Published online Dec 20, 2024. doi: 10.5662/wjm.v14.i4.95762
Revised: June 3, 2024
Accepted: June 13, 2024
Published online: December 20, 2024
Processing time: 99 Days and 15.1 Hours
Artificial intelligence (AI) technology is vital for practitioners to incorporate AI and robotics in day-to-day regional anesthesia practice. Recent literature is encouraging on its applications in regional anesthesia, but the data are limited. AI can help us identify and guide the needle tip precisely to the location. This may help us reduce the time, improve precision, and reduce the associated side effects of improper distribution of drugs. In this article, we discuss the potential roles of AI and robotics in regional anesthesia.
Core Tip: Artificial intelligence (AI) technology has been incorporated in the medical field, including anesthesia. It is vital for practitioners to incorporate AI and robotics in day-to-day regional anesthesia practice. The literature is encouraging on its applications in regional anesthesia, but the data are limited. AI can help us identify and precisely guide the needle tip to the location. This may help us reduce the time, improve precision, and reduce associated side effects of improper distribution of drugs. We discuss the potential role of AI and robotics in regional anesthesia.
- Citation: Choudhary N, Gupta A, Gupta N. Artificial intelligence and robotics in regional anesthesia. World J Methodol 2024; 14(4): 95762
- URL: https://www.wjgnet.com/2222-0682/full/v14/i4/95762.htm
- DOI: https://dx.doi.org/10.5662/wjm.v14.i4.95762
Artificial intelligence (AI) is the amalgamation of various algorithms that allows machines to generate the aptitude to analyze and perform complex functions such as problem-solving, object recognition, and decision-making[1]. With the growing advancements in the field of science and its application in medicine, AI-based technologies in medical science are on the rise. Therefore, all clinicians need to understand the application of these new technologies in medical science and use them appropriately to deliver more safe, efficient, and cost-effective patient care[2].
Anesthesiologists have been at the forefront in the initiation of patient safety initiatives. With the advent of advanced technology in the form of AI, these safety initiatives can be further refined to deliver improved quality of patient care[3]. The specialty of anesthesia can immensely benefit from recent developments in AI, as it has found applications in multiple elements of clinical practice, including perioperative and intensive care, pain management, airway management, and regional anesthesia (RA)[3,4]. Research has suggested that AI tools can have potential applications in various clinical scenarios in anesthesiology practice, including ultrasound guidance (USG) for RA and pain management[4]. Machine learning is a subset of AI characterized by improvements in performance through iterative tuning of weights or coefficients within mathematical models to identify patterns in complex data sets[5]. It can potentially provide scientific data on the role of RA in improvements of long and short-term patient-centric outcomes such as the length of in-hospital stay and patient mortality, which are otherwise challenging to furnish due to lack or incorrect data[3,4].
Despite mounting evidence in favor of AI in anesthesia, its role in RA practice is still preliminary. The utilization of AI in ultrasound-guided procedures is an essential integration of newer technology, which present-day clinicians are exploring[4]. The success of regional anesthesia relies on the sound knowledge of sonoanatomy and the expertise to carefully delineate the structure under ultrasound imaging. However, this may not appear as simple as it sounds because of varied sonoanatomy. Because ultrasound-guided nerve blocks rely on imaging, AI could potentially be used to improve image optimization and its interpretation in real-time during the procedure, which would help physicians identify the target nerve and avoid nerve block-related complications[1,4].
AI-assisted ultrasound-guided RA can facilitate the identification of anatomical structures, optimize the sonographic image, improve needle visibility, and help non-experts locate the correct USG anatomy to perform regional blocks[4]. The use of AI in USG-RA has been found to help increase the success rate of nerve blocks, improve safety, and decrease the complication rate. With the integration of AI in ultrasound, the safety profile of regional blocks can be further enhanced by detecting and labeling anatomical structures (e.g., blood vessels) to reduce or avoid unwanted injury. AI can also be used as an educational tool to train novice anesthesiologists and trainees by helping identify sonoanatomy for USG-RA[4,5].
AI can significantly help decrease the risks and complications associated with injuring vital structures while performing a regional technique. AI can detect structures for various ultrasound-guided procedures, like nerves for nerve blocks and veins for central venous cannulation[1,4]. Convolutional neural networks are the most commonly employed method of achieving ultrasound image cataloging[6].
