Take a deep breath. You survived another RSNA conference.
While a few hardy souls are still enjoying educational sessions in the cozy confines of McCormick Place, the final day of the exhibit floor yesterday marks the end of RSNA 2023 for most attendees. And what a show it was.
Predictions were that AI would dominate the scientific sessions at RSNA 2023, a forecast that largely panned out. A November 28 session was a case in point, in which a series of top-quality papers were presented on one of the most promising use cases of AI, for breast screening:
A homegrown AI algorithm that analyzed screening breast ultrasound exams in addition to FFDM and DBT mammograms boosted sensitivity for detecting cancer in 12.5k patients, with better sensitivity for women with dense breasts (71% vs. 60%) and non-dense breasts (79% vs. 63%)
AI did a good job of detecting breast arterial calcification (BAC) when used prospectively to analyze screening mammograms in 16k women across 15 sites. It found 15% of women had BAC, a possible marker for atherosclerotic disease
Swedish researchers used their VAI-B validation platform to compare three AI algorithms (Therapixel, Lunit, and Vara) in 34k women, finding that using AI with a single radiologist boosted sensitivity 10-30% compared to double reading, with a slight loss in specificity (2-7%). VAI-B could be used to validate AI implementation and guide purchasing decisions
Why does AI miss some breast cancers? South Korean researchers addressed this question by analyzing 1.1k patients with invasive cancers in which AI had a miss rate of 14%. Luminal cancers were missed most often
Adding AI analysis of prior images to current studies with FFDM and DBT boosted sensitivity for cancer detection in 30k patients, with sensitivity the highest for two years of priors compared to no priors (74% vs. 70%)
This week’s research points to an exciting near-term future in which AI will help make mammography screening more accurate while helping breast radiologists perform their jobs more efficiently. Landmark studies toward this end were published in 2023 – this week’s RSNA conference shows that we can expect the momentum to continue in 2024.
It’s no secret that insurance reimbursement drives adoption of new medical technology. But a new analysis in NEJM AI shows exactly how reimbursement is affecting the diffusion into clinical practice of perhaps the newest medical technology – artificial intelligence.
Researchers analyzed a database of over 11B CPT claims from January 2018 to June 2023 to find out how often reimbursement claims are being submitted for the use of the over 500 AI devices that had been approved by the FDA at the time the paper was finalized.
The authors chose to focus on CPT claims rather than claims under the NTAP program for new technologies because CPT codes are used by both public and private payors in inpatient and outpatient settings, while NTAP only applies to Medicare inpatient payments.
They found 16 medical AI procedures billable under CPT codes; of these, 15 codes were created since 2021 and the median age of a CPT code was about 374 days, indicating the novelty of medical AI.
Also, only four of the 16 had more than 1k claims submitted, leading the authors to state “overall utilization of medical AI products is still limited and focused on a few leading procedures,” such as coronary artery disease and diabetic retinopathy.
The top 10 AI products and number of CPT claims submitted are as follows:
HeartFlow Analysis for coronary artery disease (67,306)
LumineticsCore for diabetic retinopathy (15,097)
Cleerly for coronary atherosclerosis (4,459)
Perspectum LiverMultiScan for liver MRI (2,428)
Perspectum CoverScan for multiorgan MRI (591)
Koios DS for breast ultrasound (552)
Anumana for ECG cardiac dysfunction (435)
CADScor for cardiac acoustic waveform recording (331)
Perspectum MRCP for quantitative MR cholangiopancreatography (237)
CompuFlo for epidural infusion (67)
While radiology may rule in terms of the sheer number of FDA-approved AI products (79% in a recent analysis), the list shows that cardiology is king when it comes to paying the bills.
Amid the breathless hype around medical AI, the NEJM AI study comes as a bit of a wake-up call, showing how the cold reality of healthcare economics can limit technology diffusion – a finding also indicated in other studies of economic barriers to AI.
On the positive side, it shows that a rosy future lies ahead for those AI algorithms – like HeartFlow Analysis – that can make the leap.
One of the biggest roadblocks in medical AI development is the lack of high-quality, diverse data for these technologies to train on.
What Is the Issue with Data Access?
Artificial Intelligence (AI) has emerged as a game-changer in the realm of medical imaging, with immense potential to revolutionize clinical practices. AI-powered medical imaging can efficiently identify intricate patterns within data and provide quantitative assessments of disease biomarkers. This technology not only enhances the accuracy of diagnosis but can also significantly speed up the diagnostic process, ultimately improving patient outcomes.
