6 min versus 14 min). 47 Similarly, a retrospective observational study found that ultrasound guidance for medial branch block required less procedural time (221 s versus 383 s) and fewer needle passes (two versus five), compared to fluoroscopy. 48 Finlayson and colleagues conducted a series of comparative studies examining ultrasound-guided local anesthetic injection of nerves innervating the cervical facet joints; they reported comparable accuracy to fluoroscopy-guided injection (95%-100%), as determined by fluoroscopic confirmation of needle tip position. 34,[49][50][51] Other investigators had somewhat lower accuracy with medial branch and TON injection, according to fluoroscopic confirmation (78%-82%). 42,46,52 In a cadaver study, Lee and colleagues reported 100% successful ultrasound-guided radiofrequency cannula placement as verified by fluoroscopy, and dissection revealed successful ablation in 30 of 34 medial branches targeted; C6 and C7 were the only levels where medial branches were unsuccessfully coagulated. 53 In an observational study, ultrasound guidance was found to have a lower procedural time for radiofrequency ablation compared to CT guidance (10 min versus 14 min). 54 Safety. There was one case report of spinal cord injury following ultrasound-guided C7 medial branch block, with persistent neurologic deficits after 1 month. 55 Otherwise, there were no major complications observed in any other clinical studies. Transient minor adverse effects were infrequently observed (e.g., vasovagal reaction, pain exacerbation). In an anatomic study of 102 patients with chronic neck pain, 24% of cervical levels were found to have incidental blood vessels in the vicinity of the cervical medial branches. 44 Some studies reported a 10% to 20% rate of blood aspiration with fluoroscopy-guided medial branch or TON block, though this did not result in patient morbidity. 34,48,50 Blood aspiration was not noted with ultrasound guidance in any study.

Efficacy.
In observational studies and randomized controlled trials of ultrasound-guided injection of cervical facet joints and the nerves that innervate them, ultrasound guidance produced comparable reductions in pain and disability scores when compared with fluoroscopy or CT guidance. 40,42,[47][48][49][50]54 In a retrospective study of 126 patients, ultrasound-and fluoroscopyguided cervical medial branch injection with corticosteroid produced comparable reductions in pain severity and disability, lasting at least 6 months. 48 In the only study comparing ultrasound-and fluoroscopy-guided cervical radiofrequency ablation, 53 Awad et al. found comparable analgesia at one month followup. 54 Pulsed radiofrequency ablation of the TON has also been reported, with analgesic benefit up to 12 months. 40

Thoracic Facet Joint Intra-Articular Injection
Ultrasound-guided thoracic facet joint intra-articular injection was first described by Stulc and colleagues. 56 The patient is positioned prone (Figure 4a). To identify the correct level, the 12th rib is first visualized by scanning a sagitally oriented transducer from a caudal-to-cranial direction along a vertical line inferior to the medial border of the scapula. The 12th rib is then followed medially to first show the T12 costotransverse articulation, T12 transverse process (Figure 4b), and then the T12 lamina. The facet joint is seen as a hypoechoic cleft between the respective hyperechoic laminae and articular processes ( Figure  4c) and is bounded by the spinous process in the midline. The medial and lateral borders of the facet joint are identified by sweeping the transducer medially and laterally. The mid-point of the joint is typically targeted, using an inplane caudad-to-cephalad trajectory. Subsequent facet joint levels can be counted by scanning superiorly from T12.

Thoracic Medial Branch Block
A technique for ultrasound-guided thoracic medial branch blocks was suggested by Moon et al., though their method has not been validated. 57 The variable course of the medial branch at different levels of the thoracic vertebrae necessarily entails different bony targets depending on the region of interest ( Figure 2). The target level is identified as above, and the transducer is oriented transversely to demonstrate the costotransverse junction (Figure 4a,d). From a lateral-to-medial trajectory, the needle is directed to the periosteum at the superolateral edge of the transverse process, which is the target for T1 through T4, along with T9 and T10.
From T5 to T8, the medial branches do not touch the transverse process but are suspended in the intertransverse space. So, from the superolateral edge of the transverse process, the needle is slightly withdrawn and redirected slightly cephalad. To avoid breaching the pleura, the needle tip is advanced only to a comparable depth as the superficial edge of the transverse process.
The T11 and T12 medial branches have similar anatomical characteristics as the lumbar medial branches, described below.