Bowness et al[7] examined NerveBlox® AI software for the identification of structures while performing regional techniques. AI models accurately identified the target structure in 93.5% of cases (1519/1624) with false-negative and false-positive rates of 3.0% (48/1624) and 3.5% (57/1624), respectively. Also, the risk of block failure decreased by 81.3% (585/720). In another study, authors built their neural network to identify the sciatic nerve as it was scanned over the posterior thigh[8]. The primary aim was to train the system to detect important information and ignore irrelevant information. The performance of the convolutional network was compared with that of a traditional 2D U-Net network, and the in-house study approach performed better than the standard 2D approach. The studies mentioned above demonstrate that nerve detection for RA is realistic using AI tools. Still, for widespread clinical application, further robust literature is required to develop a more efficient tracking system[8]. Scholzen and Schroeder[9] also evaluated the usefulness of the NerveBlox AI tool for training and teaching purposes. Eleven anesthesiologists and 25 students participated and worked on standardized patients. Both faculty and resident anesthesiologists rated NerveBlox® AI software's utility as a teaching aid with ratings of 9 (IQR: 7.5-10, n = 11) and 10 (IQR: 9-10, n = 25), respectively.
In addition to recognizing specific structures in ultrasound images, researchers have used these neural networks to assist in identifying correct vertebral levels and other anatomical landmarks to aid epidural catheter placement[10].
AI can help rapidly examine big data that is otherwise difficult to interpret, including patient, operator, procedure, and drug-related details. It may be used to speculate strategic steps required to correctly perform a given nerve block[2]. There is also potential for AI to act as a supervisor or assistant to a novice anesthesiologist and rapidly identify scenarios to improve block success. AI has been used in RA to detect the correct insertion site, track the needle insertion, and facilitate needle tip localization and length localization. Needle tracking is one of the most widely used functions in computer vision. Several AI models have been documented to improve monographic anatomical target perception quality. A multiple-model data association is applicable in detecting nerves and vessels[2,11].
AI technology could improve the interpretation of USG anatomy by identifying nerve block-relevant targets (such as peripheral nerves and fascial planes) and help map optimal insertion sites by detecting the pertinent landmarks and guidance structures[1,2]. Moreover, AI allows for the standardization of clinical procedures by providing ultrasound views for anesthetists, and a real-time representation of anatomical structures for immediate decision-making during blocks can potentially allow automated nerve block techniques to be performed using a remote control system[1,11].
There are three categories of robotics in anesthesia: Pharmaceutical, cognitive, and mechanical. Pharmaceutical robots are exemplified by target-controlled anesthesia using electroencephalogram (EEG) as a feedback loop. Mechanical robots have the advantage of better precision and dexterity than humans, and cognitive robots perform as decision support systems[12].
In RA, ongoing work on computer-assisted or even robotically autonomous ultrasound-guided procedures could become real possibilities with enough literature in the future. The first robotic ultrasound-guided nerve blocks in humans were described as early as 2013 using the Magellan System with a success rate of 100%[13]. In another study on training anesthetists using a robotic arm driven by a joystick to assess learning curves on a nerve phantom, learning curves were improved compared with manual insertion. The steeper learning curve in the said study was likely due to the novelty of the technology[14].
Furthermore, there is a potential danger of too much reliance on robotic assistance during training, and the overall competence of trainees may be inadequate, exposing them to any emergencies and equipment failure. Therefore, it is crucial to carefully design robotic techniques in training as a feedback system to aid and not supplant the training process[13,14].
Future cognitive robotic systems will be able to inform the anesthetist of a problem and may also be capable of suggesting or administering the treatment. Recent examples include devices such as Safer Injection for Regional Anesthesia, which eliminates the need for an assistant during nerve block. It can aspirate during injection and cut off flow when injection pressure is more than 117 kPa[15].
AI and robotics have substantial potential applications in the field of anesthesiology. Advancements in augmented and virtual reality mixed reality technologies, including advanced sensing systems, display systems, and simulation platforms, will likely be informed by further advancements in AI[12]. Augmented and virtual environments will be more realistic in the future with the addition of sensory modalities such as motion, sight, and touch, as they will provide operator feedback and can even be incorporated into autonomous mode mechanical robots to perform tasks[16].
Motion analysis was used to guide the clinical performance of experts and novices regarding supraclavicular brachial plexus block and showed differences between time, number of movements, and needle path length. The feedback helped improve the performance of trainees[16].
Eye-tracking has recently been used in ultrasound-guided RA to objectively assess trainees' difficulties, performance levels, learning curves, and decision-making[17]. Furthermore, reflective feedback based on real-time performance can potentially accelerate the ultrasound-guided RA learning process. A novel system (RA simulator and assistant) system provides reflective feedback using combined virtual feedback using MRI or computed tomography images of actual patients with haptic feedback[12]. In addition to anatomical navigation, augmented reality may be helpful in RA training. Poor accessibility and high cost of high-fidelity cadaveric training are its main limitations, and alternatives like virtual training platforms to provide cadaver-like simulation training will be a big boon in the future. Another novel application of virtual reality to ultrasound-guided RA has focused on patient-centered anxiety reduction and training using virtual reality distraction but has had mixed results[16,17].
