Usage of augmented reality for interventional neuraxial procedures: A phantom-based study : European Journal of Anaesthesiology | EJA

Secondary Logo

Journal Logo

Regional anaesthesia

Usage of augmented reality for interventional neuraxial procedures

A phantom-based study

Reinacher, Peter Christoph; Cimniak, Anina; Demerath, Theo; Schallner, Nils

Author Information
European Journal of Anaesthesiology 40(2):p 121-129, February 2023. | DOI: 10.1097/EJA.0000000000001755
  • Open

Abstract

KEY POINTS

  • In a randomised study using a phantom spine, the use of augmented reality proved to be technically feasible.
  • In the same phantom model, the use of augmented reality improved the success of epidural access as the primary outcome and accuracy and procedure time as secondary outcomes.
  • Future studies with further technical development should investigate whether these finding might translate into relevant clinical advantages.
  • With further technical development, augmented reality might prove useful in anatomically challenging neuraxial procedures.

Introduction

Neuraxial procedures in anaesthesia and interventional pain medicine are widely used for peri-operative and posttrauma pain management and the treatment of acute and chronic pain conditions. Additionally, dural puncture for diagnostic and therapeutic purposes is routinely performed in neurosurgery and neuroradiology. The efficacy and success of these interventional procedures relies on the correct identification of the respective anatomical structures such as the epidural or the subarachnoid space (Fig. 1). To date, the usual standard approach relies on anatomical landmark techniques with varying rates of success, duration and efficacy, given the anatomical heterogeneity in patients. In certain cases, CT guidance may be employed to facilitate identification of the anatomical landmarks. However, the related radiation exposure and the high complexity render it an approach for unique cases only. During the last few years, there has been a growing utilisation of ultrasound to facilitate needle placement and for preprocedural identification of correct anatomical structures. Indeed, several studies report higher rates of success, fewer attempts and less needle redirections with the use of preprocedural ultrasound.1–7 However, the beneficial effects of real-time ultrasound use during the actual puncture remain minimal, mainly because of poor visualisation and needle guidance.8,9

F1
Fig. 1:
Anatomy of the epidural space

Augmented reality can be defined as the integration of computer-generated, virtual data into the user's ‘real world’ in real-time. Still undergoing fast technical evolution, its use in clinical scenarios remains experimental. The usability of augmented reality has been described using patients’ imaging data for planning purposes and in preclinical phantom studies including spine surgery, neurosurgical navigation and stereotactic procedures.10–14 Currently subject to scientific evaluation, emerging studies indicate possible advantages when augmented reality is used, for example, to enhance ultrasound images for needle guidance during neuraxial procedures.15 Newly developed portable eyewear-like augmented reality devices such as the Magic Leap (Magic Leap, Plantation, Florida, USA), HoloLens (Microsoft, Redmond, Washington, USA) or Moverio (Epson, Seiko Epson Corporation, Nagano, Japan) provide hardware tools to improve the clinical usability of augmented reality for neuraxial procedures, as they can be worn by the operator during a procedure, providing superimposed images to improve visualisation of target structures and procedural success.

In this technical proof-of-concept study, using a phantom of the lower thoracic and lumbar spine, we investigated whether the use of a portable augmented reality device could be used to improve success in neuraxial interventional procedures as the primary outcome. In addition, we analysed the accuracy and the procedural duration as secondary outcome measures.