While the landscape is promising, medical innovators grapple with challenges in accessing high-quality, diverse, and timely data, which is vital for training AI and driving progress.
A 2019 study from the Massachusetts Institute of Technology found that over half of medical AI studies predominantly relied on databases from high-income countries, particularly the United States and China. If models trained on homogenous data are used clinically in diverse populations, then it could pose a risk to patients and worsen health inequalities experienced by underrepresented groups. In the United States, If the Food and Drug Administration deems these risks to be too high, then they could even reject a product’s application for approval.
In trying to get hold of the best training data, AI developers, particularly startups and individual researchers, face a web of complexities, including legal, ethical, and technical considerations. Issues like data privacy, security, interoperability, and data quality compound these challenges, all of which are crucial in the effective and responsible utilization of healthcare data.
One company working to overcome these hurdles in hope of accelerated and high-quality innovations is Gradient Health.
Gradient Health’s Approach
Gradient Health offers AI developers instant access to one of the world’s largest libraries of anonymized medical images, sourced from hundreds of global hospitals, clinics, and research centers. This data is meticulously de-identified for compliance and can be tailored by vendors to suit their project’s needs and exported in machine learning-ready DICOM + JSON formats.
By partnering with Gradient Health, innovators can use these extensive, diverse datasets to train and validate their AI algorithms, mitigating bias in medical AI and advancing the development of precise, high-quality medical solutions.
Gaining access to top-tier data at the outset of the development process promises long-term benefits. Here’s how:
Expand Market Presence: Access the latest cross-vendor datasets to develop medical innovations, expanding your market share.
Global Expansion: Enter new regions swiftly with locally sourced data from your target markets, accelerating your global reach.
Competitive Edge: Obtain on-demand training data for imaging modalities and disease areas, facilitating product portfolio expansion.
Speed to Market: Quickly acquire data for product training and validation, reducing sourcing time and expediting regulatory clearances for faster patient delivery.
“After looking for a data provider for many weeks, I was not able to get even a sample delivery within one month. I was immensely glad to work with Gradient and go from first contact to final delivery within one week!” said Julien Schmidt, chief operations officer and co-founder at Mango Medical.
In recent years, medical AI has experienced significant growth. Innovations in medical imaging in particular have played a pivotal role in enabling healthcare professionals to identify diseases earlier and more accurately in patients with a range of conditions.
Gradient Health offers a data-compliant, intuitive platform for AI developers, facilitating access to the essential data required to train these critical technologies. This approach holds the potential to save time, resources, and, most importantly, lives.
More information about Gradient Health is available on the company’s website. They will also be exhibiting at RSNA 2023 in booth #5149 in the South Hall.
It seems like watershed moments in AI are happening on a weekly basis now. This time, the big news is the Biden Administration’s sweeping executive order that directs federal regulation of AI across multiple industries – including healthcare.
The order comes as AI is becoming a clinical reality for many applications.
The number of AI algorithms cleared by the FDA has been surging, and clinicians – particularly radiologists – are getting access to new tools on an almost daily basis.
But AI’s rapid growth – and in particular the rise of generative AI technologies like ChatGPT – have raised questions about its future impact on patient care and whether the FDA’s existing regulatory structure is suitable for such a new technology.
The executive order appears to be an effort to get ahead of these trends. When it comes to healthcare, itsmajor elements are summarized in a succinct analysis of the plan by Health Law Advisor. In short, the order:
Calls on HHS to work with the VA and Department of Defense to create an HHS task force on AI within 90 days
Requires the task force to develop a strategic plan within a year that could include regulatory action regarding the deployment and use of AI for applications such as healthcare delivery, research, and drug and device safety
Orders HHS to develop a strategy within 180 days to determine if AI-enabled technologies in healthcare “maintain appropriate levels of quality” – basically, a review of the FDA’s authorization process
Requires HHS to set up an AI safety program within a year, in conjunction with patient safety organizations
Tells HHS to develop a strategy for regulating AI in drug development
Most analysts are viewing the executive order as the Biden Administration’s attempt to manage both risk and opportunity.
The risk is that AI developers lose control of the technology, with consequences such as patients potentially harmed by inaccurate AI. The opportunity is for the US to become a leader in AI development by developing a long-term AI strategy.