AI comes with its limitations. AI is a newly developed tool that we must install in the right circumstances to solve a clinical problem. AI-assisted USG-RA is a novel medical technology that has recently developed and is still growing with each passing day. As this technology is relatively new, most clinicians are learning to use it in their everyday practice. Therefore, beginners' use of these technologies in daily practice may be more time-consuming with a variable learning curve. Also, we must acknowledge that using AI–based techniques may not necessarily result in superior outcomes compared to conventional methods and skills[1,2].
There can be a risk of image misinterpretation in cases of abnormal anatomy (e.g., spinal fusion or reduced interspinous distance) as AI image interpretation is operator-dependent. This may paradoxically increase the incidence of block failure and undesirable trauma to critical structures if the color on the screen misleadingly affirms the practitioner. So, AI-based tools are no replacement for clinical skills and understanding[10].
AI algorithms are susceptible to data bias. Beyond the fundamental research biases, there can be both implicit and explicit biases in the healthcare system that can impact the large-scale data fed to train AIs and meaningfully affect the types of predictions that an AI tool will make to influence clinical decisions. Therefore, clinicians must work with data scientists and engineers to ensure the appropriate interpretation of scientific data[2,18].
The practice of anesthesiology is a fusion of science and art, and much of the data that clinicians gather comes from the clinician-patient relationship that builds on patients' trust in their doctor. It won't be easy to account for these aspects, even by complex AI algorithms. Furthermore, the extent to which patients will be keen to trust these algorithms remains to be seen. Therefore, further research should focus on better understanding the ethical, societal, and cultural implications of integrating AI into clinical systems. Furthermore, reducing inter-operator variability has been a critical driver of robotics technology. However, this may be achieved with simulation training, along with the appropriate objective performance metrics[1].
Cost remains a significant barrier to robotic use in RA on a wide scale. Still, in the long term, they can be cost-effective if fewer complications occur with robotic assistance. Regulatory processes can be another barrier to AI technology[12].
Lastly, AI models will not understand the data's implications for specific patients; therefore, anesthesiologists should partner with other specialties and patients to help develop the strategy for the optimal use of AI[19].
The field of anesthesiology has enormous potential for applying AI-based tools, and anesthesiologists are actively pursuing research incorporating AI technology in various procedural and patient management programs. At present, there is minimal literature in this regard, so it will be vital for clinicians to incorporate these techniques into routine practice to assist in the practical translation of AI. However, future projections point towards using robotics in autopilot mode instead of real-time physicians, but clinical decision-making will likely remain in the human domain. There is limited data on the application of AI in RA, but the available literature is encouraging and should be explored further. Its application in ultrasound-guided nerve blocks improves the identification of structures and needle tips during in-plane needling. Mechanical robots assisting or automatically performing nerve blocks is also a realistic possibility with further technological advances and the availability of sound data.
1. | Hashimoto DA, Witkowski E, Gao L, Meireles O, Rosman G. Artificial Intelligence in Anesthesiology: Current Techniques, Clinical Applications, and Limitations. Anesthesiology. 2020;132:379-394. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 101] [Cited by in F6Publishing: 178] [Article Influence: 44.5] [Reference Citation Analysis (0)] |
2. | Ramesh AN, Kambhampati C, Monson JR, Drew PJ. Artificial intelligence in medicine. Ann R Coll Surg Engl. 2004;86:334-338. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 310] [Cited by in F6Publishing: 322] [Article Influence: 16.1] [Reference Citation Analysis (0)] |
3. | Bellini V, Valente M, Gaddi AV, Pelosi P, Bignami E. Artificial intelligence and telemedicine in anesthesia: potential and problems. Minerva Anestesiol. 2022;88:729-734. [PubMed] [DOI] [Cited in This Article: ] [Cited by in F6Publishing: 1] [Reference Citation Analysis (0)] |
4. | Balavenkatasubramanian J, Kumar S, Sanjayan RD. Artificial intelligence in regional anaesthesia. Indian J Anaesth. 2024;68:100-104. [PubMed] [DOI] [Cited in This Article: ] [Reference Citation Analysis (0)] |
