MRI Recon Gets Real with AI-Driven Protocols

AI-based data reconstruction for MRI scans took a step forward this week with studies showing how to generate 3T-like images from ultralow-field scanners, and improve scanner efficiency by cutting energy consumption.

MRI is radiology’s premier modality, but MRI scanners are cumbersome to install and expensive to operate. 

  • Ultralow-field scanners could help but some believe they lack the image quality for some clinical applications. 

Enter AI-based image reconstruction. Deep learning protocols are being developed for a wide range of imaging modalities, from PET to CT to MRI. 

  • These algorithms take images acquired with lower-quality input data – be it less CT radiation dose or lower MRI field strength – and upscale them to resemble full-fidelity images.

This trend is illustrated by research published this week in Radiology in which researchers tested a generative adversarial network algorithm called LowGAN for reconstructing data acquired on Hyperfine’s Swoop 0.064T portable ultralow-field MRI scanner.

  • Their goal was to enable Swoop to generate images resembling those acquired on a 3T system. 

After training LowGAN on paired 3T and 0.064T images, they tested the algorithm in 50 patients with multiple sclerosis and further validated it with a separate 13-patient cohort. They then judged LowGAN against several measures of MR image quality, finding that it …

  • Showed the biggest improvement on synthetic FLAIR and T1 images.
  • Improved conspicuity of white matter lesions, without introducing false lesions.
  • Increased consistency of cortical and subcortical volume measurements with 3T images.
  • But was unable to reveal brain lesions that were missed in the original low-field scans. 

AI-based data reconstruction also has environmental implications. Medical imaging is a major contributor to greenhouse gas emissions, and anyone who’s managed an MRI operation knows how much energy these massive scanners consume. 

  • A second paper published this week in Radiology described how MRI acceleration – scans acquired at a faster speed and then reconstructed for better image quality – reduced energy use, lowering carbon emissions while boosting imaging capacity. 

Researchers tried three techniques for speeding MRI acquisition – parallel acceleration, simultaneous multi-slice, and a deep learning algorithm. 

  • All three reduced energy consumption 21% to 65% and increased daily capacity by one to seven scanning slots, with deep learning showing the biggest effect.

The Takeaway

The new papers demonstrate an exciting future in which less powerful data acquisition technologies can be upscaled with AI to produce images that more closely resemble state-of-the-art scanning. The benefits will be enjoyed by both patients and the planet.

Patients Want Mammo AI, But Mostly As Backup

Patients support the idea of having AI review their screening mammograms – under certain conditions. That’s according to a new study in Radiology: Imaging Cancer that could have implications for breast imagers seeking to integrate AI into their practices.

Mammography screening has been identified as one of the most promising use cases for AI, but clinical adoption has been sluggish for reasons that range from low reimbursement to concerns about data privacy, security, algorithm bias, and transparency. 

  • Vendors and providers are working on solving many of the problems impeding greater AI use, but patient preference is an often overlooked factor – even as some providers are beginning to offer AI review services for which patients pay out of pocket.

To gain more insight into what patients want, researchers from the University of Texas Southwestern Medical Center surveyed 518 women who got screening mammography over eight months in 2023, finding …

  • 71% preferred that AI be used as a second reader along with a radiologist.
  • Only 4.4% accepted standalone AI interpretation of their images.
  • 74% wanted patient consent before AI review.
  • If AI found an abnormality, 89% wanted a radiologist to review their case, versus 51% who wanted AI to review abnormal findings by radiologists.
  • If AI missed a finding, 58% believed “everyone” should be accountable, while 15% said they would hold the AI manufacturer accountable. 

Patient preference for use of AI in collaboration with radiologists tracks with other recent research. 

  • Patients seem to want humans to retain oversight of AI, and seem to value trust, empathy, and accountability in healthcare – values associated with providers, not machines. 

The findings should also be good news for imaging services companies offering out-of-pocket AI review services. 

The Takeaway

The new findings should be encouraging not only for breast imagers and AI developers, but also for the imaging services companies that are banking on patients to shell out their own money for AI review. As insurance reimbursement for AI languishes, this may be the only way to move mammography AI forward in the short term.

