RSNA 2018 Artificial Intelligence Preview

Road to RSNA 2018: Artificial Intelligence Preview

By Erik L. Ridley, staff writer
October 29, 2018

Welcome to the first installment of this year's Road to RSNA preview of the RSNA 2018 meeting in Chicago. For the 10th year in a row, we're providing a modality-by-modality overview of selected scientific presentations to serve as your guide to events at McCormick Place.

Our journey along the Road to RSNA begins with our preview of artificial intelligence (AI), a fitting topic given its ever-growing prominence at the conference -- and in radiology overall -- over the last few years. Researchers will travel to the Windy City to share their experiences working with and developing AI applications in all aspects of the imaging process.

Interest continues to surge in applying AI as a triage tool, alerting radiologists or other physicians that imaging studies need urgent review. Several speakers will share promising results from using AI algorithms to automatically assess for fractures on spine CT studies, prescreen brain MRIs, and identify acute findings on routine abdominal CT studies, to name a few examples.

On a modality level, radiography is a particularly active area of imaging AI research. A number of RSNA 2018 presenters will describe how AI can find multiple diseases on chest x-rays, detect changes on serial imaging studies, or even power a portable automated x-ray reading system.

It's also no surprise that AI shows considerable potential in women's imaging; for years, mammography has been the only imaging modality to experience broad adoption of computer-aided detection (CAD) software. At RSNA 2018, researchers will discuss how AI can, for example, find more breast cancers than traditional CAD software, as well as analyze breast density. What's more, it can also enhance the accuracy of digital breast tomosynthesis screening.

Other hot topics this year include using AI to lower radiation dose and decrease scanning time, facilitate communication of urgent results, and analyze radiomics data to assess risk and predict treatment response and patient prognosis.

See below for previews of these and other AI-related scientific talks at this year's RSNA meeting. Of course, these are just a sample of the content on offer; a host of other scientific papers, scientific posters, refresher courses, and educational exhibits on AI topics also await attendees. For more information on those presentations and other abstracts in this year's scientific and educational program, view the RSNA 2018 meeting program.

But that's not all. Other notable AI educational opportunities at RSNA 2018 include the return of the Machine Learning Showcase, a dedicated area in the technical exhibits of the North Hall that features AI company booths and daily theater presentations. The most accurate algorithms from this year's Machine Learning Pneumonia Detection Challenge also will be highlighted in the Machine Learning Showcase.

The RSNA Deep Learning Classroom is back this year as well; certified instructors from the NVIDIA Deep Learning Institute will be on hand to help attendees learn to write AI algorithms and enhance their knowledge of AI. Want to contribute to AI research? You can once again annotate clinical images at the National Cancer Institute's Crowds Cure Cancer exhibit in the Learning Center.

