How Will AI Change Radiology?


Medical imaging is crucial to the management of different conditions and a major element in ensuring the continuum of care, from prevention, prompt detection and diagnosis, to informed treatment decisions and improved outcomes and rehabilitation.

If the first few decades of radiology were about improving the resolution of the pictures taken of the body, then the next decades will be devoted to interpreting that data to make sure nothing is missed out.

From diagnosis to treatment

Medical imaging is starting to advance from its initial focus—that of diagnosing health problems—to playing an essential part in treatment as well, especially in oncology. 

Physicians are starting to rely on imaging to assist them in observing and keeping track of the evolution of tumors and the spread of cancer cells so that they have a faster way of knowing if treatment is working. 

Imaging will contribute to the type of treatment that patients receive, but also contribute to the way doctors are working, offering timely feedback that will allow them to be more efficient.


Functional imaging becoming part of the care

We will also be seeing a transition to functional imaging, for higher precision purposes: functional imaging (or physiological imaging) is a medical imaging technique of detecting or measuring changes in metabolism, blood flow, regional chemical composition, and absorption.

As opposed to structural imaging, functional imaging focuses on revealing physiological activities within a certain tissue or organ by using medical imaging modalities that employ probes to reflect the spatial distribution of them within the body.


The ever-growing demand for radiology

Imaging has become key in the care of communicable and non-communicable diseases. Yet, there are major shortages of imaging equipment and workforce, especially in low- and middle-income countries (1.9 radiologists per million people in low-income and 97.9 in high-income countries).

On the other hand, the ever-growing demand for radiology combined with increasingly more accurate AI systems makes artificial intelligence a useful tool in improving medical imaging screenings and risk assessment.


Radiology AI

Computer algorithms have been trained to detect when images fall outside of normal parameters, and the more images they are fed, the better they become at flagging abnormalities.

The goal is to automate reading CTs, MRIs, or ultrasounds as much as possible, and save radiologists time, as they are under increasing pressure to read and interpret hundreds of images per day.

So far, the FDA (U.S. Food and Drug Administration) has approved about 420 algorithms involving imaging for various diseases (mostly cancer), but FDA still requires that a human takes the final decision on what the algorithm detects.


Use cases of AI-assisted screening in radiology

Detecting things early and getting the patient to treatment much faster is the goal in computer-assisted screening. Here are some of the use cases of AI in radiology:

  • augmenting cardiac imaging
  • classifying brain tumors
  • detecting vertebral fractures
  • diagnosing ALS
  • detecting Alzheimer’s disease
  • detecting breast cancer
  • detecting pneumonia
  • detecting aneurysms
  • spotting embolisms and signs of stroke


Tracking and monitoring the patient

While computer-assisted triaging is the first step in integrating AI support in healthcare, machine learning is also becoming a powerful way to keep track of even the smallest changes in patients’ conditions, especially in cancer care, as patients undergo chemotherapy.


Predicting adverse events

With enough imaging data, AI algorithms could even find abnormalities for any condition that no human could spot, like certain biomarkers that can flag changes when someone is likely to have a stroke or heart attack.


Improving access to AI medical imaging in low and middle-income countries

AI already has the potential to improve workflows in radiology, and it could help with automatically detecting abnormalities in chest, brain and other body regions, which will have a considerable impact in LMICs. Free access to AI-supported medical imaging for high-priority diseases, such as tuberculosis, should be advanced.


How do we get there?

To be able to make these technologies happen, we would need tons of data.

But the data silos that exist in healthcare systems make it harder to access, share, collaborate and ultimately, augment medical imaging data.


Through Medicai, we have built a decentralized cloud infrastructure that allows healthcare systems to exchange data in a secure and compliant way, thus making possible new care pathways, data flows, and consequently, AI integrations that were previously unattainable.


Sources: 1, 2, 3, 4.

About the author - Andra Bria

Andra Bria is a growth marketer at Medicai. She is interested in health equity, patient experience and care pathways. She believes in interoperability and collaboration for a more connected healthcare industry.