Evidence for Thoracic Facet Interventions
The systematic literature search found one study meeting inclusion criteria. 56 Performance-Related outcomes. Stulc and colleagues performed 20 ultrasound-guided thoracic facet intraarticular injections on a single cadaver, with an accuracy rate of 80%, as per CT imaging. 56 Safety. No studies reported safety outcomes for ultrasound-guided thoracic facet joint interventions.

Efficacy.
No studies reported efficacy outcomes for ultrasound-guided thoracic facet joint interventions.

Lumbar Facet Joint Intra-Articular Injection
Ultrasound-guided lumbar facet joint intra-articular injection was introduced almost two decades ago. 58,59 The patient is positioned prone (Figure 5a). The transducer is placed on the lumbar region in sagittal orientation. The most lateral structures seen are the transverse processes, in a characteristic "trident" configuration (Figure 5b). The transducer is swept medially, first showing the facet joints (i.e., continuous "camel hump" pattern; Figure 5c) and then spinous processes in the midline. By sliding the transducer caudad, the lumbosacral junction can be dynamically visualized, permitting identification of the desired levels ( Figure 5d). The sacrum is a curved, hyperechoic surface appearing continuous except at the sacral foramina. The L5-S1 facet joint is immediately cephalad to the sacral outline, and further levels are counted upwards sequentially.
At the target level, the transducer is rotated transversely to simultaneously show one level's spinous process, lamina, articular facets, and transverse processes. Rocking the transducer reveals the cleft between the joint's superior articular process and the inferior articular process (Figure 5e). The needle is advanced in a lateral-to-medial, in-plane trajectory. To verify needle tip placement, the transducer may be rotated sagitally to show the needle tip lying on the middle portion of the facet joint.

Lumbar Medial Branch Block
The L1 to L4 medial branches are consistently found at the junction of the transverse process and the superior Scanning from lateral to medial (a; pink arrow), the ribs may be seen transitioning to the thoracic facet joints, then the laminae and facet joints, and finally the spinous processes in midline. The in-plane inferior-to-superior needle trajectory for (a) thoracic intra-articular injection is shown with a blue arrow. The transverse transducer position and in-plane lateral-to-medial needle trajectory is also shown for (b) thoracic medial branch block, at the superolateral aspect of the desired level's transverse process. (b) Thoracic transverse process, as seen with an ultrasound transducer in saggital orientation. (c) Thoracic facet joints, as seen with an ultrasound transducer in saggital orientation. For intra-articular injection, the needle (blue arrow) is advanced inplane to the target from an inferior-to-superior trajectory. (d) Thoracic medial branch block. With an ultrasound transducer in transverse orientation, the needle (blue arrow) is advanced in-plane to the target from a lateral-to-medial trajectory. For T5 to T8 medial branch block, the transducer is swept just superior to this view (not shown) and the needle is advance in-plane from a lateral-tomedial trajectory to intertransverse space, at a depth no deeper than the surface of the adjacent transverse processes. ES = erector spinae muscles; IAP = inferior articular process; SAP = superior articular process. articular process, which may be found with the transducer oriented transversely at the desired level. The junction appears as a step-like shadow deep and lateral to the facet joint and is targeted via an in-plane lateral-tomedial trajectory (Figure 5e). Rotating the transducer to a sagittal orientation, needle tip placement may be confirmed at the superior aspect of the transverse process prior to injection.
The L5 dorsal ramus contributes to innervation of the L5-S1 facet joint, which is the most commonly involved level in lumbar facet joint-mediated spinal pain. The L5 dorsal ramus, however, is particularly challenging to block due to its depth, obscuration by the iliac crest, and variable sacral anatomy. Greher and colleagues described an early technique to inject this target. 60 With sagittal transducer orientation, the L5 transverse process and the hyperechoic sacrum are first identified. The transducer then is rotated obliquely, almost 90°, with the medial part more cranial than the lateral part (Figure 5f). In this view, the iliac crest is the most lateral structure. Looking medially, the sacral ala is seen along with the S1 superior articular process; the junction of these two structures is targeted with an in-plane trajectory. Scanning from lateral to medial (a; pink arrow), with the transducer in sagittal orientation, the transverse processes ("trident" sign) may be seen transitioning to the lumbar facet joints ("camel hump" sign), then the laminae, and finally the spinous processes in midline. The sagittal transducer position (b) for examining the lumbosacral junction is also shown. The in-plane lateral-to-medial needle trajectory (c) for lumbar intra-articular injection and medial branch block is shown with a blue arrow. Two methods for targeting the L5 dorsal ramus are demonstrated, via an (d) oblique approach and and via an (e) out-of-plane technique. (b) Lumbar transverse processes, as seen with an ultrasound transducer in saggital orientation. The "trident" sign shows the transverse processes as hyperechoic outlines with shadows where the ultrasound beam cannot pass through. (c) Lumbar facet joints, as seen with an ultrasound transducer in saggital orientation. The superior and inferior articular processes of the facet joints form a continuous hyperechoic "camel hump" line. (d) Lumbosacral junction, as seen with an ultrasound transducer in saggital orientation. In midline, the sacrum is viewed as a continuous hyperechoic line. Superior to it are the L5 spinous process and the L5-S1 interspace. Also visualized are the anterior complex (comprising the posterior aspect of the vertebral body, posterior longitudinal ligament, and anterior dura) and posterior complex (comprising the ligamentum flavum and posterior dura), which surround the intrathecal space. (e) L4 vertebra as seen with an ultrasound transducer in transverse orientation. The in-plane lateral-to-medial needle trajectories are shown for (a) facet intra-articular injection and (b) medial branch block. (f) L5 dorsal ramus block with oblique ultrasound transducer positioning. The needle (pink dot) is advanced out-of-plane with an superolateral-to-inferomedial trajectory. (g) L5 dorsal ramus block, using a "pivot" technique immediately after injecting the L4 medial branch. The ultrasound transducer is placed in a sagittal oblique position to show the L3, L4, L5 transverse processes along with the sacral ala. The needle (pink dot) is advanced out-of-plane to the superior aspect of the sacral ala, by incrementally moving the needle tip from the initial position at the L4 medial branch, at the junction of the L5 transverse process and superior articular process. ES = erector spinae muscles; IAP = inferior articular process; P = psoas muscle; SAP = superior articular process.
An alternative "pivot" technique for L5 dorsal ramus block begins with the transducer oriented transversely, as seen while targeting the L4 medial branch as described above. 61 Next, the transducer is rotated sagittally to view the L3-L4 to L5-S1 facet joints and then is swept slightly laterally to show the L3 to L5 transverse processes and sacral ala (Figure 5g). From the junction of the L5 transverse process and superior articular process, the needle is progressively redirected ("walked") inferiorly and medially down to its target, the junction of the sacral ala and superior articular process. To avoid inadvertently advancing the needle into the L5 foramen, the needle tip must not advance deeper than a line connecting the L5 transverse process and sacral ala.

