Forty-eight MRI-guided nerve root infiltrations were carried out in 40 patients between November 2017 and July 2022, on a 3 T MRI machine (Magnetom Skyra, Siemens Healthcare, Erlangen, Germany). As described in a former publication [22], an adapted semi-flexible surface coil was used in addition to the standard spine coil.

Study approval was obtained from the ethics committee of the CCER from Geneva (n°2019–02301) and all methods were performed in accordance with the relevant guidelines and regulations. All patients were informed about the study before the intervention and gave their written informed consent to participate in the study.

The inclusion criteria were patients with lumbar or sacral radiculopathy selected for MRI-guided lumbar or sacral nerve root infiltrations after multidisciplinary board discussion involving radiology, neurosurgery, rheumatology, orthopedic surgery, and pain relief teams. Patients were assessed clinically by a spinal surgeon and by standard musculoskeletal MRI using sagittal T1 and T2 turbo spin echo (TSE), axial T2 TSE, and sagittal T2 TSE FatSat Short-TI Inversion Recovery (STIR) sequences. Other tests such as electromyography were not routinely performed since the diagnostic value of the intervention was considered sufficiently high.

The exclusion criteria included all contraindications for MRI examinations, in particular claustrophobia and presence of metallic devices. These conditions could be ruled out already at the step of the standard before the intervention MRI mentioned above.

In this study, no pregnant women were recruited for infiltration. Therefore, we cannot provide specific advice regarding the use of the technique in this particular population.

In the first step, the optimal needle path was determined by a radiologist using an axial breathhold T2-weighted TSE sequence, repeated a second time for confirmation after placement of a small silicone marker on the chosen skin puncture point (Fig. 2). Infiltration of a 1% lidocaine solution via 25G and 22G needles was performed for superficial and deep anesthesia.

Axial T2-weighted MRI image of 43-year-old man with suspected left L5 radiculopathy in a context of congenital lumbar canal stenosis and L4–L5 discal hernia. A silicon marker (arrow) has been placed on the skin for planification of needle access (dotted line)

A breathhold PD-weighted sequence was then used to repeatedly control the advancement of a 20G or 22G MRI-compatible needle with a stylet (Cytocut MRI, MDL, Delebio, Italy) until the needle came to lie directly behind the exiting nerve root in the neural foramen or a sacral foramen.

On May 20, 2022, a 22G needle built from identical alloy was employed as an alternative to the standard 20G needle to compare the size of artifacts. The assessment criterion focused on visual artifacts, specifically evaluating the width of the dark band that artificially increased the apparent width of the needle as described in Scheffler et al. [22]. It is important to note that the 22G needle diameter is smaller than that of the 20G needle; therefore, the spatial extension of the magnetic field perturbation is proportionally reduced [27].

In case of lumbo-sacral transitional anomaly, the L5-S1 nerve root can be difficult to reach by a direct and strictly axial access. In this situation, MRI-guidance allowed for a three-dimensional planning and execution of access path (see Figs. 3 and 4 for examples from our own experience).

Non-contrast MRI scan obtained in nerve root infiltration in 24-year-old man. a Sagittal T2 turbo spin echo-weighted image showing oblique cranioventral access plane (green line) chosen to access L5 nerve within L5-S1 neural foramen. b Axial oblique proton density-weighted image showing 20G needle (white arrow) arriving behind left L5 nerve (black arrow)

Non-contrast MRI scan obtained in nerve root infiltration in 70-year-old woman. a Sagittal T2 turbo spin echo-weighted image showing oblique cranioventral access plane (green line) chosen to access L5 nerve within L5-S1 neural foramen. Note transitional lumbo-sacral vertebra. b Axial oblique proton density-weighted image showing 20G needle (white arrow) with tip (arrowhead, inset) arriving behind left L5 nerve (black arrow, inset)

Once the needle was in its final position, as close as possible to the root without touching it, a breathhold fat-saturated T2-weighted sequence was acquired. It was then followed by the injection of a small quantity of sterile normal saline solution, and then acquired again to compute a subtraction image showing the injected fluid clearly. Figure 5 shows the distribution of the saline solution around the nerve root, allowing to confirm the correct position of the infiltration needle and to estimate the direction of fluid diffusion. In addition, if the injected fluid is visible on the images, an extravascular position of the needle tip can be assumed. Table 1 shows the main acquisition parameters of the MRI sequences. In case of an excessive distortion of the fat-saturated T2 images by a total hip prosthesis, a free breathing STIR sequence was acquired instead. Finally, medication injections were performed around the nerve root.

Non-contrast MRI scan performed during a nerve root infiltration procedure in a 41-year-old woman. a Axial proton density-weighted image illustrating a 22G needle (white arrow) positioned behind the left L5 nerve (black arrow). b Axial T2 turbo spin density (TSE)-weighted fat-saturated image revealing a fluid deposit (white arrow) surrounding the needle, resulting from the local anesthetic injection along the needle's path. The nerve appears hyperintense (black arrow). c At the same level, an axial T2 TSE-weighted fat-saturated image was acquired after injecting a small amount of sterile saline solution, which highlighted a minor fluid accumulation (white arrow) around the nerve (black arrow), thereby confirming the extravascular position of the needle tip. d By subtracting images b and c, the fluid deposit (white arrow) around the nerve (black arrow) became more obvious and helpful for visualization

We generally used 1% lidocaine to anesthetize the needle path, followed by injections of 0.4% non-crystalline dexamethasone solution and 0.5% ropivacaine or bupivacaine.

Two aspects of the patient’s subjective pain were assessed. On the one hand, the locoregional pain as described in relation to the minimally invasive procedure itself was recorded for every patient, without counting the first injection for cutaneous anesthesia. On the other hand, chronic back pain evolution was the primary endpoint of this study. It was assessed a few days before the intervention, approximately 30 min after the infiltration on-site for every intervention, and then monthly for up to 5 months of follow-up.

A numeric rating scale (NRS), with 0 = no pain, from 1 to 3 = minor pain, from 4 to 6 = moderate pain, and from 7 to 10 = severe pain, was derived from the visual analog scale (VAS).

Patients received instructions from a staff member to quantitatively evaluate their pain level on the VAS at the aforementioned sampling points. During follow-up, patients were contacted by phone calls by the same staff member monthly from the first to the fifth month after the procedure, in order to record their pain level and other procedure-related information, such as the necessity for ultimate spinal surgery.

The success of the procedure was assessed by calculating the difference between the pain level experienced by the patient during the days before the procedure and the pain level experienced at each assessment by the physician during the 5 months following the procedure.