Methods

Model for neuraxial puncture

As the primary outcome, we aimed to assess whether the success of neuraxial access as defined below could be improved using the augmented reality-guided technique. The accuracy and the procedural duration were defined as secondary outcomes. To study this, a custom-made lower thoracic (T9 to T12) and lumbar spine with phantom iliac crest (Fig. 2a) consisting of X-ray opaque bone structures surrounded by elastic foam and surgical skin (Fig. 2b) was used (Misstrainer Spine Surgery, Creaplast Medical, Verton, France). Bony structures of the spine for the conventional landmark approach such as the spinous processes and the iliac crest were palpable through the modelled soft tissue. The phantom was placed in prone position and a computed tomography scan (CT scan) was acquired (tube voltage 130 kV, exposure 36 mAs, tilt 0°, slice thickness 0.75 mm, helical mode; Siemens Somatom Scope, Siemens Healthcare, Erlangen, Germany). The digital imaging (DICOM) file was transferred to the stereotactic planning workstation (Elements, Brainlab, Munich, Germany). Ideal trajectories for access to the epidural and subarachnoid space on the levels of T10/11, T11/12 and T12/L1 using the left and right paramedian approach were defined by a stereotactic neurosurgeon. The target point was set at the centre of the epidural space at the respective spinal level. The trajectories were transferred into a three-dimensional (3D) dataset of the phantom spine and the dataset with the superimposed trajectories was displayed on the eyewear-like augmented reality headset (Magic Leap 1) worn by the operators under study (‘probands’), consisting of two anaesthesiologists with different levels of experience with epidural and spinal anaesthesia (senior physician, third-year resident), one neuroradiologist experienced in spinal puncture for myelographies and one neurosurgeon with experience in interventional procedures at the spinal level. Each operator performed 20 punctures of the epidural space using the paramedian approach. The paramedian approach was chosen in order to directly compare both techniques for a scenario where the preferred access to the epidural space followed the direct and ideal augmented reality trajectory. The levels (T10/11, T11/12 and T12/L1) and the side (paramedian left vs. right) were assigned for each puncture using a randomisation algorithm based on atmospheric noise (random.org). A 18-G, 90 mm Tuohy needle (Pajunk, Geisingen, Germany) was used for all punctures. Ten punctures were performed using augmented reality, whereas the other 10 punctures were carried out using the conventional anatomical landmark technique. For the latter, the probands had to use existing landmarks (iliac crest and spinous processes) to identify the assigned level. For the augmented reality-guided approach using the headset, a hologram of the phantom's 3D-image displayed on the augmented reality headset was aligned with the real phantom placed in prone position on the CT table (Fig. 3a). The probands then had to perform the puncture along the virtual ideal trajectory displayed at the correct level and side. Needle correction was allowed only during the process of puncture but not after the supposed final position had been reached. As in-vivo tissue characteristics allowing confirmation of needle insertion into the epidural space (loss of resistance and hanging drop) were absent in the phantom model, the specific insertion depth for each level and side along the individual ideal trajectory was stipulated to the probands. Specifically, the depth to reach the epidural space from the phantom's outer surface was indicated using the depth measured along the ideal trajectory from the 3D-image (see Fig. 2). For the conventional landmark approach, the probands performed the punctures without the augmented reality headset using the usual anatomical landmarks to identify the correct level and site of needle insertion. Correction of needle direction was only allowed during the process of puncture but not once either bony structures or the presumed final position had been reached. Specifically, the angle of puncture during the conventional approach was allowed to be corrected during the initial phase of the puncture but not once the probands experienced bony contact or were in the position that they determined as being the final epidural position. After the needle was placed in the final position, a CT scan was performed to analyse the needle position and whether the correct spinal level was reached (Fig. 3b). To avoid unwanted effects of habituation, the sequence of both augmented reality vs. conventional puncture and the spinal level/side were randomised for each proband. The procedural time was defined starting from approaching the phantom with the needle until final needle placement was recorded. The CT scans were transferred to the stereotactic planning workstation to measure the deviation of the needle tip position from the predefined target points as the Euclidean distance (Euclidean distance = shortest distance between needle tip and target point, Fig. 3b; middle image, orange line) and the lateral deviation of the trajectory performed by the puncture from the predefined ideal trajectory at the level of the target point (lateral deviation = distance between ideal trajectory and actual needle trajectory at the target point, Fig. 3b; left and middle image, blue line). ‘Success’ was defined as a puncture along a trajectory that reached the epidural space without bony contact. Data acquisition took place between 29 January 2021 and 14 March 2021. Analysis was completed by 15th October 2021.

F2
Fig. 2:
Thoracolumbar spine phantom
F3
Fig. 3:
Augmented reality-guided puncture of the epidural space and control computed tomography scan

Study approval

No patients were involved in this study. Therefore, no ethical approval was required.

Statistics

The data were analysed with a computerised statistics program (GraphPad Prism Version 9). Data are presented as median with [interquartile range] and 95% confidence intervals (95% CI) of the median. Truncated violin plots were used to visualise true dispersion of the data points. Normality of data was tested using Shapiro–Wilk test. As data were not normally distributed, two groups were statistically compared using Mann–Whitney U test for nonparametric comparison. The rate of success was compared using a contingency table and Fisher's exact test for nonparametric comparison of ratios. A P value smaller than 0.05 was considered to be statistically significant.