The question is whether an industry that’s as fast-moving as AI – with headlines changing by the week – will lend itself to the sort of centralized long-term planning envisioned in the Biden Administration’s executive order. Time will tell.
It’s no secret that the rapid growth of AI in radiology is being fueled by venture capital firms eager to see a payoff for early investments in startup AI developers. But are there signs that VCs’ appetite for radiology AI is starting to wane?
Maybe. And maybe not. While one new analysis shows that AI investments slowed in 2023 compared to the year before, another predicts that over the long term, VC investing will spur a boom in AI development that is likely to transform radiology.
Total funding has also fallen for the second straight year, to $501M year-to-date in 2023. That compares to $771M through the third quarter of 2022, and $1.1B through the corresponding quarter of 2021.
On the other hand, the average deal size has grown to an all-time high of $20.9M, compared to 2022 ($15.4M) and 2021 ($18M).
And one company – Rapid AI – joined the exclusive club of just 14 AI vendors that have raised over $100M with a $75M Series C round in July 2023.
In a look forward at AI’s future, a new analysis in JACRby researchers from the ACR Data Science Institute (DSI) directly ties VC funding to healthcare AI software development, predicting that every $1B in funding translates into 11 new product approvals, with a six-year lag between funding and approval.
And the authors forecast long-term growth: In 2022 there were 69 FDA-approved products, but by 2035, funding is expected to reach $31B for the year, resulting in the release of a staggering 350 new AI products that year.
Further, the ACR DSI authors see a virtuous cycle developing, as increasing AI adoption spurs more investment that creates more products available to help radiologists with their workloads.
The numbers from Signify and ACR DSI don’t match up exactly, but together they paint a picture of a market segment that continues to enjoy massive VC investment. While the precise numbers may fluctuate year to year, investor interest in medical imaging AI will fuel innovation that promises to transform how radiology is practiced in years to come.
What is autonomous artificial intelligence, and is radiology ready for this new technology? In this paper, we explore one of the most exciting autonomous AI applications, ChestLink from Oxipit.
What is Autonomous AI?
Up to now, most interpretive AI solutions have focused on assisting radiologists with analyzing medical images. In this scenario, AI provides suggestions to radiologists and alerts them to suspicious areas, but the final diagnosis is the physician’s responsibility.
Autonomous AI flips the script by having AI run independently of the radiologist, such as by analyzing a large batch of chest X-ray exams for tuberculosis to screen out those certain to be normal. This can significantly reduce the primary care workload, where healthcare providers who offer preventive health checkups may see up to 80% of chest X-rays with no abnormalities.
Autonomous AI frees the radiologist to focus on cases with suspicious pathology – with the potential of delivering a more accurate diagnosis to patients in real need.
One of the first of this new breed of autonomous AI is ChestLink from Oxipit. The solution received the CE Mark in March 2022, and more than a year later it is still the only AI application capable of autonomous performance.
How ChestLink Works
ChestLink produces final chest X-ray reports on healthy patients with no involvement from human radiologists. The application only reports autonomously on chest X-ray studies where it is highly confident that the image does not include abnormalities. These studies are automatically removed from the reporting workflow.
ChestLink enables radiologists to report on studies most likely to have abnormalities. In current clinical deployments, ChestLink automates 10-30% of all chest X-ray workflow. The exact percentage depends on the type of medical institution, with primary care facilities having the most potential for automation.
ChestLink Clinical Validation
ChestLink was trained on a dataset with over 500k images. In clinical validation studies, ChestLink consistently performed at 99%+ sensitivity.
“The most surprising finding was just how sensitive this AI tool was for all kinds of chest disease. In fact, we could not find a single chest X-ray in our database where the algorithm made a major mistake. Furthermore, the AI tool had a sensitivity overall better than the clinical board-certified radiologists,” said study co-author Louis Lind Plesner, MD, from the Department of Radiology at the Herlev and Gentofte Hospital in Copenhagen, Denmark.
In this study ChestLink autonomously reported on 28% of all normal studies.
In another study at the Oulu University Hospital in Finland, researchers concluded that AI could reliably remove 36.4% of normal chest X-rays from the reporting workflow with a minimal number of false negatives, leading to effectively no compromise on patient safety.
Safe Path to AI Autonomy
Oxipit ChestLink is currently used in healthcare facilities in the Netherlands, Finland, Lithuania, and other European countries, and is in the trial phase for deployment in one of the leading hospitals in England.
