Through close collaborations with remarkable organizations, Intel is providing expertise and world-class hardware to get transformative AI healthcare solutions off the ground. Outlined below are two such partnerships going on in Wales.
LineSafe – Ensuring the Safe Placement of Naso-Gastric Tubes
The LineSafe National Imaging Academy of Wales is a new radiology academy with a goal to increase research and innovation in medical science. In response to an open call for AI projects from the Welsh government (an initiative supported by Intel), radiology trainee Dr. Asad Saeed and Dr. Syed Junaid applied for funding to harness the power of AI to tackle a common and significant problem. Their aim is to keep patients safe by ensuring correct placement of naso-gastric (NG) tubes, which carry food and medicine to the stomach, through their nasal cavity.
Currently, placement requires a certain level of expertise for doctors and other healthcare professionals to assess if they’ve fed the tube correctly into the esophagus, or incorrectly into the windpipe of a patient. The consequences of feeding into the windpipe can be damaging at best, and life threatening at worst.
Doctors often use X-ray imaging to determine the correct placement of the NG tube, but as the esophagus and windpipe are directly in front of each other, it can be hard to tell the two apart. Junior doctors with little specialist training are often required to make the decision, rather than experienced radiologists, which can be stressful and carries risk. LineSafe aims to solve this problem by using AI to train from X-ray image data and accurately determine whether NG tubes have been placed correctly.
The team of radiologists can access thousands of chest X-rays within the National Imaging Academy Wales, which they’ve labelled and fed into an AI model trained on Intel® Xeon® Scalable processors that’s now able to determine the correct placement of NG tubes. Dr. Asad Saeed of the National Imagine Academy confirmed they have exceeded 90% percent accuracy and the results are continuing to improve.
This may be improved even further before the solution is validated and ultimately rolled out to patients within the NHS.
Intel has been an essential collaborator throughout the project. In early 2019, the Intel Health and Life Sciences (HLS) team in the UK ran a series of workshops throughout the project lifecycle. This gave LineSafe the opportunity to work with Intel experts to ensure their project was a success. Intel collaborated with the team to determine an optimal hardware configuration from its rich AI portfolio, running the latest AI optimized software stacks for LineSafe’s training and inference workloads.
ASPIRE – Treating Esophageal Cancer More Effectively and Efficiently
ASPIRE is a project that has brought together scientists from Velindre University NHS Trust (Prof. Tom Crosby and Dr. Kieran Foley) and the Cardiff University School of Engineering (Dr. Emiliano Spezi) to develop deep learning systems that automate radiotherapy planning of esophageal cancer, and once successful, develop the project to use AI to plan radiotherapy in other tumor sites.
The normal process for treatment planning in esophageal cancer is time consuming, involving a diagnostic computed tomography (CT) scan that allows a specially trained oncologist to identify where the growth is. It can take days. For patients and their families who are facing a poor prognosis, this wastes valuable time.
By training an AI model on more than 1,000 3D scans with labelled structures, ASPIRE aims to quickly and accurately determine cancerous tumors based on scans, reducing the pressure on oncologists. Although it’s in early stages, ASPIRE aims to vastly improve the speed, accuracy, and consistency of diagnoses.
Once complete, the project will improve experiences:
- For patients – by cutting time to treatment and providing higher accuracy, they have the best possible chance of a positive outcome.
- For clinicians – by improving workflow efficiency within the NHS, they’re able to treat more patients.
By working very closely with the ASPIRE team, supporting them to determine the correct hardware and software configuration, and providing a number of hands-on workshops to train and optimize performance and accuracy, Intel has played a critical and invaluable role. The ASPIRE team are planning to continue their close collaboration with Intel by working on even more ambitious projects in the future. With the wealth of cancer data available within the National Health Service (NHS), there’s tremendous scope for similar projects to harness this untapped potential.