Results

Descriptive data on punctures carried out

Table 1 provides descriptive data on the sequence of the actual levels and sides at which the punctures were conducted.

Table 1 - Sequence of epidural puncture
Proband 1 2 3 4
1 T11/12 r, conv T12/L1 l, conv T10/11 r, conv T11/12 r, conv
2 T11/12 l, AR T10/11 r, conv T10/11 l, conv T12/L1 r, conv
3 T11/12 r, AR T11/12 r, AR T11/12 r, conv T11/12 r, AR
4 T11/12 l, conv T12/L1 r, conv T11/12 r, AR T11/12 r, AR
5 T10/11 r, conv T10/11 l, conv T11/12 r, conv T10/11 r, conv
6 T12/L1 r, AR T10/11 r, conv T11/12 r, AR T10/11 l, AR
7 T10/11 l, conv T11/12 l, AR T11/12 l, AR T11/12 l, AR
8 T12/L1 r, conv T11/12 r, conv T11/12 l, conv T10/11 l, conv
9 T10/11 r, AR T11/12 l, AR T10/11 l, AR T11/12 l, conv
10 T10/11 l, AR T10/11 l, AR T10/11 r, AR T11/12 l, conv
11 T11/12 l, conv T10/11 r, AR T10/11 l, conv T12/L1 r, AR
12 T11/12 l, AR T11/12 l, conv T11/12 l, conv T11/12 r, conv
13 T10/11 l, conv T10/11 r, AR T12/L1 l, AR T12/L1 l, AR
14 T10/11 r, AR T10/11 l, AR T12/L1 r, AR T12/L1 l, conv
15 T11/12 r, conv T11/12 r, conv T10/11 l, AR T10/11 r, AR
16 T10/11 r, conv T10/11 l, conv T11/12 l, AR T10/11 l, conv
17 T11/12 r, AR T11/12 l, conv T10/11 r, AR T11/12 l, AR
18 T12/L1 l, AR T12/L1 r, AR T12/L1 r, conv T10/11 r, conv
19 T12/L1 l, conv T11/12 r, AR T12/L1 l, conv T10/11 l, AR
20 T10/11 l, AR T12/L1 l, AR T10/11 r, conv T10/11 r, AR
Probands 1 to 4 performed 20 punctures each (10 conventional and 10 AR-guided) in a randomised fashion regarding the sequence, side and level. AR, augmented reality-guided technique; conv., conventional landmark technique; l, left; r, right; T10/11, thoracic level 10/11; T11/12, thoracic level 11/12; T12/L1, thoracic/lumbar level 12/1.

Success of puncture

Successful entry into the epidural space on the first puncture attempt as defined above was achieved in 16 out of 40 attempts in the conventional landmark approach, compared with 33 out of 40 successful attempts using the augmented reality-guided technique (Fig. 4), which translates into a success rate of 40% (conventional approach) and 82.5% (augmented reality approach). The difference in the frequency of distribution between success and no success was significant (P = 0.0002) with regard to the technique used. Expressed as the odds for a successful puncture of the epidural space, the odds ratio for the augmented reality-guided technique compared with the conventional landmark approach was 7.07 (95% CI, 2.61 to 18.05).

F4
Fig. 4:
Success rate of the conventional vs. augmented reality-guided approach

Euclidean distance

We recorded a median Euclidean deviation from the ideal trajectories of 12 [8.925 to 19.33] mm (95% CI, 9.9 to 15.9) for the conventional landmark technique. Using augmented reality guidance, the Euclidean deviation was 6.1 [4.125 to 7.875] mm (95% CI, 4.8 to 7.1), which was significantly smaller compared with the conventional landmark approach (P < 0.0001, Fig. 5a).

F5
Fig. 5:
Spatial error and duration of the conventional vs. augmented reality-guided approach

Lateral distance

The lateral error from the ideal trajectories using the conventional landmark approach measured 9.2 mm [6.0 to 16.9] (95% CI, 7.3 to 14.1). When augmented reality was used for the puncture, the lateral error was 3.7 mm [2.8 to 5.775] (95% CI, 3.0 to 4.8), which was significantly smaller (P < 0.0001, Fig. 5b).

Duration of procedure

The median duration of puncture using the conventional landmark approach was 67.5 s [53.25 to 82.50] (95% CI, 57 to 78). With augmented reality-guided puncture, this was 52.5 s [36.00 to 67.75] (95% CI, 44 to 62), significantly faster (P = 0.0015, Fig. 5c).