Evidence for Lumbar Facet Interventions
Thirty of the included studies were focused on facet joint interventions, of which 18 were observational studies and 4 were randomized controlled trials (Table 1). Two studies used a combination of different methodological designs. 26,27 Fourteen studies examined interventions targeting nerves innervating the facet joints (i.e., medial branches, L5 dorsal ramus), and 16 studies assessed facet joint intra-articular injection. There was 1 study of ultrasound-guided medial branch radiofrequency ablation in cadavers 62 and 1 study of ultrasound-guided medial branch cryoneurolysis in patients. 39 Performance-Related outcomes. Based on fluoroscopic confirmation of needle tip placement, the success rate for ultrasound-guided facet joint intra-articular injection ranged from 86% to 100%. 58,59,[63][64][65] There was not a consistent advantage in procedural time when ultrasound was compared to fluoroscopy or CT guidance. -59,62,66-68 In one randomized controlled trial, ultrasound-guided facet joint intra-articular injection was found to be more accurate (86% versus 31%) than a landmark-based technique (i.e., needle entry site at predefined distance from palpated spinous process). 65 The accuracy for needle tip placement in ultrasoundguided lumbar medial branch intervention was highly variable in both clinical and cadaver studies. In these studies, patients with obesity were often excluded. One observational study, focused on patients with body mass index over 30, reported 62% accuracy on fluoroscopy confirmation and concluded that ultrasound guidance alone in this population is unreliable. 69 In other studies, the accuracy rate ranged from 72% to 97%. [26][27][28]39,60,61,63,64,[69][70][71][72][73][74][75][76] The L5 dorsal ramus was identified as a particularly challenging target, owing to its unique and more variable anatomy, -60,69,70 and techniques for reaching this target continue to be developed. 60,61 One cadaver study of ultrasound-guided lumbar radiofrequency ablation, using a sophisticated magnetic needle localization system, demonstrated 97% accuracy on fluoroscopy. 62 Safety. There were no major complications observed. Transient minor adverse effects were infrequently observed (e.g., vasovagal reaction, superficial hematoma, pain exacerbation). One observational study reported blood aspiration during 7% of fluoroscopy-guided medial branch injections, without subsequent sequelae. 74 Blood aspiration was not reported during any ultrasound-guided interventions.
Efficacy. Observational studies [77][78][79] and randomized trials 26,58,64,66,80,81 alike attest to comparable reduction in pain scores and disability between ultrasound-and fluoroscopy-guided lumbar facet intra-articular injections with corticosteroid; these benefits persisted through the follow-up period of each study, which tended to be 3 months or shorter. Two randomized trials of lumbar facet joint intra-articular corticosteroid injection compared ultrasound guidance to landmark-based techniques (e.g., needle entry site at a predefined distance from palpated spinous process); one of these studies demonstrated improved pain reduction with ultrasound guidance up to 6-week follow-up, 81 but the other study found that the superiority of ultrasound guidance did not persist past the immediate postprocedural period. 64 Though medial branch block is more often used diagnostically rather than therapeutically, some authors reported prolonged analgesia and functional improvement after both ultrasound-and fluoroscopy-guided lumbar medial branch injection with corticosteroid, lasting weeks to months. 71,73 A case series of satisfactory ultrasound-guided lumbar medial branch cryoneurolysis was reported, with up to 12-month follow-up. 38