Getting Paid for AI – Will It Get Easier?

Reimbursement is one of the major stumbling blocks holding back wider clinical adoption of artificial intelligence. But new legislation was introduced into the U.S. Congress last week that could ease AI’s reimbursement path. 

For AI developers, getting an algorithm approved is just the first step toward commercial acceptance. 

  • Perhaps even more important than FDA clearance is Medicare reimbursement, as healthcare providers are reluctant to use a product they won’t get paid for. 

Reimbursement drives clinical AI adoption, as evidenced by a 2023 analysis listing the top algorithms by CPT claims submitted (HeartFlow Analysis topped the list). 

  • But CMS uses a patchwork system governing reimbursement, from temporary codes like New Technology Add-On Payment codes that expire after 2-3 years to G-codes for procedures that don’t have CPT codes, on up to the holy grail of medical reimbursement: Category I codes. 

The new legislationS.1399 or the Health Tech Investment Act – would simplify the situation by setting up a dedicated Medicare coverage pathway for AI-enabled medical devices approved by the FDA (called “algorithm-based healthcare services”), as follows … 

  • All FDA-approved products would be assigned a Category III New Technology Ambulatory Payment Classification in the HOPPS program.
  • NTAPC codes would last for five years to enable collection of cost data before a permanent payment code is assigned. 
  • Payment classifications will be based on the cost of service as estimated by the manufacturer. 

The bill at present has co-sponsors from both political parties, Sen. Mike Rounds (R-SD) and Sen. Martin Heinrich (D-NM). 

  • The legislation has also drawn support from industry heavyweights like GE HealthCare and Siemens Healthineers, as well as industry groups like AdvaMed and others.

The Takeaway

The new bill sounds like a great idea, but it’s easy to be skeptical about its prospects in today’s highly charged political environment – especially when even bipartisan compromises like the 2025 Medicare fix got scuttled. Still, S.1399’s introduction at least shows that the highest levels of the U.S. government are cognizant of the need to improve clinical AI reimbursement.

AI Helps Radiologists Read Prostate MRI

MRI is changing how prostate cancer is detected, diagnosed, and followed up. But even a technology as powerful as MRI could use a little help, as evidenced by a new study in Radiology showing that a commercially available AI algorithm could help radiologists diagnose clinically significant prostate cancer. 

Workup of suspicious prostate lesions is being reshaped by MRI in meaningful ways.

  • For example, MRI-guided biopsy is replacing systemic prostate biopsy without guidance, especially for patients with low to intermediate risk of prostate cancer. 

But prostate MRI isn’t perfect – yet. Radiologist performance can vary due to differences in experience, as well as variations in MRI acquisitions, tumor location, and cancer prevalence. Could AI help even out these variations? 

  • To find out, researchers from South Korea tested Siemens Healthineers’ syngo.via Prostate MR algorithm in 205 patients suspected of prostate cancer who were scheduled for biopsy based on clinical information (including previous MRI scans).

The AI algorithm’s performance was compared to that of experienced radiologists, and researchers also estimated its impact on radiologist interpretation if used as a reading aid, finding that for clinically significant prostate cancer… 

  • AI had lower sensitivity versus radiologists (80% vs. 93%).
  • But higher positive predictive value (58% vs. 48%).
  • Adding AI to radiologists’ interpretation more than doubled specificity (44% vs. 21%).
  • There were no cancer cases among lesions rated by both the algorithm and radiologists as not likely to be cancer (PI-RADS 1 or 2).

AI’s higher PPV indicates that it could help reduce unnecessary prostate biopsies, while also detecting clinically significant cancer that might have been missed by radiologists.  

The Takeaway

The new findings echo previous studies that demonstrate the value of AI for MRI of prostate cancer, but differ in that they investigate a commercially available algorithm – indicating that tools for better prostate MRI are becoming accessible to radiologists. 