Subgroup analysis of outcomes

A proband-specific evaluation of the outcomes regarding success rate, Euclidean and lateral distance and duration of puncture is presented in Table 2. For probands 1 to 3, all outcomes were significantly different in the conventional versus augmented reality-guided technique in line with the overall analysis except for the duration, which was significantly different only in proband 1. In contrast, higher success rate and accuracy for the usage of augmented reality was not evident for proband 4 (stereotactic neurosurgeon). When analysing the three specific spinal levels used during the study (thoracic level 10 to lumbar level 1), success rates on the first attempt were compared. With the conventional technique 44% at T10/T11 and both 38% for T11/T12 and T12/L1. For the augmented reality technique, these were 82% at T10/T11, 75% at T11/T12 and 88% at T12/L1. Level-specific differences for the other reported outcomes (Euclidean and lateral distance, duration) were not detectable.

Table 2 - Subgroup analysis of probands 1 to 4
Proband 1 2 3 4
Success
 conv (%) 40 30 20 70
 AR (%) 100 80 90 60
 OR (95% CI) ∞ (2.59-∞) 9.33 (1.00 to 56.53) 36.00 (3.24 to 429) 0.64 (0.12 to 3.81)
P 0.0108 0.0246 0.0055 >0.9999
Euclidean (mm)
 conv 10.0 16.2 18.7 8.9
 IQR [8.2 to 17.1] [11.4 to 25.5] [14.0 to 31.7] [5.9 to 10.5]
 95% CI (6.2 to 22.3) (10.9 to 38.9) (10.0 to 33.5) (5.2 to 10.9)
 AR 4.5 6.3 6.7 6.7
 IQR [3.2 to 6.1] [4.3 to 11.6] [4.2 to 8.2] [4.7 to 7.6]
 95% CI (3.0 to 6.5) (3.9 to 13.2) (3.4 to 8.4) (4.2 to 7.9)
P 0.0002 0.0015 <0.0001 0.1093
lateral (mm)
 conv 8.5 13.2 17.2 5.9
 IQR [5.6 to 16.0] [8.7 to 20] [10.1 to 30.1] [3.3 to 7.5]
 95% CI (4.8 to 21.0) (6.9 to 30.2) (9.1 to 32.6) (2.5 to 7.9)
 AR 3.0 5.1 3.0 4.9
 IQR [2.1 to 4.3] [3.5 to 10.7] [2.0 to 3.8] [4.0 to 6.5]
 95% CI (2.1 to 5.6) (2.8 to 11.7) (1.7 to 4.5) (3.6 to 6.8)
P 0.0019 0.0095 <0.0001 0.6690
duration (s)
 conv 80.5 56.5 76.5 54.0
 IQR [76.3 to 98.3] [49.3 to 64.8] [67.8 to 104.3] [43.5 to 71.3]
 95% CI (74.0 to 102.0) (47.0 to 70.0) (67.0 to 105.0) (42.0 to 72.0)
 AR 34.5 48.5 61.5 51.5
 IQR [27.0 to 46.3] [32.3 to 67.3] [58.5 to 89.3] [39.8 to 56.8]
 95% CI (27.0 to 50.0) (24.0 to 68.0) (57.0 to 96.0) (30.0 to 59.0)
P <0.0001 0.3051 0.1007 0.4247
Proband-specific analysis of the reported outcomes. Results are reported as percentage of successful attempts with the OR for success with AR guidance (95% CI) and as median (bold font) with [interquartile range] and 95% confidence intervals (95% CI) of the median for deviations and duration. AR, augmented reality-guided technique; conv, conventional landmark technique; IQR, interquartile range. Proband 1: senior anaesthesiologist; proband 2: neuroradiologist; proband 3: third-year resident anaesthesiology; proband 4: stereotactic neurosurgeon.

Discussion

The results of this technical proof-of-concept study demonstrate that, using a phantom of the thoracolumbar spine, the utilisation of a portable augmented reality device with superimposed 3D images can facilitate the identification of the epidural space. Both the rate of success and accuracy were significantly higher with the use of augmented reality compared with the conventional landmark technique with preprocedural CT imaging. Additionally, the procedural time was significantly faster when using augmented reality compared with the conventional technique.

