Pranav Rajpurkar, PhD, co-author of a new study on AI’s impact on radiologist performance.

Imaging Wire Sponsors

Annalise.ai  •  Arterys  •  Bayer  •  Blackford  •  CARPL.ai  •  Change Healthcare  •  Clearpath  •  Enlitic  •  GE HealthCare  •  Intelerad  •  Medality  •  Merge by Merative  •  Nuance  •  PocketHealth  •  Riverain Technologies  •  Siemens Healthineers  •  Subtle Medical  •  United Imaging  •  Us2.ai  •  Visage Imaging

Artificial Intelligence

Does ‘Automation Neglect’ Limit AI’s Impact?

Radiologists ignored AI suggestions in a new study because of “automation neglect,” a phenomenon in which humans are less likely to trust algorithmic recommendations. The findings raise questions about whether AI really should be used as a collaborative tool by radiologists. 

How radiologists use AI predictions has become a growing area of research as AI moves into the clinical realm. Most use cases see radiologists employing AI in a collaborative role as a decision-making aid when reviewing cases. 

But is that really the best way to use AI? In a paper published by the National Bureau of Economic Research, researchers from Harvard Medical School and MIT explored the effectiveness of radiologist performance when assisted by AI, in particular its impact on diagnostic quality.

They ran an experiment in which they manipulated radiologist access to predictions from the CheXpert AI algorithm for 324 chest X-ray cases, and then analyzed the results. They also assessed radiologist performance with and without clinical context. The 180 radiologists participating in the study were recruited from US teleradiology firms, as well as from a health network in Vietnam. 

It was expected that AI would boost radiologist performance, but instead accuracy remained unchanged:

• AI predictions were more accurate than two-thirds of the radiologists
• Yet, AI assistance failed to improve the radiologists’ diagnostic accuracy, as readers underweighted AI findings by 30% compared to their own assessments
• Radiologists took 4% longer to interpret cases when either AI or clinical context were added
• Adding clinical context to cases had a bigger impact on radiologist performance than adding AI interpretations

The findings show automation neglect can be a “major barrier” to human-AI collaboration. Interestingly, the new article seems to run counter to a previous study finding that radiologists who received incorrect AI results were more likely to follow the algorithm’s suggestions – against their own judgment. 

The Takeaway

The authors themselves admit the new findings are “puzzling,” but they do have intriguing ramifications. In particular, the researchers suggest that there may be limitations to the collaborative model in which humans and AI work together to analyze cases. Instead, it may be more effective to assign AI exclusively to certain studies, while radiologists work without AI assistance on other cases.

Memorial MRI’s Choice for Patient Comfort

Texas has one of the highest obesity rates in the US. So to ensure patient comfort, Memorial MRI & Diagnostic in Houston turned to United Imaging and its 3.0T uMR OMEGA MRI scanner with 75cm ultra-wide-bore. Learn more about their story.

Duke’s Case for ClearRead CT

This Riverain Technologies case study details how Duke University Medical Center integrated ClearRead CT into its chest CT workflows, reducing read times by 26% and improving nodule detection by 29%.

The Wire

• AWS Launches HealthImaging: AWS has officially launched its cloud storage offering, AWS HealthImaging. The launch comes with a slight rebranding of the service, which was originally introduced as AWS HealthLake Imaging in November 2022. HealthImaging enables healthcare providers to run medical imaging applications from a single copy of patient imaging data in the cloud, and AWS expects it will reduce total cost of storage ownership by 40%. Simultaneously, AWS launched AWS HealthScribe, which uses generative AI to create clinical documentation.

• National Review Targets ACR Racism Course: Conservative website National Review is taking aim at a course on systemic racism and white privilege available on ACR’s website that includes content on “identity and privilege.”  National Review accuses ACR of “pushing” the course on its members and of wading into “social-justice programming and the politicization of medicine.” In response, an ACR spokesperson said the group “supports health outcomes equity for all persons our members serve,” and pointed out that ACR cannot require its members to take any specific course.

