Mobile Assisted Reconstruction In Orthopaedics: extended to the fully automated surgical navigation system

By 2040, 11.4 % of all adults will experience arthritis-attributed activity limitations globally. These limitations lead to pain, social isolation and loneliness. The incidence of disability from knee pain secondary to osteoarthritis is high at 20% of adults over 50 and 40% over 80 years. Total knee arthroplasty (TKA) is one of the most cost-effective and consistently successful surgeries performed to treat osteoarthritis of the knee. Patient-reported outcomes are shown to improve dramatically with respect to pain relief, restoration of function, and improved quality of life. In 2019, around 100,000 and 1,000,000 knee replacements were performed in the UK and the US, respectively. This number is increasing due to demographic and lifestyle changes. However, 10-20% of TKA patients are not fully satisfied, and this is partly attributed to variations in surgery.

The advanced technologies that improve the surgery are robotic technologies and patient-specific instrumentation. These are expensive (robots) and difficult to implement (patient-specific instrumentation). By taking advantage of the evolution of smartphone technology, we are developing Mobile Assisted Reconstruction In Orthopaedics (M.A.R.I.O), a simple, low-cost, intra-operative navigation system to help surgeons place implants more precisely.

Pre-operative planning is an essential prerequisite for the success of the orthopaedic procedure. We have a developed Statistical Shape Model (SSM) technology that can predict the intact, nonpathological bone shape from X-rays or CT scan images, and we use this to generate a better surgical plan automatically. This plan is utilised by a smartphone-based navigation application during surgery.

The segmentation of medical images is a key step before SSM can take place and the current gold standard is to manually draw the region-of-interest. Users' and implant manufacturers' feedback is encouraging us to develop an automated image segmentation tool now, as part of our M.A.R.I.O device. Machine-learning systems are slowly replacing the manual tasks being carried out by medical experts and radiologists with higher accuracy, reduced human effort, and greater speed. The current state-of-the-art research on machine learning-based medical image segmentation at ICL, ICL's world-leading medical device translation infrastructure and the advent of high-speed computers and their modest price is placing us in an ideal position to automate the segmentation of bone and cartilage shapes, which has remained unachieved.

The overall cost of the current project is equivalent to 50 revision surgeries, we will prove the feasibility of a fully automated pre-surgical planning software to benefit patients, surgeons, and the NHS in the longer term.