Future Developments in Ultrasound Guidance
Several studies may be highlighted for their use of novel technology, including fusion imaging (i.e., real-time ultrasound coupled with prior CT or magnetic resonance imaging data) 65,67,82 and magnetic needle tip tracking. 61 Sophisticated image guidance systems hold promise to improve the accuracy of ultrasound-guided medial branch targeting, which may permit blockade or radiofrequency ablation. However, few data exist to inform the use of such technology, and further investigation is required before widespread adoption is considered.

Discussion
This systematic literature search revealed a diverse and rich body of human studies concerning ultrasoundguided facet joint interventions. Cervical facet joint intra-articular injection and medial branch or TON block were particularly amenable to the modality, with favorable accuracy (78%-100%), lower procedural time compared to fluoroscopy or CT guidance, and comparable pain relief. Ultrasound guidance provided excellent accuracy for lumbar facet joint intra-articular injection (86%-100%), whereas medial branch and dorsal ramus block had more variable accuracy (72%-97%); the analgesic effect was comparable to that obtained with fluoroscopy and CT guidance.

Limitations of Ultrasound-Guided Facet Joint Interventions
Although ultrasound presents an opportunity to refine techniques for managing facet joint-mediated spinal pain, enthusiasm for this imaging modality should be tempered with a realistic understanding of its limitations.

Limitations Relative to Fluoroscopy
Compared to fluoroscopy, ultrasound-guided interventions may be more challenging when deeper targets affect needle tip visualization (i.e., lumbar or lower cervical levels). Though some degree of error may be reasonable for facet joint intra-articular injection, the diagnostic specificity of the medial branch block depends on accurate needle tip placement, given that minute local anesthetic volumes are administered. Whereas fluoroscopy permits evaluation of appropriate contrast spread during medial branch block and can exclude intravascular injection, this is not possible using ultrasound alone. Indeed, a recent meta-analysis reported an 11% to 13% absolute risk increase for incorrect needle tip placement using ultrasound compared to fluoroscopic guidance. 83 Additionally, care must be taken to carefully identify the desired spinal levels for intervention. Fluoroscopy or CT is well suited to showing a wide field of view, but ultrasound displays a relatively limited area, which risks targeting the wrong level. 43,50 Patients with transitional lumbar anatomy (e.g., sacralized L5) are particularly at risk of misidentification of spinal levels using ultrasound. 13,60 Training Requirements Interestingly, one study described a statistical model for the acquisition of ultrasound-guided lumbar medial branch block proficiency ("learning curve") by experienced regional anesthesiologists who did not have prior experience in interventional pain medicine. The model estimated that the procedure would need to be performed more than 47 times to achieve an 85% success rate in the technique for nonobese patients. 76 This is a challenging learning curve compared to some other ultrasound-guided interventions (i.e., approximately 30 injections to become proficient in sacroiliac joint intraarticular injection). 84 Ultimately, some clinicians may find the reliability of ultrasound-guided lumbar medial branch block to be unacceptable compared to fluoroscopic guidance using well-established and consistent radiographic landmarks for targeting these nerves. 83