Scientific and Educational Presentations
AI algorithm could decrease number of thyroid biopsies
Sunday, November 25 | 10:55 a.m.-11:05 a.m. | SSA12-02 | Room S406B
In this presentation, researchers will describe how the use of an artificial intelligence (AI) algorithm could obviate the need for many biopsies performed on thyroid nodules.
Free-text reports boost AI performance in chest x-rays
Sunday, November 25 | 11:55 a.m.-12:05 p.m. | SSA12-08 | Room S406B
A team from the U.S. National Institutes of Health (NIH) will describe how using free-text radiological reports in training an artificial intelligence (AI) framework can improve performance in classifying multiple results on chest radiographs.
AI taps prognostic power of CT lung cancer screening
Monday, November 26 | 10:30 a.m.-10:40 a.m. | SSC03-01 | Room E451A
By analyzing CT lung cancer screening exams, an artificial intelligence (AI) algorithm can predict the likelihood of five other major diseases, researchers from California will report in this presentation.
Deep learning elevates mammography CAD performance
Monday, November 26 | 11:10 a.m.-11:20 a.m. | RC215-13 | Arie Crown Theater
Mammography computer-aided detection (CAD) software based on deep learning can perform comparably to radiologists in detecting breast cancer and at a higher level than traditional mammography CAD applications, according to Dutch researchers.
AI speeds up DBT reading time, helps find more cancers
Monday, November 26 | 11:20 a.m.-11:30 a.m. | RC215-14 | Arie Crown Theater
In this talk, researchers will report that the concurrent use of artificial intelligence (AI) software while interpreting digital breast tomosynthesis (DBT) screening exams leads to higher radiologist accuracy and much faster reading times.
Deep learning can spot findings early on chest x-rays
Monday, November 26 | 11:40 a.m.-11:50 a.m. | SSC09-08 | Room E450A
Researchers from artificial intelligence software developer will share how their deep-learning algorithms can identify some abnormalities on chest radiographs even before they can be visualized by radiologists.
Portable automated x-ray system targets tuberculosis
Monday, November 26 | 11:50 a.m.-12:00 p.m. | SSC09-09 | Room E450A
In this talk, researchers will present a portable x-ray system and machine-learning image analysis application designed to make it easier to diagnose tuberculosis in developing and underdeveloped nations.
AI enables CT screening for osteoporosis fracture risk
Monday, November 26 | 3:00 p.m.-3:10 p.m. | SSE14-01 | Room E353C
Artificial intelligence (AI) algorithms could enable screening for the risk of osteoporosis-attributed bone fracture by analyzing CT scans that are already being performed for other purposes, according to researchers from Switzerland.
Deep learning may yield sharply lower dose from DBT
Monday, November 26 | 3:00 p.m.-3:10 p.m. | SSE23-01 | Room S502AB
In this scientific session, a multi-institutional team of researchers will share how deep-learning technology could lead to nearly 80% lower radiation dose to patients from digital breast tomosynthesis (DBT) studies.
Deep learning spots urgent results on radiology reports
Tuesday, November 27 | 10:30 a.m.-10:40 a.m. | SSG06-01 | Room N230B
Deep learning and natural language processing can automatically detect urgent findings on free-text radiology reports, U.S. researchers will report in this scientific presentation.
AI can detect 5 conditions on chest x-rays
Tuesday, November 27 | 10:30 a.m.-10:40 a.m. | SSG13-01 | Room S404AB
An artificial intelligence (AI) algorithm system can be highly accurate for detecting five specific findings on chest radiography studies, according to South Korean researchers.
Machine learning bests CAD-RADS for risk assessment
Tuesday, November 27 | 10:40 a.m.-10:50 a.m. | SSG02-02 | Room S104B
In this talk, researchers will report that machine learning performs better than the Coronary Artery Disease Reporting and Data System (CAD-RADS) for predicting death and coronary events on coronary CT angiography.
AI combs radiology reports for urgent findings
Tuesday, November 27 | 10:40 a.m.-10:50 a.m. | SSG06-02 | Room N230B
In this presentation, researchers will share another approach based on artificial intelligence (AI) for automatically flagging radiology results that need to be urgently communicated to referring physicians.
AI algorithm can triage suspected spine fracture cases
Tuesday, November 27 | 10:50 a.m.-11:00 a.m. | SSG08-03 | Room S102CD
A team of researchers from California found that an artificial intelligence (AI) algorithm can assess for the presence of spine fractures on CT exams, serving as a triage tool in the acute trauma setting.
Deep-learning algorithms need real-world testing
Tuesday, November 27 | 11:20 a.m.-11:30 a.m. | RC305-11 | Room S406B
Researchers from Boston will show why it's important for deep-learning algorithms to be tested on real-world cases prior to being deployed in clinical practice for image analysis tasks.
AI can prescreen brain MRIs as normal or abnormal
Tuesday, November 27 | 3:00 p.m.-3:10 p.m. | SSJ18-01 | Room E451B
Artificial intelligence (AI) can prescreen brain MRI studies and triage abnormal exams for neuroradiologists, according to researchers from New York.
Machine learning reduces MRI scanning times
Tuesday, November 27 | 3:30 p.m.-3:40 p.m. | SSJ18-04 | Room E451B
In this presentation, researchers will show how a machine learning-based image reconstruction algorithm can sharply lower scanning times for brain and lumbar MRI studies.
Deep-learning networks help classify breast density
Wednesday, November 28 | 10:40 a.m.-10:50 a.m. | SSK02-02 | Room E451B
Deep-learning networks can help classify breast tissue density, according to researchers from NYU Langone Medical Center.
In-house data train breast cancer detection algorithm
Wednesday, November 28 | 11:20 a.m.-11:30 a.m. | SSK02-06 | Room E451B
Researchers from California will share how deep-learning algorithms for breast cancer detection can be trained using an in-house image database.
Deep learning bolsters interpretation of chest x-rays
Wednesday, November 28 | 11:20 a.m.-11:30 a.m. | SSK05-06 | Room N227B
Researchers from South Korea will present their deep-learning algorithm for the automatic detection of major thoracic abnormalities on chest radiographs.
AI-based CAD can improve breast cancer detection
Wednesday, November 28 | 11:30 a.m.-11:40 a.m. | SSK02-07 | Room E451B
Computer-aided detection (CAD) software based on artificial intelligence (AI) can increase radiologist breast cancer detection rates, according to this scientific presentation.
Algorithm accurately detects changes on chest x-rays
Wednesday, November 28 | 11:40 a.m.-11:50 a.m. | SSK05-08 | Room N227B
In this talk, researchers will report that a machine-learning algorithm can perform better than radiologists in detecting changes in findings on serial chest radiographs.
AI algorithm can triage routine abdominal CT exams
Thursday, November 29 | 10:20 a.m.-10:30 a.m. | RC608-07 | Room E451B
A Swiss team will describe how an artificial intelligence (AI) algorithm can identify acute findings on routine abdominal CT scans, enabling radiologists to prioritize reading of these urgent exams.