5. | Bowness J, El-Boghdadly K, Burckett-St Laurent D. Artificial intelligence for image interpretation in ultrasound-guided regional anaesthesia. Anaesthesia. 2021;76:602-607. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 17] [Cited by in F6Publishing: 34] [Article Influence: 8.5] [Reference Citation Analysis (0)] |
6. | Fukushima K. Neocognitron: a self organizing neural network model for a mechanism of pattern recognition unaffected by shift in position. Biol Cybern. 1980;36:193-202. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 2930] [Cited by in F6Publishing: 1192] [Article Influence: 27.1] [Reference Citation Analysis (0)] |
7. | Bowness JS, Burckett-St Laurent D, Hernandez N, Keane PA, Lobo C, Margetts S, Moka E, Pawa A, Rosenblatt M, Sleep N, Taylor A, Woodworth G, Vasalauskaite A, Noble JA, Higham H. Assistive artificial intelligence for ultrasound image interpretation in regional anaesthesia: an external validation study. Br J Anaesth. 2023;130:217-225. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 11] [Cited by in F6Publishing: 21] [Article Influence: 21.0] [Reference Citation Analysis (0)] |
8. | Cai N, Wang G, Xu L, Zhou Y, Chong H, Zhao Y, Wang J, Yan W, Zhang B, Liu N. Examining the impact perceptual learning artificial-intelligence-based on the incidence of paresthesia when performing the ultrasound-guided popliteal sciatic block: simulation-based randomized study. BMC Anesthesiol. 2022;22:392. [PubMed] [DOI] [Cited in This Article: ] [Reference Citation Analysis (0)] |
9. | Scholzen EA, Schroeder KM. Use of Artificial Intelligence Software Helpful for Regional Anesthesia Education in Self-Reported Questionnaire in Academic Medical Center Setting. 2023 Preprint. Available from: Research Square:rs.3.rs-2790929. [DOI] [Cited in This Article: ] |
10. | Pesteie M, Lessoway V, Abolmaesumi P, Rohling RN. Automatic Localization of the Needle Target for Ultrasound-Guided Epidural Injections. IEEE Trans Med Imaging. 2018;37:81-92. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 33] [Cited by in F6Publishing: 23] [Article Influence: 3.8] [Reference Citation Analysis (0)] |
11. | Karmakar A, Khan MJ, Abdul-Rahman ME, Shahid U. The Advances and Utility of Artificial Intelligence and Robotics in Regional Anesthesia: An Overview of Recent Developments. Cureus. 2023;15:e44306. [PubMed] [DOI] [Cited in This Article: ] [Cited by in F6Publishing: 1] [Reference Citation Analysis (0)] |
12. | McKendrick M, Yang S, McLeod GA. The use of artificial intelligence and robotics in regional anaesthesia. Anaesthesia. 2021;76 Suppl 1:171-181. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 10] [Cited by in F6Publishing: 21] [Article Influence: 7.0] [Reference Citation Analysis (0)] |
13. | Hemmerling TM, Taddei R, Wehbe M, Cyr S, Zaouter C, Morse J. Technical communication: First robotic ultrasound-guided nerve blocks in humans using the Magellan system. Anesth Analg. 2013;116:491-494. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 21] [Cited by in F6Publishing: 22] [Article Influence: 2.0] [Reference Citation Analysis (0)] |
14. | Morse J, Terrasini N, Wehbe M, Philippona C, Zaouter C, Cyr S, Hemmerling TM. Comparison of success rates, learning curves, and inter-subject performance variability of robot-assisted and manual ultrasound-guided nerve block needle guidance in simulation. Br J Anaesth. 2014;112:1092-1097. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 23] [Cited by in F6Publishing: 24] [Article Influence: 2.4] [Reference Citation Analysis (0)] |
15. | Bodhey A, Nair A, Seelam S. SAFIRA pump: A novel device for fixed injection pressure and to control local anesthetic injection during peripheral nerve block. J Anaesthesiol Clin Pharmacol. 2023;39:146-147. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 2] [Reference Citation Analysis (0)] |
16. | Marhofer P, Eichenberger U. Augmented reality in ultrasound-guided regional anaesthesia: useful tool or expensive toy? Br J Anaesth. 2023;131:442-445. [PubMed] [DOI] [Cited in This Article: ] [Reference Citation Analysis (0)] |
17. | Harrison TK, Kim TE, Kou A, Shum C, Mariano ER, Howard SK; ADAPT (Anesthesiology-Directed Advanced Procedural Training) Research Group. Feasibility of eye-tracking technology to quantify expertise in ultrasound-guided regional anesthesia. J Anesth. 2016;30:530-533. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 21] [Cited by in F6Publishing: 19] [Article Influence: 2.4] [Reference Citation Analysis (0)] |
18. | Chen Y, Clayton EW, Novak LL, Anders S, Malin B. Human-Centered Design to Address Biases in Artificial Intelligence. J Med Internet Res. 2023;25:e43251. [PubMed] [DOI] [Cited in This Article: ] [Cited by in F6Publishing: 18] [Reference Citation Analysis (0)] |
19. | Singh M, Nath G. Artificial intelligence and anesthesia: A narrative review. Saudi J Anaesth. 2022;16:86-93. [PubMed] [DOI] [Cited in This Article: ] [Cited by in Crossref: 3] [Cited by in F6Publishing: 15] [Article Influence: 7.5] [Reference Citation Analysis (0)] |