Discover the news about Guerbet Digital Solutions!
You can also find them or discover more information on

Meaning of AI in Radiology

9 June 2022 - Meaning of AI in Radiology

Artificial Intelligence (AI) and big data have transformed our basic human understanding of the technologies with which we interact every day. From the products, we order online, to the type of social media content we see each day – algorithm and AI, by design enables us to shift towards more digitized innovation, creating a trail by which companies are now following suit.

Originally, many considered the use, and implementation of Artificial Intelligence somewhat foreign, regarding the misuse of its scope of practice as a way to replace human labor by all means. While these accusations may be considered somewhat novel, AI has become more than computerized coding and software.

While we’re well aware of its current capabilities, and how it’s already enhanced our daily lifestyles, AI in the field of radiology, and healthcare, in general, is now playing a key part in the advancement of uncovering new patterns of diagnosis.

It’s now become somewhat ambiguous to think, or even consider that AI or machine learning can completely replace the human labor force.

Perhaps it’s rather more important to consider how these technologies have helped eliminate mundane and outdated tasks from the work environment, enabling practitioners and radiologists to increase productivity, improve imaging analysis and discover the development of additional conditions within patients.

To ensure a better understanding of the general concept regarding how AI can become and has already become beneficial in the field of radiology and medical imaging, a review on the practice thereof has been examined below.


A survey published by Definitive Healthcare revealed that roughly one-third of hospitals and imaging centers currently make use of Artificial Intelligence (AI) throughout the developing and data analysis process., with imaging centers reporting slightly higher utilization.

But although some have questioned the rapid shift towards the full implementation or rather partial utilization of AI, a majority or 57.3% of surveyed professionals claim that AI drives better patient care and improves overall accuracy.

Striking balance between having AI as a means of accuracy, while the radiologist still covers the broader spectrum or scope of practice is possible, and it’s now become more standard throughout most imaging practices to accept the adoption of technological aids.

Here’s how machine learning radiology and radiology AI can help in the practice environment.


The pandemic may have caused severe economic and socio displacements for millions of inhabitants, it’s important to look at how prolonged lockdowns and government isolation protocols increased pressure on healthcare professionals after the height of the pandemic between March 2020 and October 2020.

Although ongoing efforts to help curb the spread and the distribution of vaccines have lifted a lot of the barriers that stood between development and putting a damper on the pandemic; healthcare professionals were faced with the brunt of the pandemic.

Excessive working hours, labor shortages, and exposure to high environments are among some of the chronic reasons why healthcare facilities were unable to cope with an unprecedented amount of hospitalizations as found by a World Health Organization report.

But AI and other machine learning or big data software platforms have been able to take these challenges head-on.

A research study, focusing on radiologists in practice, found that 26.5% of them mentioned that AI helps to improve workflow. These implementations now mean that radiologists can assist more patients throughout their day and diagnose a higher number of publications per day.

This does not only mean more accurate and precise analysis, it also assists professionals to deal with increased demand, and a decreasing labor force.


Between 2017 and 2018, the number of imagining publications viewed per practitioner stood at more than 1000-1100 per year, that number was merely 100-150 per year between 2007 and 2008. What this indicates is that AI has not only brought innovation, it’s enabled radiologists to improve workflow and process facilitation immensely.

What AI has incorporated is how it does not only automate the majority of the process, but it can build an algorithm that can support the ongoing analysis. Additionally, we see that the radiologist, with the aid of artificial software, can upload and directly link records with patient files or portfolios.

This makes it easier, for both parties, to actively share imaging tests, without the need for a physical in-person consultation. This draws back to the importance of how rapidly AI has advanced the field of medical imagery, as the technology can enhance decision-making procedures, automate scanning and present it more visually through means of established communication platforms.


A common theme among radiologists is to uncover patterns and structures within their images. The analysis is the process by which the radiologist can see different patterns of illness and chronic disease development in a wide range of patients.