We chose this experimental setting to systematically and reproducibly assess the feasibility of augmented reality guidance in epidural access. Additionally, in this experimental setting, we were able to analyse accuracy and success via CT scans without the hazard of radiation exposure. The custom-made phantom provided realistic conditions in terms of palpable landmarks.

Access to the epidural and subarachnoid space is needed for a wide array of therapeutic and diagnostic procedures. Depending on patient and user-based factors, access using the conventional landmark technique can be challenging and might even make the planned procedure impossible. At the same time, because of the proximity of the spinal cord, traumatic and lengthy attempts pose a rare but severe risk for epidural hematoma, infection or dural puncture.16 The reported deviations from the ideal target point for the conventional approach (Euclidean 12 mm and lateral 9 mm) compared with the deviations with augmented reality guidance (6 mm and 4 mm, respectively) are large enough to make the argument that the risk of inadvertent events during the procedure such as dural puncture could be reduced with the usage of augmented reality. Several techniques have been established to facilitate access to the epidural or subarachnoid space. Myelographies and CT-guided interventions such as percutaneous blood patching for the treatment of CSF leakage is performed successfully but requires repetitive radiation exposure.17,18 At the current state of development, the augmented reality-guided approach requires only one initial CT image for the matching of the augmented reality and the real image. This could even be done with low-dose CT imaging or rotational fluoroscopy,19 further reducing radiation exposure. Future technical developments such as magnetic or laser-based technologies (i.e. Lidar: Light imaging, detection and ranging) will have to overcome the current downside that no real-time synchronising of the real and the augmented reality image is possible. However, for potential future clinical applications, it would be desirable to be able to adjust augmented reality-imaging to the inevitable movements within the spine during the procedure on an actual patient in an upright position.

As a preprocedural tool to identify relevant anatomical structures, the use of ultrasound has proved to be helpful in anaesthesia-related procedures with higher rates of success, fewer punctures and less needle redirections.1–3,5,6 However, simultaneous use of ultrasound with the actual puncture remains of questionable benefit because of challenges of needle guidance and visualisation.8,9 Ameri et al. recently reported the technical implementation of augmented reality-guided ultrasound by integrating live tracking of an epidural needle into a B-Mode ultrasound image to enhance needle tip visibility and guidance into the epidural space using magnetic sensors and a custom-made needle magnetic transducer.15 However, the experimental setup was rather complex and no true overlay of virtual content on the actual phantom was done in this study. The augmented reality headset would permit superimposition of virtual images onto the physical body within the user's field of vision, enhancing the applicability of the augmented reality technology.

The initial success rate in our study with the conventional approach (40%) was low compared with clinical studies evaluating first-attempt success of epidural puncture, which is reported as 40 to 60% for the conventional landmark approach.6,20–22 Relatively low conventional success rates in our study could be explained by the experimental setup on a phantom in an unusual prone-positioning, which differs somewhat from a standard clinical situation. Using augmented reality guidance, the success rate of 83% is within the range reported for ultrasound-guidance in some studies (75 to 85%),20,21 indicating that augmented reality guidance could in fact be of additional value.21,22

The augmented reality-guided technique might provide additional benefits in teaching the challenging technique of epidural puncture. Although the benefit of epidural simulators for teaching and training purposes remains debatable,23,24 direct needle visualisation during the training leads to improved learning performance.25 Augmented reality usage not only enables direct needle visualisation but also superimposed visualisation of the relevant interior anatomical structures, providing additional benefit for teaching scenarios.

A separate analysis of outcomes for each proband individually revealed a heterogeneous picture with proband 4 (stereotactic neurosurgeon) presenting as an outlier, while probands 1 to 3 show similar results in all outcomes. In this proband-specific analysis, proband 4 was the only proband with no significant improvement in outcomes using augmented reality compared with the conventional technique. One explanation could be the high level of clinical experience of proband 4 regarding spatial orientation in three-dimensional anatomical structures for stereotactic procedures (e.g. trigeminal nerve thermocoagulation). Additionally, as this proband was the only one showing lack of increased success rate using augmented reality, technical problems with manual referencing or display of the superimposed images cannot be excluded. This detailed analysis indicates that depending on preprocedural experience, augmented reality guidance might not benefit all users.