ChestLink follows a three-stage framework for clinical deployment.
Retrospective analysis. ChestLink analyzes a couple of years worth (100k+) of historic chest x-ray studies at the medical institution. In this analysis the product is validated on real-world data. It also realistically estimates what fraction of reporting scope can be automated.
Semi-autonomous operations. The application moves into prospective settings, analyzing images in near-real time. ChestLink produces preliminary reports for healthy patients, which may then be approved by a certified clinician.
Autonomous operations. The application autonomously reports on high-confidence healthy patient studies. The application performance is monitored in real-time with analytical tools.
Are We There Yet?
ChestLink aims to address the shortage of clinical radiologists worldwide, which has led to a substantial decline in care quality.
In the UK, the NHS currently faces a massive 33% shortfall in its radiology workforce. Nearly 71% of clinical directors of UK radiology departments feel that they do not have a sufficient number of radiologists to deliver safe and effective patient care.
ChestLink offers a safe pathway into autonomous operations by automating a significant and somewhat mundane portion of radiologist workflow without any negative effects for patient care.
So should we embrace autonomous AI? The real question should be, can we afford not to?
The ongoing tug of war over AI’s value to radiology continues. This time the rope has moved in AI’s favor with publication of a new study in JAMA Network Open that shows the potential of a new type of AI language model for creating radiology reports.
Headlines about AI have ping-ponged in recent weeks, from positive studies like MASAI and PERFORMS to more equivocal trials like a chest X-ray study in Radiology and news from the UK that healthcare authorities may not be ready for chest X-ray AI’s full clinical roll-out.
In the new paper, Northwestern University researchers tested a chest X-ray AI algorithm they developed with a transformer technique, a type of generative AI language model that can both analyze images and generate radiology text as output.
Transformer language models show promise due to their ability to combine both image and non-image data, as researchers showed in a paper last week.
The Northwestern researchers tested their transformer model in 500 chest radiographs of patients evaluated overnight in the emergency department from January 2022 to January 2023.
Reports generated by AI were then compared to reports from a teleradiologist as well as the final report by an in-house radiologist, which was set as the gold standard. The researchers found that AI-generated reports …
Had sensitivity a bit lower than teleradiology reports (85% vs. 92%)
Had specificity a bit higher (99% vs. 97%)
In some cases improved on the in-house radiology report by detecting subtle abnormalities missed by the radiologist
Generative AI language models like the Northwestern algorithm could perform better than algorithms that rely on a classification approach to predicting the presence of pathology. Such models limit medical diagnoses to yes/no predictions that may omit context that’s relevant to clinical care, the researchers believe.
In real-world clinical use, the Northwestern team thinks their model could assist emergency physicians in circumstances where in-house radiologists or teleradiologists aren’t immediately available, helping triage emergent cases.
After the negative headlines of the last few weeks, it’s good to see positive news about AI again. Although the current study is relatively small and much larger trials are needed, the Northwestern research has promising implications for the future of transformer-based AI language models in radiology.
In another blow to radiology AI, the UK’s national technology assessment agency issued an equivocal report on AI for chest X-ray, stating that more research is needed before the technology can enter routine clinical use.
The report came from the National Institute for Health and Care Excellence (NICE), which assesses new health technologies that have the potential to address unmet NHS needs.
The NHS sees AI as a potential solution to its challenge of meeting rising demand for imaging services, a dynamic that’s leading to long wait times for exams.
But at least some corners of the UK health establishment have concerns about whether AI for chest X-ray is ready for prime time.
The NICE report states that – despite the unmet need for quicker chest X-ray reporting – there is insufficient evidence to support the technology, and as such it’s not possible to assess its clinical and cost benefits. And it said there is “no evidence” on the accuracy of AI-assisted clinician review compared to clinicians working alone.
As such, the use of AI for chest X-ray in the NHS should be limited to research, with the following additional recommendations …
Centers already using AI software to review chest X-rays may continue to do so, but only as part of an evaluation framework and alongside clinician review
Purchase of chest X-ray AI software should be made through corporate, research, or non-core NHS funding
More research is needed on AI’s impact on a number of outcomes, such as CT referrals, healthcare costs and resource use, review and reporting time, and diagnostic accuracy when used alongside clinician review
The NICE report listed 14 commercially available chest X-ray algorithms that need more research, and it recommended prospective studies to address gaps in evidence. AI developers will be responsible for performing these studies.
