Skip to main content
Home » Imaging » AI, big data and deep learning – the future of radiology
Future of Imaging 2020

AI, big data and deep learning – the future of radiology

iStock / Getty Images Plus / gorodenkoff

Dr Sridhar Redla

Honorary Secretary and President Elect, The British Institute of Radiology

We are witnessing the third big transformation in the form of artificial or augmented intelligence. The discussion is no longer about “if it comes”, but “when it comes.”

Radiology has transformed to an extraordinary degree since Roentgen’s discovery of X-rays in 1895, comments Dr Jane Phillips-Hughes, President of the British Institute of Radiology.

The 1970s and 1980s saw the introduction of cross-section imaging, in the form of ultrasound, computerised tomography (CT), magnetic resonance imaging (MRI), etc.

The second transformation of the profession happened in the 1990s, with the introduction of picture archiving and communication system (PACS), which moved us on from the analog to the digital age.

We can now access radiological investigations and results from most work computers. This transformed the working lives of not just radiologists, but the medical workforce as a whole. Now we are seeing the third major revolution in imaging!

AI will complement the work of radiologists

Without a doubt, AI will have a major positive impact on diagnostics, especially in regards to radiology and imaging.

It has now been widely accepted that AI will not replace radiologists, but complement and augment the care provided. AI is expected to help with prioritisation, improving efficiency and accuracy – and thereby patient safety.

The future will hopefully see AI seamlessly integrate into our PACS workstations, acting as a ‘third eye’, providing a more timely and accurate diagnosis.

It is to be recognised, however, that the role of AI, its benefits and usage, will vary from country to country, based on the local healthcare system and its challenges.

Data is the new ‘energy source’ for AI

For AI to be successfully implemented, we must develop algorithms that are practical to the profession. For that we need clean data.

The National Health Service is a huge source of data; given its organised structure and the fact that it is a freely accessible healthcare providers.

Naturally, the big and small vendors would like this data to develop their machine learning programs and algorithms. This puts the radiology professions in a prime position, to direct the vendors and big players to work to our advantage.

Applications of AI in radiology

Any new algorithm should be properly validated and should be ‘vendor neutral’, so that it can be seamlessly integrated into any imaging system. One side of AI should help with the day-to-day, mundane tasks like workflow, scheduling, protocols, reporting, quality assessment, dose modulation etc. However, the machine learning aspect should be concentrated on the clinical applications.

The NHS has among the lowest numbers of doctors, nurses and hospital beds per capita in the western world and radiology has an acute workforce crisis. The proposed Government funded AI lab should help ease the pressure on these healthcare professionals, provided the priorities are right.

Next article