Bridging Quality and Efficiency: Why Radiology Groups Are Adopting AI for Mammography Workflows

By Dr. Roger Yang, President, University Radiology Group, and Mo Abdolell, CEO, Densitas

Radiology groups offering mammography services operate under ever-tightening demands, including MQSA EQUIP and ACR accreditation standards. Manual case selection, cumbersome paperwork, and lengthy review cycles often divert radiologists and technologists from what matters most – patient care.

But change is coming. By leveraging AI and mammography workflow automation, private radiology groups are reshaping how they manage quality, reduce administrative overhead, and advance patient care. 

AI-powered platforms can significantly streamline mammography quality management by:

  • Automating case selection for EQUIP reviews.
  • Measuring positioning metrics in near real-time.
  • Centralizing documentation to simplify compliance.

Some practices have reported up to a 90% reduction in EQUIP review time and 80% workload reduction in ACR accreditation using AI. But time savings are only part of the story.

Rather than waiting months for sporadic audits, technologists gain instant insights into positioning accuracy. This rapid feedback loop…

  • Accelerates targeted training.
  • Encourages continuous quality improvement.
  • Empowers technologists to self-monitor performance and identify gaps earlier. 

Today’s vendor-agnostic AI solutions integrate seamlessly with diverse imaging systems across multiple sites. 

  • Standards-based platforms can grow from a single mammography unit to dozens, helping radiology groups expand without adding complexity.

In a crowded marketplace, radiology practices that adopt AI-driven mammography quality management and automation stand out as forward-thinking leaders. Advantages include…

  • Enhancing patient perception: Offering efficient exams and high-quality imaging underscores a commitment to excellence, boosting satisfaction and referrals.
  • Leveraging analytics: Aggregated data on image quality and positioning helps leadership identify trends, optimize workflows, and highlight innovation.
  • Attracting top talent: Skilled technologists and radiologists gravitate toward practices with cutting-edge tools.

By integrating AI early, private practices can differentiate themselves, paving the way for growth and success.

Successful AI adoption and mammography workflow automation relies on more than just software. It requires:

  • Deep mammography expertise from vendors.
  • Robust training programs for staff.
  • Change training programs for staff.
  • Responsive customer support that fosters trust.

Mammography workflow automation cuts administrative burdens, curtails physician burnout, and speeds accreditation. Technologists receive clear, timely feedback, improving morale and performance. 

  • Meanwhile, patients benefit from streamlined workflows and consistent image quality, reinforcing trust in the practice.

The Takeaway

By embracing AI-driven mammography workflow automation and quality management, radiology groups can stay focused on delivering exceptional patient care while meeting regulatory requirements. This strategic investment propels private practices toward sustained growth and innovation, securing a competitive edge in a rapidly evolving healthcare landscape. Learn more.

Will FDA Staff Cuts Slow AI Adoption?

The Trump Administration’s campaign to cut the federal workforce arrived at the FDA last weekend – in particular its division regulating AI in healthcare. Multiple staff cuts were reported at the Center for Devices and Radiological Health, which had been in the midst of a major overhaul of AI regulation. 

A February 15 article in STAT News first reported the layoffs, which as with other recent staff reductions concentrated on FDA employees with probationary status and was part of a larger initiative that has also affected the CDC and NIH. 

The rapid growth of medical AI has had a major impact on the center, which as of its last report had given regulatory authorization to over 1k AI-enabled devices (76% of which are for radiology). 

  • To deal with the deluge, CDRH reportedly had been hiring many new staffers who were still on probationary status, making them targets for layoffs (permanent federal employees have civil service protections that make them harder to fire). 

FDA also has been retooling its regulatory approach to AI with new initiatives that reflect the fact that AI products continue learning (and changing) after they’ve been approved, and thus require more aggressive post-market surveillance than other medical devices…

So what impact – if any – will the layoffs have on the rapidly growing medical AI segment? 

  • The FDA may simply scale back its new AI initiatives and regulate the field under more traditional avenues that have served the medical device industry well for decades.

In another scenario, the FDA’s frenzied pace of AI approvals and initiatives could slow as the agency struggles to handle a growing number of product submissions with less staff. 