• Alzheimer’s Imaging Underrepresentation: Minority groups are underrepresented in US-based studies using brain imaging to diagnose Alzheimer’s disease. So say USC researchers in an article in Communications Medicine in which they analyzed nearly 2.5k articles, finding that non-Hispanic whites made up 84-87% of study participants even though they make up less than 60% of the US population. The finding is especially concerning as Hispanics are 1.5X and African-Americans are 2X more likely to develop Alzheimer’s than Whites. 

• Mammo DL Predicts Breast Cancer Risk: MIT’s Mirai deep-learning tool performed better in predicting breast cancer risk for low- and intermediate-grade cancers compared with high-grade cancers (AUC=0.64 and 0.66 vs. 0.60). In Radiology: Artificial Intelligence, researchers noted that their dataset was enriched with 50% of cases from African-American women, in addition to cases from women with benign breast disease and BRCA mutation carriers, giving important insight into deep learning’s impact on high-risk populations.

• AI Falls Short for Dense Breasts: Meanwhile, AI fell short in a new study in AJR that compared three techniques for screening women with dense breasts. In 1.3k women, there was no statistically significant difference in sensitivity for mammography + AI versus mammography + ultrasound and mammography + AI and US (p>0.05). Meanwhile, mammo + AI had lower specificity than mammo + US (85.8% vs. 89.1%), lower accuracy (85.7% vs. 89.2%), and a higher recall rate (14.9 vs. 11.7), differences that were statistically significant (p<0.05).

• PocketHealth Update Tackles No-Shows: PocketHealth has released an update to its image-sharing platform designed to reduce missed appointments, which cost healthcare providers $200 per no-show. PocketHealth’s new Appointment Reminders feature sends personalized, automated communications to patients that include prep instructions and scheduling reminders to ensure continuity of care and operational efficiency. The reminders are sent via email or text messages, and PocketHealth estimates they can reduce no-shows by about 30%. 

• AI Expands Lung Screening Predictions: AI can provide important data on body composition measurements by analyzing CT lung cancer screening exams – and improving mortality predictions. Researchers in Radiology used AI to analyze data from nearly 21k people scanned in the National Lung Screening Trial (NLST), producing metrics like quantitative emphysema, coronary artery calcification, and fat mass and distribution. Adding AI analysis improved prediction of death from lung cancer and cardiovascular disease, as well as all-cause mortality. The study adds to evolving knowledge of AI’s value for opportunistic screening.

• FDA Clears Bunkerhill AI for CAC: The FDA has cleared an AI algorithm for detecting incidental coronary artery calcium (CAC) on routine non-gated chest CT scans. The iCAC algorithm was developed by Bhavik Patel, MD, when at Stanford University and is being commercialized on their behalf by Bunkerhill Health, a healthcare startup that translates AI algorithms developed by researchers to clinical use. The algorithm could play a role in opportunistic screening for heart disease risk. 

• ViewRay May Be Delisted: Beleaguered radiation therapy firm ViewRay said the Nasdaq stock market has begun the process to delist the company’s shares from the exchange. The move was prompted by ViewRay’s decision earlier this month to file for Chapter 11 bankruptcy protection, as well as concerns about its ability to maintain a minimum bid price of $1.00. Trading in ViewRay stock was scheduled to be suspended on July 26, and to begin trading on the over-the-counter (OTC) markets. 

• Personalized Radiology Reports: Looking for a way to deliver more personalized radiology reports to patients? Canadian startup Brell Health has launched a new reporting platform that enables practices to personalize reports by adding live specialist consultations with an integrated image viewer, as well as recorded video clips of anatomy and image findings. Practices can also add third-party AI applications, such as Brell’s own dictionary that explains medical terminology. Brell’s software includes optional integration with PACS, RIS, and EMR. Brell also offers radiology second opinions.

• Curium Inks AI Pact with EXINI: Curium is building momentum for an upcoming European launch of its PSMA-targeting PET radiotracer for diagnosing prostate cancer, Pylclari (piflufolastat fluorine-18). The company signed a deal to license and customize AI software from EXINI Diagnostics for interpreting and reporting of PSMA PET/CT scans, with the goal of establishing a reporting standard in the US and Europe. Pylclari was recommended for European marketing authorization in May 2023, and Curium already has a relationship with EXINI’s parent company, Lantheus. 

