AI News, Articles in Press: Journal of the American College of Radiology artificial intelligence

The Current State of Artificial Intelligence in Medical Imaging and Nuclear Medicine

AI holds the potential of extracting pertinent information from such sources to allow radiologists access to key information for reporting.29 Since its introduction, CT has evolved as an important imaging modality which is now used 24/7 in hospitals.

Deep learning techniques are now able to map what structures look like in low dose imaging compared to regular dose imaging, generating diagnostic quality comparable to regular CT imaging.38–40 Whilst MRI does not have the same radiation issues, one of the biggest limitations is the relatively long acquisition time.

Chaudhari et al used deep learning to reduce acquisition time by improving image quality of thicker MR sections, comparable to that of thin sections by interpolating data.41 Furthermore, Hyun et al achieved faster MRI by presenting a deep learning method that used 29% of k-space data to generate images comparable to standard MRI reconstruction with fully sampled data.42

The impact of artificial intelligence in medicine on the future role of the physician

According to the authors, machine learning research is becoming more important for the future because medicine is now a data-rich quantitative field due to the increasing prevalence of Electronic Medical Records (EMR) and such research is important to make sense of this data and to create discovery-support systems that support the work of a human physician (Patel et al., 2009).

The author believes that the next breakthrough will come not from some new scanner technology but as a courtesy of AI utilizing the imaging data that is already available from imaging technologies such as ultrasound, CT, MRI, and PET (King Jr, 2018).

Though the initial reaction of radiologists to the automated interpretation of images may not be welcoming, the radiologists overall had a sense of optimism about the future of radiology with the machine learning based technologies augmenting the quality of care (Kruskal et al., 2017).

Radiologists, it was felt, should remain active in the development of these new technologies by engaging with industry to ensure the ethical usages and clinical relevance of developing technologies (Kruskal et al., 2017).

Deep CNNs automatically learn mid-level and high-level abstractions obtained images and recent results show that CNNs are extremely effective in object recognition and localization in natural images (Greenspan, Van Ginneken &

(2017) summarize that while deep learning techniques in quantitative brain MRI has made big strides, “it is still challenging to have a generic method that can deal with all variations in brain MR images from different institutions and MRI scanners”.

recommends that physicians should learn how AI works in healthcare, companies such as IBM should increase awareness in the general public about the advantages and risks of using AI in medicine, and healthcare institutions should measure the effectiveness of the AI-based system (Meskó, 2017).

One of these areas is cognitive assisted robotics which is considered minimally invasive whereby large incisions are replaced with a series of quarter-inch incisions and utilize miniaturized surgical instruments.

The authors conclude that while AI is very promising for BC screenings, to translate this promise into reality requires very large data-sets of imaging examinations that are associated with clinical factors to train AI models (Houssami et al., 2017).

The authors report that critics of screening have suggested that a significant percentage of screen-detected breast cancers constitute overdiagnosis or cases that would not have become clinically significant during women’s lifetimes (Trister, Buist &

The authors report that AI algorithms in digital mammography have converted single whole digital images of the breast into automatically extracted quantitative, pixel-level variables which are unrecognizable to the human eye (Trister, Buist &

The ultimate promise is that AI can combine these pixel-level variables and associations with patient clinical data, including any known patient risk factors, to develop predictive algorithms that may someday provide equal or better accuracy than human screening mammography (Trister, Buist &

The author writes that processing healthcare related big data offers the promise of unlocking new insights and accelerating breakthroughs in medicine which in turn can help improve clinical practice.

Physicians employ personal medical histories, physical exams, individual biomarkers, and scores such as CURB-65 (severity score that estimates mortality of community-acquired pneumonia) to diagnose patients (Krittanawong, 2018).

AI on the other hand can assist physicians by diagnosing diseases using “complex algorithms, hundreds of biomarkers, imaging results from millions of patients, aggregated published clinical research, and thousands of physician’s notes from EMRs”

The author reports that researchers from the Google Brain project had reported in 2016 that the deep learning AI system had taught itself to accurately detect diabetic retinopathy (DR) and diabetic macular edema in fundus photographs (Roach, 2017).

According to the article, AI-based systems are likely to be used by ophthalmologists as just another tool in their tool set to diagnose eye diseases (Roach, 2017) and to improve the sensitivity to detect patients at risk and the specificity to correctly identify those without disease.

Who is winning the artificial intelligence race?

Read the full story on the Stanford Engineering website: In a recent talk at Stanford, Kai-Fu Lee says China has taken the lead. Lee is ..

Black-Box Medicine: Legal and Ethical Issues

Black-box medicine—the use of opaque computational models to make care decisions—has the potential to shape health care by improving and aiding many ...

Lesson 1: Deep Learning 2019 - Image classification

Note: please view this using the video player at instead of viewing on YouTube directly, to ensure you have the latest information. If you have ..

Dean Ornish: "Transforming Lives and Healthcare" | Talks at Google

For over 35 years, Dr. Dean Ornish's series of scientific research studies have been empowering the way millions of people think about the simple lifestyle ...

2018 ABOR OFR Session 1 of 2

The Arizona Board of Regents unanimously accepted the new strategic plan for the University of Arizona following this detailed presentation by UA President ...

Updates from SIIM with Cheryl and Jim, Feb 2019 - Episode 2

Health Care Stakeholder Discussion: Physicians

Director of the White House Office on Health Reform Nancy-Ann DeParle holds a stakeholder discussion with over 30 medical leaders from across the country to ...

IBM Watson and Personalized Medicine: Dr. Eliot Siegel

Dr. Eliot Siegel from the Department of Diagnostic Radiology, University of Maryland presents: The Jeopardy match between the two best human players of all ...

David Arky - Still Life & X-Ray Photographer

New York City-based award-winning photographer David Arky is best known for his conceptual still life and x-ray images. Arky's endlessly inventive photographs ...

Leadership in Medicine, 2020 and Beyond--What Will Doctors Be?

March 27, 2019 Christine K. Cassel Even as the University of Virginia and other medical schools across the U.S. prepare to graduate a new wave of physicians, ...