Looking at the application of AI in radiology brings to life a comparable algorithm that helps the radiologists through the process of diagnosing pneumonia with chest X-rays, or the detection of vertebral fractures through the use of standard spinal radiology.

These are among some of the most common patterns radiologists can build and uncover as advancements improve, and technology becomes more accessible within the field of practice.

Ultimately, examinations are improving on their accuracy, and radiologists can establish a basic platform of structures and patterns, which for the short term results delivers efficiency.


When it comes to radiology AI, it’s not possible to list all of its capable benefits without mentioning the most important which have already seen making immense improvements in the practice itself and overall medical imaging procedures.


In the simplest ways, we understand that radiology imaging, from MRI and CAT scans to the more standard X-rays is simply about extracting data and information from the patient.

This has been the notion for quite some time, and while it’s helped advance the industry and push it towards new and better innovations, it’s now possible for practitioners and radiologists to extract more accurate data through discrete medical images.

It’s not just on the most basic principles that we see how AI is already changing the way how radiologists and scientists can extract and stack important imagery data or information.

A recent study from the Department of Diagnostic Radiology and Nuclear Medicine at the University of Maryland School of Medicine in Bal­timore found that deep learning can extract copious amounts of information regarding stroke and evaluation and overall procedure management.

Although on the surface, the complexity of this research indicates how far we’ve already mastered the profession, yet, how much is still to be discovered and unraveled.


Although the Federal Drug Administration has until recently only approved 28 algorithms, others that have also proven to be as successful can assist in tracing supplementary conditions such as the identification of pulmonary nodules or perhaps mitigating the required dose of CT.

Interestingly enough, at the time of publication in 2019, studies found that artificial data and synthetic algorithms can discover a whole different range of conditions that have gone somewhat unnoticed by radiologists.

At the time, pneumothorax, rib fractures, pulmonary embolism, and other abdominal algorithms were still awaiting FDA approval but have already been used within discretion to offer radiologists a solution-based approach.

While primary objectives are still somewhat relevant in the current execution process, and utilization has slowly taken better form – there are more creator possibilities for the currently approved algorithms which will hopefully see widespread adoption in the coming years.


While from an imaging and practitioner viewpoint, Artificial Intelligence proves to offer better imaging quality and data collection. These assets, among others, can be directly linked to patients assisting them in having a better understanding of their current condition, speeding up diagnosis, and allowing access to more affordable imaging.


The adoption of several types of technologies has made it easier for patients to have access to affordable and high-quality health care. With this is also the consideration that technology and software-based practices make it a lot faster for them to have reliable results without having to endure painstaking waiting periods.

Research in this field revealed that imaging diagnosis can help speed up more accurate treatment, the management thereof, and provide optimal outcomes for patients. These factors are applicable in high, middle, and low-income environments, with a more prominent focus, shifted towards lower income brackets.


The current state of AI in radiology has been primarily focused on two interventions, the first being to improve the overall workflow for radiologists, and the second to present more accurate diagnoses.

In a separate branch, time sensitivity for severe conditions was seen to help deliver improved diagnostic accuracy through the means of decreasing a missed diagnosis and preventing unnecessary human intervention.

We can now see that AI and computerized-algorithms work in combined efforts to decrease the burden of the disease on the patient.

While patients themselves might not completely be in control of the diagnosis process, as this is left to the radiologist and second-party healthcare professionals, it enables them to have better transparency and knowledge of their current state or condition.

These efforts can help eliminate, or perhaps alleviate disease stress, anxiety, or uncertainty which can be common traits among patients awaiting further results on their medical screening and imaging.


It’s perhaps now more clear that Artificial Intelligence, with overbearing assistance from human capabilities, has the chance to enhance and improve the scope of radiology practice. From practitioner to patient, various aspects can positively impact the advancement of radiology.

