Artificial Intelligence

Artificial intelligence (AI) is becoming a crucial component of healthcare to help augment physicians and make them more efficient. In medical imaging, it is helping radiologists more efficiently manage PACS worklists, enable structured reporting, auto detect injuries and diseases, and to pull in relevant prior exams and patient data. In cardiology, AI is helping automate tasks and measurements on imaging and in reporting systems, guides novice echo users to improve imaging and accuracy, and can risk stratify patients. AI includes deep learning algorithms, machine learning, computer-aided detection (CAD) systems, and convolutional neural networks.

Even AI struggles to work under stress, study suggests
Apparently, it isn’t just humans who occasionally struggle to work under stress. According to a recent study, the performance of AI flounders, too.
ChatGPT both passes and fails at translating free-text into structured reports
While the latest ChatGPT models show promise in easing the reporting burden on radiologists in the future, they aren't up to the task just yet.
GPT-4 spots radiology report errors in less time, at a lower cost
The large language model can identify report errors seven times more quickly than human readers, new study reveals.
AI able to assess invasiveness of lung lesions to aid in surgery
In a study, the most accurate model combined deep-learning with a radionomics approach.
First-in-human trial shows potential of guiding CABG with cardiac CT and AI—no ICA required
An independent heart team blinded to ICA results was able to deliver helpful guidance for CABG procedures for 99.1% of patients using just CCTA and FFR_CT alone. This approach is safe and feasible, researchers wrote, and the next step is to gather additional data.

An overview of artificial intelligence (AI) in radiology with Keith Dreyer with the ACR. Images shows a COVID-19 lung CT scan reconstruction from Siemens Healthineers. #AI #radAI #ACR

VIDEO: Overview of radiology AI by Keith Dreyer

Keith J. Dreyer, DO, PhD, FACR, American College of Radiology (ACR) Data Science Institute Chief Science Officer, explains the state of AI in radiology in 2022.

August 18, 2022

Example of a radiology diagnostic aid artificial intelligence (AI) algorithm with Lunit's mammography cancer lesion detection system.

VIDEO: Segmenting the Radiology Artificial Intelligence Market by Function

Keith J. Dreyer, DO, American College of Radiology (ACR) Data Science Institute chief science officer, breaks down radiology AI down into 4 areas and discusses where these areas stand with regulatory approval.

August 17, 2022

Addressing 'model drift' to recover AI performance before it leads to report errors

“Although regularly assessing and updating these models is necessary to ensure accurate performance, there is no standard approach to addressing model drift.”

August 16, 2022

Example of an artificial intelligence (AI) app store on the Sectra website, where Sectra PACS users can select the AI algorithms they want that are already integrated into the Sectra System. Other vendors have followed a similar approach to AI developed by many smaller vendors they partner with.

VIDEO: Development of AI app stores to enable easier access

Keith J. Dreyer, DO, PhD, FACR, American College of Radiology (ACR) Data Science Institute Chief Science Officer, explains how radiology vendors have developed AI app stores to make it easier to access new FDA cleared AI algorithms.

August 16, 2022

Keith J. Dreyer, DO, PhD, FACR, American College of Radiology (ACR) Data Science Institute Chief Science Officer, explains artificial intelligence (AI) for radiology. Dreyer also holds the positions of vice chairman of radiology at Massachusetts General Hospital, chief data science and information officer for the departments of radiology for both Massachusetts General Hospital and Brigham and Women's Hospital.

VIDEO: Where will radiology AI be in 5 years?

Keith J. Dreyer, DO, PhD, FACR, American College of Radiology (ACR) Data Science Institute Chief Science Officer, explains 5 developments to watch for in radiology artificial intelligence (AI).

August 16, 2022

How do radiologists really feel about adopting AI? New data offer insight

Up to 60% of radiologists have intentions of adopting artificial intelligence tools into clinical practice in the near future.

August 15, 2022

Educational platform releases AI course geared toward radiology administrators, technologists

The beginner course includes information on AI terminology, information technology principles, medical and legal considerations that accompany AI implementation and HIPAA compliance pertaining to the technology’s use in clinical settings.

August 10, 2022

Deep learning, subtraction technique ideal for evaluating stent re-stenosis on coronary CTA

Detecting in-stent restenosis via coronary CTA with hybrid iterative reconstruction has historically been an exercise in avoiding false positives.

August 10, 2022

Around the web

Radiology Business

Google teams with Bayer to address radiologist burnout via AI

"Gen AI can help tackle repetitive tasks and provide insights into massive datasets, saving valuable time," Thomas Kurian, CEO of Google Cloud, said Tuesday.

Cardiovascular Business

Cardiology, radiology groups urge insurance provider to rethink imaging policy

SCAI and four other major healthcare organizations signed a joint letter in support of intravascular ultrasound.

Cardiovascular Business

Radiology AI company gains FDA clearance for new CT offerings focused on PE, stroke

The newly approved AI models are designed to improve the detection of pulmonary embolisms and strokes in patients who undergo CT scans.

Artificial Intelligence

Even AI struggles to work under stress, study suggests

ChatGPT both passes and fails at translating free-text into structured reports

GPT-4 spots radiology report errors in less time, at a lower cost

AI able to assess invasiveness of lung lesions to aid in surgery

First-in-human trial shows potential of guiding CABG with cardiac CT and AI—no ICA required

Around the web