Breaking News
December 11, 2018 - Krystal 2000 microplate design improves fluorescence and luminescence measurement
December 11, 2018 - FDA clears mobile medical app to help increase retention in recovery program for opioid use disorder
December 11, 2018 - Overcoming Challenges in High-Speed Centrifugation Experiments
December 11, 2018 - Study shows link between neighborhoods’ socioeconomic status and dietary choices
December 11, 2018 - Lower BMI before obesity surgery predicts greater post-operative weight loss, study finds
December 11, 2018 - Obesity May Be Driving Rise in Uterine Cancers
December 11, 2018 - Antioxidants may prevent cognitive impairment in diabetes
December 11, 2018 - Oral cancer prognostic signature identified
December 11, 2018 - How Can I Find Out What Caused My Miscarriage?
December 11, 2018 - Novel personalized medicine tool for assessing inherited colorectal cancer syndrome risk developed
December 11, 2018 - Study uncovers 11 new genes associated with epilepsy
December 11, 2018 - Filling research gaps could help develop more disability-inclusive workplaces
December 11, 2018 - Cartilage tissue engineering brings good news for patients with cartilage defects
December 11, 2018 - Novel 3D printing workflow helps predict leaky heart valves
December 11, 2018 - Imagination can help overcome fear and anxiety-related disorders, shows study
December 11, 2018 - Are caries linked to political regime?
December 11, 2018 - Leader in Diabetes Clinical Trials Wins Naomi Berrie Award
December 11, 2018 - Scientists discover cellular mechanism that triggers pneumonia in humans
December 11, 2018 - Increasing mental health problems related to drug use in over 55’s
December 11, 2018 - High-intensity interval exercise could help combat cognitive dysfunction in obese people
December 11, 2018 - Annual flu shot can save lives of heart failure patients
December 11, 2018 - Researchers compare health outcomes for VA and non-VA hospitals
December 11, 2018 - Recommendations Developed for Psoriatic Arthritis Treatment
December 11, 2018 - Genetic analysis links obesity with diabetes, coronary artery disease
December 11, 2018 - Study shows that having genetic information can affect how the body responds
December 11, 2018 - UNAIDS Report: 9 Million Are Likely HIV Positive And Don't Know It
December 11, 2018 - Lund University researchers succeed in obtaining dendritic cells by direct reprogramming
December 11, 2018 - Breast tumors recruit bone marrow cells to boost their growth, study reveals
December 11, 2018 - Updated breast cancer screening guideline highlights importance of shared decision-making
December 11, 2018 - EHR-related stress associated with physician burnout
December 11, 2018 - AHA: 12-Year-Old Heart Defect Survivor Inspires NFL Player’s Foundation
December 11, 2018 - Breast cancer patients who take heart drug with trastuzumab have less heart damage
December 11, 2018 - Providing aid to those humans – and animals – affected by the California fires
December 11, 2018 - Even without proof, CBD is finding a niche as a cure-all
December 11, 2018 - Drawing leads to better memory than writing
December 11, 2018 - Researchers report novel findings on plant hormone
December 10, 2018 - A Tale of Two Labels
December 10, 2018 - Triple combination cancer immunotherapy improves outcomes in preclinical melanoma model
December 10, 2018 - A 14-year-old explains what it’s like to get a new heart
December 10, 2018 - Team Players Honored with 2018 Baton Awards
December 10, 2018 - Global report highlights how the changing world is affecting children’s physical activity levels
December 10, 2018 - Genes play a role in physical activity and sleep
December 10, 2018 - DDT in Alaskan fish shown to increase risk of cancer
December 10, 2018 - Laws to curb use of cell phones have greatly reduced fatalities for motorcyclists
December 10, 2018 - Argenx Provides Detailed Data from Phase 2 Clinical Trial of Efgartigimod in Immune Thrombocytopenia and Phase 1/2 Clinical Trial of Cusatuzumab in Acute Myeloid Leukemia
December 10, 2018 - University of Maryland doctors treat first breast cancer patients with GammaPod radiotherapy
December 10, 2018 - The heartbeat seat: Demoing new well-being technologies in a car
December 10, 2018 - Leading Cancer Researcher to Direct Herbert Irving Comprehensive Cancer Center
December 10, 2018 - Researchers explore how glial cells develop in the brain from neural precursor cells
December 10, 2018 - Study compares pain-related diagnoses in First Nations and non-First Nations children, youth
December 10, 2018 - Experts address sleep disorders following traumatic brain injury
December 10, 2018 - Scientists find answers to how cancer spreads
December 10, 2018 - Study explores why older people read more slowly
December 10, 2018 - Smart life-collar could save lives of young children
December 10, 2018 - Asbestos found in most NHS hospitals finds BBC inquiry
December 10, 2018 - Researchers use new technique to probe hydrogen bonds
December 10, 2018 - Music improves social communication in autistic children
December 10, 2018 - Some Brain Tumors May Respond to Immunotherapy, New Study Suggests
December 10, 2018 - Banning junk food ads to combat childhood obesity
December 10, 2018 - Skin Autofluorescence Predicts T2DM, Heart Disease, Mortality
December 10, 2018 - Largest autism sequencing study to date yields 102 genes associated with ASD
December 10, 2018 - Statins associated with low risk of side effects
December 10, 2018 - Episodic memory tests help in predicting brain atrophy and Alzheimer’s disease
December 10, 2018 - Study explores how schools address adolescent self-harming practices
December 10, 2018 - Pregnancy in adolescence linked to increased risks of complications in young mothers
December 10, 2018 - Risk Analysis publishes special issue on communicating about Zika virus
December 10, 2018 - Botox May Help Prevent Post-Op A-Fib
December 10, 2018 - African-American mothers rate boys higher for ADHD
December 10, 2018 - Graphic warning labels cancel out cigarettes’ appeal to young people
December 10, 2018 - Australian researchers to study gas inhalational anaesthetic and likelihood of cancer return
December 10, 2018 - Individual neurons located within the brain have implications for psychiatric diseases
December 10, 2018 - Researchers improve bariatric surgery scoring system to extend prediction time for diabetic remission
December 10, 2018 - HPV type 16 or 18 associated with cervical cancer risk in young women
December 10, 2018 - Cervical cancer risk is higher in women with positive HPV, but no cellular abnormalities
December 10, 2018 - Combo therapy not needed if low RA disease activity achieved
December 10, 2018 - Novel therapeutic targets based on biology of aging show promise for Alzheimer’s disease
December 10, 2018 - UC San Diego professor receives NCI Outstanding Investigator Award for cancer research
December 10, 2018 - Study evaluates placental mesenchymal stem cell sheets for myocardial repair and regeneration
December 10, 2018 - Blueprint Medicines Announces Updated Results from Ongoing EXPLORER Clinical Trial of Avapritinib Demonstrating Broad Clinical Activity and Significant Symptom Reductions in Patients with Systemic Mastocytosis
December 10, 2018 - Study clarifies ApoE4’s role in dementia
AI matched, outperformed radiologists in screening X-rays for certain diseases | News Center