Unfortunately, there are some challenges and concerns which have been raised, both within and without the radiology community. While these are perhaps sometimes overshadowed by the widespread support for AI, it’s highly recommended that attention-driven solutions can be established to resolve novelty issues currently facing the industry.


As already mentioned, there are currently a few dozen algorithms that have already been approved by the FDA. While this makes it possible for faster and more widespread adoption, it’s not completely possible to cover the entire scope of practice if minimal standardization procedures have yet been established.

This does not only place a burden on the patient but the overall concept of radiology AI as well. A lack of standardization can lead to issues regarding;

  • Imaging accuracy across practices
  • Transparent algorithm utilization
  • Radiologist bias
  • Algorithm bias
  • Radiologists reliability


Personal information and patient confidential material are perhaps the biggest concern and challenges when it comes to the use of AI and other deep learning mechanisms.

For starters, it’s understandable that patients would enjoy privacy and data protection when it comes to their medical procedures and doings, now the same is said for the information obtained through AI and widespread governance thereof.

It creates an open-ended question that puts the patient information at the center. When new algorithms are established to help discover various conditions, how certain can patients be that their information and data are kept private and confidential?

Are radiologists openly allowed to make use of their findings to publish research material thereof, or perhaps utilize it as a way to further entice new advancements in their field?


A third challenge to consider, is in the event of false or inaccurate data delivery, who remains liable for the mistreatment thereof? Do we blame machines or perhaps physicians or radiologists for accounted errors?

While AI can see that algorithms are applicable where necessary, it remains in the best interest of the patient and their condition that radiologists oversee the final results and accuracy of their findings.

This remains a major challenge for many radiologists, as machines are necessary for deeper execution, but could also tend to deliver inaccurate data.


While many deem the use of Artificial Intelligence a necessity within the practice of radiology, there is still much we still need to uncover before further widespread adoption can be made. So far, the utilization of AI has proven to deliver positive results and help improve the overall workflow for radiologists who are working with increased demand, and higher number of images.   

AI has managed to establish a prominent position within the practice of radiology, enhancing imaging quality, accuracy, and data collection and making medical imaging more affordable and accessible.

It’s possible that the future of radiology could perhaps become completely computer-based, but this would still take quite a bit of time before proven successful. Until then, it will remain within the hands of radiologists to strike a balance between humans and machine.

Blog Press Release

27 June 2018 - ECR2018

Last and final day of #ECR2018! There is still time to meet our experts and attend our live product demos! Booth #220

Contrast&Care Dose&Care

15 June 2018 - Digital Solutions to control patient x-ray exposure

Gilles Hameury, Digital Services Global Product Manager at Guerbet, speaks about Digital Solutions to control patient x-ray exposure.


15 June 2018 - [RSNA17] Guerbet’s ambitions in Digital Services

Levi Cheng, Head of ISS Business Unit (Imaging Solutions and Services), highlights the Guerbet’s ambitions in Digital Services.


15 June 2018 - [RSNA17] Digital Services family and its new brand identity

Adan Martin, Head of Strategic New Business ISS (Imaging Solutions and Services), tells us more about the Digital Services family and its new brand identity at RSNA 2017


31 May 2018 - GEAR 2023

Guerbet Group management held a Capital Markets Day yesterday to release its GEAR 2023 strategic plan. Find the presentation in video by clicking on the following link:


30 May 2018 - CEO Yves L’Epine

Watch our CEO Yves L’Epine commenting on @GuerbetGroup’s 2017 annual results and ambitions:


22 May 2018 - #AOCNR2018

Hello #AOCNR2018! Our #DigitalServices team will perform products demos all along the congress, come to visit us at our booth and meet our experts!


I certify to be a Healthcare Provider Licensed to Practice


* Mandatory information
By clicking here, you are agreeing to receive marketing emails and other promotional communications from Guerbet from time to time.
You have the right to access and rectify your personal information.
For more information relating to our privacy practices, click here