Our study contains several limitations. Strictly speaking, we did not compare an augmented reality-guided approach to a classical epidural puncture but rather to a neuraxial access after preprocedural CT imaging. Although in this scenario, augmented reality guidance proved to be superior to preprocedural CT-imaging alone, we cannot conclude that it will be advantageous in a classical epidural anaesthesia scenario. Second, with the current state of technology, augmented reality synchronisation with the physical environment remains a challenge and has to be done manually. Further, synchronisation is static and does not automatically adapt to movement of the physical environment. However, this would be desirable, taking into account potential application in moving, soft tissue structures. Additionally, a reference image for synchronisation and co-registration such as a CT scan is currently necessary to align the physical environment with the superimposed augmented reality image. This would hardly be justifiable in a clinical setting to be done on a routine basis. With regard to the spine, the CT scan would require to be done in the same position as the actual procedure, unless in the future dynamic co-registration will be possible and images already existing could be used. Future technology might even enable dynamic co-registration with images posing less radiation hazards to patients such as ultrasound.

Additionally, the phantom did not feature any in-vivo characteristics of the identification of the epidural space (e.g. loss of resistance). Therefore, another main limitation was that the probands were supplied with the desired depth of insertion in order to reach the epidural space, which might have influenced performance especially with the conventional approach. Under these experimental conditions, augmented reality guidance was superior to the conventional landmark approach. However, the scenario, which did not resemble clinical routine, could have introduced a bias influencing the performance of the probands and may explain low initial success rates with the conventional technique, especially in those probands used to different clinical routines for epidural puncture.

It is important to point out that our model and the technique presented in this article are not yet technically mature to be used in a real-life clinical scenario for neuraxial procedures. The primary aim of this technical proof-of-concept study was not the simulation of such in-vivo scenarios but rather to investigate the applicability of this new technology in the hands of clinicians. The structure of this study, with a preprocedural CT scan and neuraxial puncture in prone position, is cumbersome. Future technical developments with dynamic real-time referencing using new laser technology must aim to enable realistic scenarios in a sitting position with the use of alternative referencing techniques such as ultrasound.

Therefore, it is not yet possible to answer the question whether the reported higher accuracy and higher success rate on the phantom will contribute to higher success rates in vivo or will reduce the risk of undesired needle placement in adjacent structures such as the subarachnoid space, paravertebral space or neuronal structures. This limitation also holds true regarding the reported time advantages for the usage of augmented reality, as the study structure does not reflect current clinical practice for neuraxial procedures. Future studies will have to critically determine whether augmented reality will really mean faster procedure times and whether this translates into a clinically significant advantage.

Conclusions

In summary, we provide experimental proof in a phantom-based setting, that augmented reality can greatly facilitate access to the epidural space in terms of accuracy and success when performed during therapeutic and diagnostic procedures. This first technical proof-of-concept needs to be developed further in order to be applicable to real clinical scenarios.

Acknowledgements relating to this article

Assistance with the article: none.

Financial support and sponsorship: this work was supported by departmental funding.

Conflicts of interest: PCR receives research support from Else-Kroener-Fresenius Foundation (Germany) and Fraunhofer Society (Germany), received personal honoraria for lectures or advice from Boston Scientific (USA) and Brainlab (Germany) and is a consultant for Boston Scientific (USA), Inomed (Germany) and Brainlab (Germany). NS received research support from the German Research Foundation (DFG). TD received funding from the Research Commission, Medical Centre, University of Freiburg. The remaining authors declare that they have no conflict of interest.

Presentations: none.

This manuscript was handled by Matthias Desmet.