Taken with last week’s disappointing news on AI for radiology, the NICE report is a wakeup call for what had been one of the most promising clinical use cases for AI. The NHS had been seen as a leader in spearheading clinical adoption of AI; for chest X-ray, clinicians in the UK may have to wait just a bit longer.
There’s no question AI is the future of radiology. But AI’s drive to widespread clinical use is going to hit some speed bumps along the way.
This week is a case in point. Two studies were published showing AI’s limitations and underscoring the challenges faced in making AI an everyday clinical reality.
In the first study, researchers found that radiologists outperformed four commercially available AI algorithms for analyzing chest X-rays (Annalise.ai, Milvue, Oxipit, and Siemens Healthineers) in a study of 2k patients in Radiology.
Researchers from Denmark found the AI tools had moderate to high sensitivity for three detection tasks:
airspace disease (72%-91%)
pleural effusion (62%-95%).
But the algorithms also had higher false-positive rates and performance dropped in cases with smaller pathology and multiple findings. The findings are disappointing, especially since they got such widespread play in the mainstream media.
Foundation models are designed to address the challenge of finding enough high-quality data for AI training. Most algorithms are trained with actual de-identified clinical data that have been labeled and referenced to ground truth; foundation models are AI neural networks pre-trained with broad, unlabeled data and then fine-tuned with smaller volumes of more detailed data to perform specific tasks.
Researchers in the new study found that a chest X-ray algorithm trained on a foundation model with 800k images had lower performance than an algorithm trained with the CheXpert reference model in a group of 42.9k patients. The foundation model’s performance lagged for four possible results – no finding, pleural effusion, cardiomegaly, and pneumothorax – as follows…
Lower by 6.8-7.7% in females for the “no finding” result
Down by 10.7-11.6% in Black patients in detecting pleural effusion
Lower performance across all groups for classifying cardiomegaly
The decline in female and Black patients is particularly concerning given recent studies on bias and lack of generalizability for AI.
This week’s studies show that there’s not always going to be a clear road ahead for AI in its drive to routine clinical use. The study on foundation models in particular could have ramifications for AI developers looking for a shortcut to faster algorithm development. They may want to slow their roll.
How can you predict whether an AI algorithm will fall short for a particular clinical use case such as detecting cancer? Researchers in Radiologytook a crack at this conundrum by developing what they call an “uncertainty quantification” metric to predict when an AI algorithm might be less accurate.
But AI isn’t infallible. And unlike a human radiologist who might be less confident in a particular diagnosis, an AI algorithm doesn’t have a built-in hedging mechanism.
So researchers from Denmark and the Netherlands decided to build one. They took publicly available AI algorithms and tweaked their code so they produced “uncertainty quantification” scores with their predictions.
They then tested how well the scores predicted AI performance in a dataset of 13k images for three common tasks covering some of the deadliest types of cancer:
1) detecting pancreatic ductal adenocarcinoma on CT 2) detecting clinically significant prostate cancer on MRI 3) predicting pulmonary nodule malignancy on low-dose CT
Researchers classified the highest 80% of the AI predictions as “certain,” and the remaining 20% as “uncertain,” and compared AI’s accuracy in both groups, finding …
AI led to significant accuracy improvements in the “certain” group for pancreatic cancer (80% vs. 59%), prostate cancer (90% vs. 63%), and pulmonary nodule malignancy prediction (80% vs. 51%)
AI accuracy was comparable to clinicians when its predictions were “certain” (80% vs. 78%, P=0.07), but much worse when “uncertain” (50% vs. 68%, P<0.001)
Using AI to triage “uncertain” cases produced overall accuracy improvements for pancreatic and prostate cancer (+5%) and lung nodule malignancy prediction (+6%) compared to a no-triage scenario
How would uncertainty quantification be used in clinical practice? It could play a triage role, deprioritizing radiologist review of easier cases while helping them focus on more challenging studies. It’s a concept similar to the MASAI study of mammography AI.
Like MASAI, the new findings present exciting new possibilities for AI implementation. They also present a framework within which AI can be implemented more safely by alerting clinicians to cases in which AI’s analysis might fall short – and enabling humans to step in and pick up the slack.
Get every issue of The Imaging Wire, delivered right to your inbox.
You're signed up!
It's great to have you as a reader. Check your inbox for a welcome email.