
5 minute read
Use of Artificial Intelligence and Machine Learning in Spine Surgery
The rapid development of artificial intelligence (AI) is reaching all aspects of our world. Artificial intelligence is being used to make our web searches more focused, make the ads that we see more personalized, and even assist us in the workplace. Artificial intelligence has already made inroads into spine surgery. Consider that the basis of science is to generate data, and the practice of medicine is to synthesize the vast amount of data produced in bench research, animal studies, and clinical trials into a well-organized plan to treat patients. One area in which AI can assist in spine surgery is machine learning, a process in which a computer program evaluates datasets to identify relationships in the data and then assist in surgical planning.
Spinal Deformity
Spinal deformity is a field of spine surgery in which a significant amount of research has been done. There are certain parameters that can improve surgical outcomes in spinal deformity surgery. For example, studies have shown that obtaining sagittal balance can lead to improved outcomes.[1-3] Surgical techniques that can help achieve sagittal balance include the use of Smith-Peterson osteotomies and pedicle subtraction osteotomies.[1-3] The selection of levels to fuse can also determine sagittal balance.[1-3] Data are currently being gathered comparing preoperative and postoperative radiographs and correlating them with outcomes. Based on a growing database of scoliosis cases, programs now exist that can help generate a surgical plan that can give surgeons and patients the best possible outcome. Such programs can analyze a vast amount of data that would be too tedious and repetitive for humans. One unfortunate aspect of medicine is that physicians work their entire careers to accrue knowledge, but when physicians retire, that knowledge often is not passed on. The benefit of machine learning programs is that this knowledge can be stored in the programs, and by ingesting that information, the programs will continue to improve and be able to assist surgeons well on into the future, such as helping surgeons synthesize a plan for complex scoliosis cases.
As a case example, R.O. is a 65-year-old man with a history of congenital scoliosis. Over the years, the scoliosis has worsened, and he would like to consider surgery (Figure 1). The planning for surgery in this case would be very challenging. However, the preoperative films were sent to one of the programs, which sent back a surgical plan (Figure 2). Currently, the program is only able to give recommendations in the sagittal plane, but in time, the program’s capabilities will expand and become more refined with each new case that is added.



Disc Degeneration and Stenosis
Another area of spine surgery in which a computer program may be able to assist surgeons is in the selection patients with lumbar disc degeneration and stenosis for surgery. The surgical indications in these cases are commonly recognized to be worsening radicular pain, loss of sensation or motor weakness, and loss of bowel or bladder control. While surgeons try to adhere to these principles, surgical results still remain highly variable.[4-6] Many patients do well after surgery, but some patients are not satisfied with the surgery they had. This can be due to many reasons, such as patients experiencing instability after decompression[4-6] or not having adequate decompression.
Spine surgeons often rely on clinical judgment and experience, in addition to scientific studies, to determine which individuals are good candidates for surgery and which surgical procedure is most appropriate. Nevertheless, the field of medicine is still an art and not entirely a science at this point. As long as there is some intuition involved in surgical decision-making, there will be variability in patient selection methods and surgical technique utilization. For example, a 65-year-old woman diagnosed with lumbar stenosis and a grade 1 spondylolisthesis at L4-5 will most likely have a decompression and fusion at L4-5. In contrast, an 85-year-old woman with lumbar stenosis and a grade 1 spondylolisthesis may have many different treatment recommendations because of her age and possible medical comorbidities. In situations like this, having a program that could predict outcomes and assist spine surgeons in decision-making would be a huge benefit to surgeons and would improve patient satisfaction.
Conclusion
New technology improves the field of medicine. The study of human anatomy made surgery possible. The discovery of penicillin and antibiotics made surgery safer. The discovery of radiographs and advanced imaging made surgery more precise. The use of instrumentation made corrections more powerful. It will be interesting to see what gains spine surgery will see from artificial intelligence.
References
1. de Kleuver M, Faraj SSA, Haanstra TM, et al; COSSCO study group. The Scoliosis Research Society adult spinal deformity standard outcome set. Spine Deform. 2021;9(5):1211-1221.
2. Ilharreborde B. Sagittal balance and idiopathic scoliosis: does final sagittal alignment influence outcomes, degeneration rate or failure rate? Eur Spine J. 2018;27(Suppl 1):48-58.
3. Koller H, Pfanz C, Meier O, et al. Factors influencing radiographic and clinical outcomes in adult scoliosis surgery: a study of 448 European patients. Eur Spine J. 2016;25(2):532-548.
4. Phillips FM, Slosar PJ, Youssef JA, Andersson G, Papatheofanis F. Lumbar spine fusion for chronic low back pain due to degenerative disc disease: a systematic review. Spine (Phila Pa 1976). 2013;38(7):E409-E422.
5. Furunes H, Storheim K, Brox JI, et al. Total disc replacement versus multidisciplinary rehabilitation in patients with chronic low back pain and degenerative discs: 8-year follow-up of a randomized controlled multicenter trial. Spine J. 2017;17(10):1480-1488.
6. Mannion AF, Brox JI, Fairbank JC. Comparison of spinal fusion and nonoperative treatment in patients with chronic low back pain: long-term follow-up of three randomized controlled trials. Spine J. 2013;13(11):1438-48.
AUTHOR
Yu-Po Lee, MD
From UCI Health in Orange County, California