Safety Considerations
Real-time ultrasound has been proposed to theoretically reduce the risk of injury to cervical vascular structures. 85 Incidentally observed blood vessels may often be found around the lower cervical articular pillars. 43,48 Nonetheless, no specific safety benefit has been conclusively demonstrated in well-powered reports and, in general, data on clinical outcomes after ultrasoundguided cervical interventions are limited. 86 Additionally, identified cervical vessels may still be at risk of injury if needle tip visualization is poor during needle manipulation. Errant needle tip movement may also risk nerve root or spinal cord injury; our systematic literature search revealed one case report of spinal cord compression due to hematoma following ultrasoundguided C7 medial branch block. 54 Ultimately, the effectiveness and safety of ultrasound-guided interventions remain highly operator dependent.

The Choice of Imaging Modality
Based on our literature search, ultrasound appears to be fairly comparable fluoroscopy or CT guidance for cervical and lumbar facet joint intra-articular injection, at least with respect to accuracy, safety, and clinical efficacy. Though medial branch block appears technically feasible, the higher failure rate for deeper structures (e. g., lumbar or lower cervical regions) may result in more false-negative diagnostic injections and risk unnecessarily precluding otherwise appropriate patients from accessing radiofrequency ablation. Negative ultrasound-guided medial branch blocks may need to be repeated to reduce this risk of false-negative results influencing management; however, this extra step may mitigate ultrasound's potential benefits of improved access and convenience.
Anatomical factors may also play a role in the choice of imaging modality. For patients with obesity, many structures may be deeper and more difficult to visualize. Additionally, ultrasound-guided facet joint interventions have not been well studied in the presence of spinal instrumentation or unusual anatomy (e.g., severe scoliosis), given that these were exclusion criteria in numerous studies. In these scenarios, it would be worth considering fluoroscopy or CT guidance.
With the exception of one prospective study finding similar efficacy and safety for ultrasound-and CTguided cervical medial branch radiofrequency ablation, 53 the study of ultrasound guidance for medial branch radiofrequency ablation has generally been limited to small studies of cadaveric specimens. 52,63 Additionally, radiofrequency ablation requires the needle tip to be adjacent and parallel to the target medial branch along bony structures, which is generally considered more feasible with fluoroscopy guidance. 12,13,86 However, there may yet be a role for ultrasound to improve the safety of fluoroscopic-or CT-guided cervical interventions, especially around the cervical facet joints where incidental blood vessels are commonly observed. 33,43,50 A preprocedural ultrasound scan could reveal vulnerable blood vessels that would not be otherwise seen on fluoroscopy or CT and aid in planning the approach for needle advancement.
Beyond clinical considerations, the choice of imaging modality is influenced by external factors. For instance, in the United States, numerous insurance companies require fluoroscopic or CT guidance for reimbursement of facet joint interventions. 13 Such restrictions potentially limit the use and ongoing refinement of ultrasound guidance for these interventions.

Areas for Further Study
This systematic literature search highlights some substantial gaps. For example, there is a conspicuous lack of clinical studies assessing ultrasound-guided thoracic facet joint interventions with respect to procedural outcomes, safety, and efficacy. In general, there is a lack of clinical studies examining thoracic facet joint interventions with any imaging modality, 87 even though a substantial proportion of thoracic spinal pain is facet joint mediated.
Additionally, ultrasound-guided radiofrequency ablation has been infrequently studied. Ultrasound guidance generally lacks the ability to precisely position a radiofrequency cannula parallel to the path of nerves innervating the facet joints, as currently recommended. 12,13 Yet, this limitation may potentially be overcome with further procedural refinement and technological advances (e.g., fusion imaging). Interestingly, an ex vivo study reported that the burn characteristics of certain multitined radiofrequency cannulas were potentially favorable for cervical medial branch radiofrequency ablation; however, such a technique has not been clinically studied. 88

Conclusion
Techniques for ultrasound-guided facet joint interventions have been developed and improved over the past two decades. Desirable accuracy, safety, and efficacy have been observed in some applications (e.g., lumbar facet joint intra-articular injection). However, some ultrasoundguided techniques remain challenging or impractical (e.g., C7 medial branch or L5 dorsal ramus block), and obesity may present a substantial challenge. For certain interventions (e.g., radiofrequency ablation, thoracic facet joint interventions), there is a paucity of literature to inform practice.

Disclosure Statement
No potential conflict of interest was reported by the author(s).