The Takeaway

The FDA layoffs couldn’t have come at a worse time for medical AI, which is on the cusp of wider clinical acceptance but still suffers from shaky confidence and poor understanding on the part of both providers and patients (see story below). The question is whether providers, organized radiology, or developers themselves will be able to step into the gap being left.

AI Enables Single-Click Cardiac MRI

Cardiac MRI is one of the most powerful imaging tools for assessing heart function, but it’s difficult and time-consuming to perform. Could automated AI planning offer a solution? A new research paper shows how AI-based software can speed up cardiac MRI workflow

Cardiac MRI has a variety of useful clinical applications, generating high-resolution images for tissue characterization and functional assessment without the ionizing radiation of angiography or CT.

  • But cardiac MR also requires highly trained MR technologists to perform complex tasks like finding reference cardiac planes, adjusting parameters for every sequence, and interacting with patients – all challenges in today’s era of workforce shortages. 

Cardiac MRI’s complexity also increases the number of clicks required by technologists to plan exams. 

  • This can introduce scan errors and produces inter-operator variability between exams. 

Fortunately, vendors are developing AI-based software that automates cardiac MR planning – in this case, Siemens Healthineers’ myExam Cardiac Assist and AI Cardiac Scan Companion. 

  • The solution enables single-click cardiac MR planning with a pre-defined protocol that includes auto-positioning to identify the center of the heart and shift the scanner table to isocenter, as well as positioning localizers to perform auto-align without manual intervention. 

How well does it work in the real world? Researchers tested the AI software against conventional manual cardiac MR exam planning in 82 patients from August 2023 to February 2024, finding that automated protocols had … 

  • A lower mean rate of procedure errors (0.45 vs. 1.13).
  • A higher rate of error-free exams (71% vs. 45%).
  • Shorter duration of free-breathing studies (30 vs. 37 minutes).
  • But similar duration of breath-hold exams (42 vs. 44 minutes, p=0.42).
  • While reducing the error gap between more and less experienced technologists. 

In their discussion of the study’s significance, the researchers note that most of the recent literature on AI in medical imaging has focused on its use for image reconstruction, analysis, and reporting.

  • Meanwhile, there’s been relatively little attention paid to one of radiology’s biggest pain points – exam preparation and planning. 

The Takeaway

The new study’s results are exciting in that they offer not only a method for performing cardiac MR more easily (potentially expanding patient access), but also address the persistent shortage of technologists. What’s not to like?

VC Investors Pivot to Quality

Venture capital investors in digital health firms pivoted to quality in 2024, with fewer deals done but a higher median deal size compared to 2023. That’s according to a new report from market analysis firm CB Insights that also documented a record high for both the number and value of AI-focused deals.

Digital health investment has fluctuated in the years since the COVID-19 pandemic, with the number of deals hitting a peak in 2021 but then receding. 

  • The first half of 2024 was particularly slow in the radiology AI sector, but funding seemed to accelerate in the second half, with more and larger deals getting done.

So where did venture capital funding for digital health end up for all of 2024? The CB Insights report found that relative to 2023 there was …

  • A 23% drop in the number of digital health funding rounds, to 1.2k deals, the lowest number since 2014, versus 1.6k deals.
  • A 3% increase in the total dollar value of investments, to $15.6B versus $15.1B.
  • A median deal size of $5.3M, up 39% versus $3.8M.
  • AI-focused companies secured 42% of funding and 31% of deals, up from 37% and 26%. 
  • The biggest imaging-related deal was a $106M Series C round raised by cardiac AI developer Cleerly.

The numbers are a sign of VC investors looking for quality companies that meet heightened benchmarks.

  • Investors want demonstrated progress in terms of clinical validation, commercial traction, and regulatory readiness before they’ll sign checks. 

The Takeaway

The new report illustrates the opportunities and challenges of the current investment environment for digital health. AI developers will find the wind shifting in their favor, but they will need to do their homework and show real progress in the clinical, commercial, and regulatory spaces before securing venture capital investment.