• Nanox to Raise $30M: Israeli digital imaging developer Nanox plans to raise $30M in a direct share offering with a single institutional investor. The company said it will use the funds for general corporate purposes as well as for further development of its Nanox.ARC digital X-ray system and Nanox.Cloud software. Nanox received FDA clearance in April for a version of Nanox.ARC with multiple X-ray sources, which the company believes will enable it to offer 3D tomographic images at a lower cost than existing imaging systems.

How Aster DM Healthcare Leveraged CARPL

See how Dubai-based healthcare leader Aster DM Healthcare leveraged the CARPL platform to connect its doctors, data scientists, and imaging workflows, and support its AI projects and development infrastructure.

Delighting Patients with Medical Image Sharing

A new platform from Clearpath now enables healthcare providers to delight their patients by sharing images and medical records digitally. Find out how it integrates simply into your practice.

Change Healthcare’s Secure Cloud

Did you know one quarter of healthcare organizations have experienced a cyber-attack in the last year? This Change Healthcare animation explains how 3rd-party certified cloud-native enterprise imaging can help secure IT infrastructure that might be exposed with re-platformed imaging systems.

The Resource Wire

• How is fast, quantitative SPECT/CT supporting quicker assessment of treatment response in patients with metastatic prostate cancer? Find out in this case study from New York-Presbyterian Hospital/Weill Cornell Medical Center, provided by Siemens Healthineers. 

• There are about 40 million imaging-related diagnostic errors occurring every year. Find out how Annalise.ai’s comprehensive AI solutions can reduce errors and improve diagnostic performance in this demo. 

• This Bayer Radiology white paper details how the right people, plan, and systems can help imaging teams achieve their dose management goals.

• What were the top five topics from SIIM 2023? Brad Levin of Visage Imaging takes a look at what was hot in Austin this summer, with interest in CloudPACS topping the list. 

• In a session from SIIM 2023, radiology industry thought leaders discussed AI’s return on investment. We covered some of the high points in this interview with Matt Lungren, MD, and Sander Kloet of Nuance Communications. 

• Efficiency and quality are the name of the game at RadNet, and that’s exactly what the imaging center giant achieved when it adopted Subtle Medical’s SubtleMR solution, optimizing its already-accelerated MRI protocols by 33-45% while maintaining consistent diagnostic image quality.

• What kinds of medical imaging exams are radiologists reading today, and how confident do they feel across different subspecialties? Find out in the 2023 Radiology Practice Development Report from Medality.  

• How can echo AI be mobilized to automate the fight against heart disease? Us2.ai President Yoran Hummel, PhD, discusses how echocardiography can be optimized in an interview on The Medical AI Podcast. 

• Today’s state-of-the-art CT scanners are advancing diagnostic capabilities by leveraging spectral acquisition to enhance lesion detection, tissue characterization, and metal artifact reduction. Learn about photon-counting CT’s potential in this GE HealthCare white paper. 

• Interest is growing in collecting patient-reported outcome measures (PROMs) in healthcare, but how can your hospital implement a PROM collection system? Learn how to get started in this article from PocketHealth. 

• What tools are available to help radiologists work remotely? In this case study, teleradiology provider 4ways Healthcare of the UK describes how they used Merative’s Merge PACS 8.0 platform to improve their service to clients while supporting remote radiologists.

• More than 80% of health systems are reporting shortages in their radiology departments. Learn the countermeasures you can take to reduce burnout in this ITN Online article by Intelerad President Morris Panner. 

• Telemedicine Clinic (TMC) is a leading teleradiology provider in Europe, and it saw a 10% increase in productivity after partnering with Enlitic and its Curie|ENDEX technology for eliminating data variability and standardizing hanging protocols. 

• Check out this Blackford Analysis white paper detailing how children’s hospital imaging teams can leverage AI to improve modality throughput and imaging device availability.

• See how AI and the cloud combine to alleviate IT challenges and amplify radiologist performance in this Arterys white paper.