The future of surgical tech may conjure up certain images: robot surgeons assisting humans; hologram anatomies projected above patients. But it is the question of images – the visuals healthcare staff use and need for patient care – which increasing numbers of tech players are trying to solve, mainly using artificial intelligence (AI).
In surgery, imagery is vital. But according to Proprio co-founder and CEO Gabe Jones, surgery is “very static and very flat,” relying heavily on 2D images that are not real-time.
“Surgeons are forced to do a lot of calculations in their heads,” he tells Verdict, “including assessing whether what they’re looking at is accurate and up-to-date.”
Proprio’s flagship product is a surgical navigation system which allows an enhanced and real-time 3D view of the anatomy, made possible in part by the AI solution of computer vision (CV) which gives machines the power of visual recognition in a way that emulates human sight. With this system, as Jones explains, surgeons have the opportunity to see more of a patient’s anatomy, and improve their situational awareness.
“With the integration of AI, augmented reality, and real-time surgical intelligence, surgeons will be able to use Proprio to see around obstacles and align their movements with digitally displayed 3D surgical path plans.” Jones describes the surgical tech as an “intelligent co-pilot for the most difficult moments in a surgery”, aiming to maximise surgeon efficiency and accuracy.
“AI enables individual elements of anatomy to be identified and tracked independently, so a surgeon can have improved awareness of their precise location in the body and potentially avoid unnecessary injury to the patient.”
While extended reality (XR) has entered surgical tech through Hololens-wielding startups such as Medivis, the CV behind Proprio’s augmented reality (AR) solution allows surgeons to see inside the anatomy while operating in real-time.
“AR and virtual reality (VR) are already proving impactful across the care continuum from training surgeons, to patient engagement, to pre-operative planning and simulation. A commonly-held view is that these are the only applications for immersive technologies in medicine.
“Proprio though leverages AR/VR in combination with CV and robotic navigation to create an adaptive imaging and navigation system for use before, during and after surgical procedures, and throughout the most critical and high-risk moments of a case. AR will be the immersive platform of choice to interact with that kind of information, which is best viewed in 3D and interactively,” Jones believes.
In agreement with Jones on the potential of this tech is Dr Owase Jeelani. Dr Jeelani is a Consultant Paediatric Neurosurgeon at Great Ormond Street Hospital (GOSH), and is well known for performing one of the most innovative surgeries in recent times, separating craniopagus conjoined twins Safa and Marwa Bibi in 2019.
“There is little doubt that virtual reality, augmented reality and computer vision in general are going to play a leading role in further developments in surgery and healthcare over the coming decades,” Jeelani tells Verdict.
“We are already using a number of these technologies in surgical theatres, albeit in a rudimentary fashion given the technology’s potential.”
Since 2012 the doctor has also been the co-director of Facevalue, a research programme based at UCL which specialises in designing machine learning algorithms to improve surgical outcomes. But, while he is a surgeon familiar with AI, Jeelani posits another acronym as more useful to the future of surgical tech: natural intelligence (NI). This covers all the systems of control present in biology, complementing its antithesis of AI and vice versa.
“AI is a myth, a most exciting one for sure, but a myth none the less. Firstly, AI did not spring from the ether, it originated within and from NI. It was built to augment, and indeed undertake permutations, that NI on its own motherboards was not capable of.
“AI’s biggest promise is in structuring algorithms and crunching astronomical amounts of data to reveal associations and causation – in other words to give us answers. That done, these answers need to be incorporated back into the real world to deliver real outcomes. This will, for the foreseeable future, need NI.
“Consider the following real life scenario: the surgeon’s gut says something is not right, but the algorithm states with 99.9% accuracy it is – should the surgeon still poke the biopsy needle a further 2cm into the brain?” Jeelani asks.
Jeelani’s view is that AI-as-panacea is myth, exercising more of a need for NI in surgical tech. Strict regulations have also hindered the wide adoption of AI in medical devices, as Globaldata medical analyst Aliyah Farouk explains.
“The EU has recently announced plans to impose additional requirements on the use of AI in medtech,” says Farouk. “Any company that does not comply with the regulations will be faced with huge penalty. Meeting regulatory requirements is already a costly task and this additional requirement may deter smaller companies from entering the market.
“Additionally, the FDA has only approved devices that use ‘locked’ algorithms so far, and machine learning algorithms need updates. However, they have issued an action plan for regulating modifications to AI and machine learning-based Software as a Medical Device (SaMD), so things are looking up in that direction.”
Such regulations may be why other, less life-and-death areas of healthcare are making use of computer vision’s various capabilities. These include image recognition, in which image recognition algorithms have been trained to identify differences in digital images of different classes.
Image recognition is how an AI system by Alibaba is able to diagnose Covid-19 cases in seconds with 96% accuracy using CT scans. With computer vision analysing CT scans and X-rays, the National Institutes of Health can help evaluate the severity of Covid in patients and predict how they will react to different treatments.
This same tech has been used to detect tumours and skin conditions, make invasive medical procedures such as endoscopies safer and enable autonomous tracking of health product inventory.
CV also works with the moving image: video recognition software is able to analyse video clips before comparing them to a database of content in search of a match. A tool by AiCure determines by phone whether patients and guinea pigs are taking prescribed medicines and doing so correctly. Patients film themselves taking the medicine via smartphone; using CV, the AiCure app can determine whether the correct patient has taken the correct medicine through identification of both face and pill.
Surgical platform developer Theator raised a $15.5m series A round this year off the back of its surgical tech video tools. The startup’s surgical intelligence platform scans video footage of real-world procedures and identifies key “intraoperative” moments in order to annotate them with metatags. Through desktop and mobile apps, surgeons have access to an indexed library of over 400,000 minutes of surgical video encompassing over 80,000 moments.
Having such a useful, accessible database can save surgeons both time in research, consultation and surgery. For Jones, this is a fundamental part of the Proprio mission.
“The artful and scientific combination of AI and AR technologies is poised to reinvent medicine. AI is a powerful set of tools that can be used to analyse all of the information around a surgery and the operating room – think Google Maps for the human body. However the most ambitious applications of AI in medicine will free up human cognition from the most mundane tasks and enable humans and computers to focus on what each does best.”
Currently pending FDA clearance for its product, Proprio already boasts partnerships with Intel and NVIDIA and has worked with top research hospitals like the University of Washington and NYU.
“We have conducted numerous studies, simulations and tests over the past five years. Neurosurgeons and orthopaedic surgeons in our pilot studies are experiencing significant improvements in performance (as) Proprio enhances what they’re able to do in surgery.
“Humans will be performing surgery for quite some time – but AI will enable them to do it even better,” Jones concludes.
Find the GlobalData Artificial Intelligence in Healthcare – Thematic Research report here.