By Ron Soferman, CEO and founder, RSIP Vision.
Artificial intelligence is transforming the healthcare industry – it is creating opportunities that have been never thought possible while opening up the realm of new possibilities beyond human capabilities.
Powered by increasing availability of healthcare data and advances in machine learning, artificial intelligence aims to mimic human cognitive functions, assisting physicians to make better clinical decisions or even replace human judgement in certain functional areas of healthcare. A major part of AI involves the use sophisticated algorithms to ‘learn’ features from a large volume of healthcare data, and then use the obtained insights to assist clinical practice. It can also be equipped with learning and self-correcting abilities to improve its accuracy based on feedback. It can assist physicians by providing up-to-date medical information from journals, textbooks and clinical practices which can help to reduce diagnostic and therapeutic errors that are inevitable in the human clinical practice.
Here are some recent advances of artificial intelligence in practical use.
“Mapping” the heart and, in many cases, mapping the signal of the heart allows physicians to understand specific problems before deciding on a solution. Taking arrhythmia as an example, by using AI, you can use the mapping to get much clearer understanding of what is the exact problem that causes the irregular heartbeat. Another example involves planning interventions with a catheter. Mapping provides the exact anatomical structure of the arteries so you make decisions on the exact kind of catheter to be used and the exact behavior of the arteries at the specific point where you have to do the intervention. Mapping usually occurs prior to an operation but sometimes it can be used during the operation itself, when you have images from the fluoroscopy; then you can do analysis of the images and get precise information about the location and the structure of the arteries.
This is a very interesting and challenging application where you have to be very precise, especially when you are putting in screws into the vertebrae. Precision cannot be gained very easily just by what the surgeon sees because, in many cases, it’s a percutaneous procedure. It’s difficult to see exactly where the vertebra line up. Artificial intelligence assists in the navigation by utilizing pre-op scanning, along with information provided by the x-ray in the operating room. Algorithms can combine those two sources of information which allows the surgeon to accurately navigate to the exact point of insertion.
This also holds true for hip or knee replacement. Using AI algorithms, during the planning phase, you can decide on a specific implant that will be for a particular patient. Mapping also provides very good segmentation of the bones prior to the operation. This helps avoid doing generic work with an implant that might not fully fit the knee or hip and the patient will suffer from future problems.
In the clinical research phases of pharmaceutical development, a lot of effort is invested in assessing the influence of new drugs by collecting CTs that are done for the patient during the use of the drug. AI algorithms can take a lot of information from all the CT scans, producing an automatic scoring (which is called RECIST) that analyze and measures the influence of the drug, an example being whether a lesion has disappeared or shrunk or whether it stayed the same. AI produces the results of these scans within a few hours, as opposed to several months, which used to be the case providing better and more immediate decision making.
Artificial intelligence helps provide the exact segmentation of the pulmonary airways, which is prerequisite to planning an intervention surgery, such as a biopsy or, in some cases, an ablation. Again, utilizing multiple CTs and using deep learning, airway segmentation can be performed, even the tiny airways found in the rest of the lungs. This creates a mapping of the lungs which allows a surgeon to go and plan actions with the catheter, and in some cases with a robotic catheter through the lungs to get to the exact place. Accuracy is very important, because you don’t want to miss and take the biopsy from adjacent places instead of the lesion itself. And AI also supplies more information about the process like how to eliminate puncturing blood vessels, the fissure between the lobes, etc.
These examples point to artificial intelligence providing extremely fast and almost unlimited image analysis capabilities, removing a seemingly endless bottleneck of tedious doctor tasks. It will continue to provide accurate assistance in surgery (both pre-op and in-op). Artificial intelligence will also foster more patient-specific healthcare through its speedy compilation of patient anatomical data augmented by observations and healthcare correlations that generally aren’t easy to be found. The beauty of AI is in its consistent response, with no inter and intra variations and with relatively short learning curve. A physician needs to see thousands of cases, over several years, to become an expert, while AI can start by seeing millions and become an expert immediately. It will only get better in helping to resolve many different healthcare issues for patients.
The big challenge with AI still revolves around the shortage of trained engineers able to develop new algorithms to solve ever more difficult situations. The good news is healthcare project managers do have the option of utilizing skilled engineers on an outsourcing basis to accomplish algorithm cultivation. Artificial intelligence has been around for 30 years but there is so much more work to be done. However, in the last five years, there has been a big change in the healthcare market where the understanding of the value of AI and computer vision has been made crystal clear and will result in the integration of newer and better AI technology into the healthcare system.