President, The British Institute of Radiology
As the pace of technological advancements continues to increase, so does the opportunity to improve the quality of medical imaging to help transform healthcare.
The use of AI to manage doctors workloads
In all areas of our lives the pace of technological possibilities and change is forever increasing. Healthcare is no different. Where we see advanced artificial intelligence (AI) potentially changing the way we drive, we also see AI driving advances in personalised medicine.
Imaging procedures are increasing in number and the number of staff in post to deal with the extra demand is not keeping pace. In other words, medical imaging is certainly not immune from all the advances and societal pressures that touch our lives.
Ground-breaking advances using imaging
One recent advance in the use of imaging is a new technology called the MRI-Linac. A Linac is the machine used in radiotherapy to fire high-energy radiation beams at cancerous cells to kill the cancer or provide palliation to the patient from their symptoms. Improvements in imaging the location of the tumour have enabled higher doses of radiation to be delivered with less radiation hitting nearby healthy tissues, thus reducing unnecessary side-effects. Until recently this imaging has been done with X-rays but the integration of an MRI scanner with a Linac has vastly improved accuracy. This is truly a ground-breaking advance, whose full potential we are only just beginning to realise.
Using Artificial Intellligence with imaging data
AI and machine learning are now gaining traction in imaging. To predict patient outcomes, we require large datasets of patient information, all interconnected to gain the maximum benefit. However, clinical data sets are notorious for having incorrect manual entries and missing data. The better the data is, the more robust the final conclusion, so data integrity issues will be massively important in the future.
Clinical data sets are notorious for having incorrect manual entries or missing data
Another issue not often appreciated is that imaging data is more useful to AI techniques when the scans are performed consistently. There are global initiatives to harmonise the way complex scans are performed to ‘prepare’ the data we use in the future. The ultimate way to control data acquisition techniques is to be found in the BioBank initiative where 100,000 individuals are being imaged at a single facility in exactly the same way and lifestyle and health information tracked over many years to provide a vast data set of useful clinical data for future researchers. All of which raise valid ethical and privacy issues that have yet to be solved.
The use of X-rays brings with it the need to minimise any potential harm that may be caused by those X-rays (in fact, a very, very small increased risk of cancer). There are many ways to minimise such harm that range from ensuring we image the correct patient, to identifying variations in practice that are not justified. This area of radiation control is also undergoing technological advances as new software tools are emerging to collect radiation dose information for every patient from imaging machines for data analysis in far greater numbers than we have previously been able to!
This increase in information will definitely lead to better insights into how to improve the quality of medical imaging.
We really are on the cusp of seeing healthcare IT and associated technologies move forward at pace, hopefully for the benefit of us all.