AI Guides Lung Ultrasound

Healthcare professionals with no experience in lung ultrasound were able to acquire diagnostic-quality scans comparable to those of experts thanks to AI guidance in a new paper in JAMA Cardiology

Ultrasound is one of the most versatile and cost-effective imaging modalities, but it is operator-dependent and many of its more challenging clinical applications require highly trained personnel. 

  • Echocardiography AI has already been shown to help novice healthcare personnel improve their skill to that of expert users – could AI also have applications in other areas, like lung ultrasound? 

To find out, researchers used Caption Health’s AI technology to guide lung ultrasound scans in 176 patients with clinical concerns for pulmonary edema from July to December 2023. 

  • Patients were scanned twice, once by an expert in lung ultrasound without AI guidance and once by a healthcare professional (registered nurses or medical assistants without formal ultrasound training) who received a short training session with lung guidance AI software. 

In analyzing the results, the researchers found …

  • Nearly all the scans acquired by healthcare professionals with AI assistance were of diagnostic quality. 
  • There was no statistically significant difference in quality between scans acquired by healthcare personnel and those of experts (98% vs. 97%, p=0.31).
  • AI-aided personnel actually performed better than experts in the lung area around the heart (91% vs. 77%), perhaps due to AI guidance. 
  • At 15 minutes, median scan acquisition times were longer than those reported in the literature (six and eight minutes). 

The findings could have major implications around access-to-care issues, with handheld ultrasound scanners distributed to low-resource areas where AI-guided healthcare professionals could perform scans sent to tertiary care centers for interpretation. 

The Takeaway

The new study demonstrates an exciting use case for AI in ultrasound that builds on previous research in echo AI. By giving more healthcare professionals access to the power of ultrasound, it promises to democratize access to care in many resource-challenged areas.  

AI As Malpractice Safety Net

One of the emerging use cases for AI in radiology is as a safety net that could help hospitals avoid malpractice cases by catching errors made by radiologists before they can cause patient harm. The topic was reviewed in a Sunday presentation at RSNA 2024

Clinical AI adoption has been held back by economic factors such as limited reimbursement and the lack of strong return on investment. 

  • Healthcare providers want to know that their AI investments will pay off, either through direct reimbursement from payors or improved operational efficiency.

At the same time, providers face rising malpractice risk, with a number of recent high-profile legal cases.

  • For example, a New York hospital was hit with a $120M verdict after a resident physician working the night shift missed a pulmonary embolism. 

Could AI limit risk by acting as a backstop to radiologists? 

  • At RSNA 2024, Benjamin Strong, MD, chief medical officer at vRad, described how they have deployed AI as a QA safety net. 

vRad mostly develops its own AI algorithms, with the first algorithm deployed in 2015. 

  • vRad is running AI algorithms as a backstop for 13 critical pathologies, from aortic dissection to superior mesenteric artery occlusion.

vRad’s QA workflow begins after the radiologist issues a final report (without using AI), and an algorithm then reviews the report automatically. 

  • If discrepancies are found the report is sent to a second radiologist, who can kick the study back to the original radiologist if they believe an error has occurred. The entire process takes 20 minutes. 

In a review of the program over one year, vRad found …

  • Corrections were made for about 1.5k diagnoses out of 6.7M exams.
  • The top five AI models accounted for over $8M in medical malpractice savings. 
  • Three pathologies – spinal epidural abscess, aortic dissection, and ischemic bowel due to SMA occlusion – would have amounted to $18M in payouts over four years.
  • Adding intracranial hemorrhage and pulmonary embolism creates what Strong called the “Big Five” of pathologies that are either the most frequently missed or the most expensive when missed.

The Takeaway

The findings offer an intriguing new use case for AI adoption. Avoiding just one malpractice verdict or settlement would more than pay for the cost of AI installation, in most cases many times over. How’s that for return on investment?

Get every issue of The Imaging Wire, delivered right to your inbox.

You might also like..

Select All

You're signed up!

It's great to have you as a reader. Check your inbox for a welcome email.

-- The Imaging Wire team

You're all set!