References

1. Ansari T, Yousef A, El Gamassy A, et al. Ultrasound-guided spinal anaesthesia in obstetrics: is there an advantage over the landmark technique in patients with easily palpable spines? Int J Obstet Anesth 2014; 23:213–216.
2. Evansa I, Logina I, Vanags I, et al. Ultrasound versus fluoroscopic-guided epidural steroid injections in patients with degenerative spinal diseases: a randomised study. Eur J Anaesthesiol 2015; 32:262–268.
3. Kallidaikurichi Srinivasan K, Iohom G, Loughnane F, et al. Conventional landmark-guided midline versus preprocedure ultrasound-guided paramedian techniques in spinal anesthesia. Anesth Analg 2015; 121:1089–1096.
4. Lim YC, Choo CY, Tan KT. A randomised controlled trial of ultrasound-assisted spinal anaesthesia. Anaesth Intensive Care 2014; 42:191–198.
5. Nassar M, Abdelazim IA. Prepuncture ultrasound guided epidural insertion before vaginal delivery. J Clin Monit Comput 2015; 29:573–577.
6. Tawfik MM, Atallah MM, Elkharboutly WS, et al. Does preprocedural ultrasound increase the first-pass success rate of epidural catheterization before cesarean delivery? A randomized controlled trial. Anesth Analg 2017; 124:851–856.
7. Vallejo MC, Phelps AL, Singh S, et al. Ultrasound decreases the failed labor epidural rate in resident trainees. Int J Obstet Anesth 2010; 19:373–378.
8. Arzola C, Mikhael R, Margarido C, et al. Spinal ultrasound versus palpation for epidural catheter insertion in labour: a randomised controlled trial. Eur J Anaesthesiol 2015; 32:499–505.
9. Ambulkar R, Patil V, Doctor JR, et al. Accuracy of ultrasound imaging versus manual palpation for locating the intervertebral level. J Anaesthesiol Clin Pharmacol 2017; 33:348–352.
10. Eftekhar B. App-assisted external ventricular drain insertion. J Neurosurg 2016; 125:754–758.
11. Elmi-Terander A, Burstrom G, Nachabe R, et al. Pedicle screw placement using augmented reality surgical navigation with intraoperative 3D imaging: a first in-human prospective cohort study. Spine (Phila Pa 1976) 2019; 44:517–525.
12. Maruyama K, Watanabe E, Kin T, et al. Smart glasses for neurosurgical navigation by augmented reality. Oper Neurosurg (Hagerstown) 2018; 15:551–556.
13. Rau A, Roelz R, Urbach H, et al. Application of augmented reality in percutaneous procedures-rhizotomy of the Gasserian ganglion. Oper Neurosurg (Hagerstown) 2021; 21:160–164.
14. Watanabe E, Satoh M, Konno T, et al. The trans-visible navigator: a see-through neuronavigation system using augmented reality. World Neurosurg 2016; 87:399–405.
15. Ameri G, Rankin A, Baxter JSH, et al. Development and evaluation of an augmented reality ultrasound guidance system for spinal anesthesia: preliminary results. Ultrasound Med Biol 2019; 45:2736–2746.
16. Cameron CM, Scott DA, McDonald WM, et al. A review of neuraxial epidural morbidity: experience of more than 8,000 cases at a single teaching hospital. Anesthesiology 2007; 106:997–1002.
17. Mamlouk MD, Shen PY, Sedrak MF, et al. CT-guided fibrin glue occlusion of cerebrospinal fluid-venous fistulas. Radiology 2021; 299:409–418.
18. Mihlon F, Kranz PG, Gafton AR, et al. Computed tomography-guided epidural patching of postoperative cerebrospinal fluid leaks. J Neurosurg Spine 2014; 21:805–810.
19. Reinacher PC, Kruger MT, Coenen VA, et al. Determining the orientation of directional deep brain stimulation electrodes using 3D rotational fluoroscopy. AJNR Am J Neuroradiol 2017; 38:1111–1116.
20. de Filho GR, Gomes HP, da Fonseca MH, et al. Predictors of successful neuraxial block: a prospective study. Eur J Anaesthesiol 2002; 19:447–451.
21. Jagadish A, Swaminathan S, Bidkar PU, et al. Ease of lumbar epidural catheter insertion with prepuncture ultrasound as guidance compared with conventional palpatory technique when performed by anesthesiology residents: a randomized controlled trial. J Anaesthesiol Clin Pharmacol 2021; 37:216–220.
22. Qu B, Chen L, Zhang Y, et al. Landmark-guided versus modified ultrasound-assisted Paramedian techniques in combined spinal-epidural anesthesia for elderly patients with hip fractures: a randomized controlled trial. BMC Anesthesiol 2020; 20:248.
23. Friedman Z, Siddiqui N, Katznelson R, et al. Clinical impact of epidural anesthesia simulation on short- and long-term learning curve: high- versus low-fidelity model training. Reg Anesth Pain Med 2009; 34:229–232.
24. Lim G, Krohner RG, Metro DG, et al. Low-fidelity haptic simulation versus mental imagery training for epidural anesthesia technical achievement in novice anesthesiology residents: a randomized comparative study. Anesth Analg 2016; 122:1516–1523.
25. Zivkovic N, van Samkar G, Hermanns H, et al. Face and construct validity of TU-Delft epidural simulator and the value of real-time visualization. Reg Anesth Pain Med 2019; 44:298–302.
Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the European Society of Anaesthesiology and Intensive Care.