AI matched, outperformed radiologists in screening X-rays for certain diseases | News Center

image_pdfDownload PDFimage_print

“Usually, we see AI algorithms that can detect a brain hemorrhage or a wrist fracture — a very narrow scope for single-use cases,” said Matthew Lungren, MD, MPH, assistant professor of radiology. “But here we’re talking about 14 different pathologies analyzed simultaneously, and it’s all through one algorithm.”

The goal, Lungren said, is to eventually leverage these algorithms to reliably and quickly scan a wide range of image-based medical exams for signs of disease without the backup of professional radiologists. And while that may sound disconcerting, the technology could eventually serve as high-quality digital “consultations” to resource-deprived regions of the world that wouldn’t otherwise have access to a radiologist’s expertise. Likewise, there’s an important role for AI in fully developed health care systems too, Lungren added. Algorithms like CheXNeXt could one day expedite care, empowering primary care doctors to make informed decisions about X-ray diagnostics faster, without having to wait for a radiologist.

“We’re seeking opportunities to get our algorithm trained and validated in a variety of settings to explore both its strengths and blind spots,” said graduate student Pranav Rajpurkar. “The algorithm has evaluated over 100,000 X-rays so far, but now we want to know how well it would do if we showed it a million X-rays — and not just from one hospital, but from hospitals around the world.”

A paper detailing the findings of the study was published online Nov. 20 in PLOS Medicine. Lungren and Andrew Ng, PhD, adjunct professor of computer science at Stanford, share senior authorship. Rajpurkar and fellow graduate student Jeremy Irvin are the lead authors.

Practice makes perfect

Lungren and Ng’s diagnostic algorithm has been in development for more than a year. It builds on their work on a previous iteration of the technology that could outperform radiologists when diagnosing pneumonia from a chest X-ray. Now, they’ve boosted the abilities of the algorithm to flag 14 ailments, including masses, enlarged hearts and collapsed lungs. For 11 of the 14 pathologies, the algorithm made diagnoses with the accuracy of radiologists or better.

Back in the summer of 2017, the National Institutes of Health released a set of hundreds of thousands of X-rays. Since then, there’s been a mad dash for computer scientists and radiologists working in artificial intelligence to deliver the best possible algorithm for chest X-ray diagnostics.

We need to be thinking about how far we can push these AI models to improve the lives of patients anywhere in the world.

The scientists used about 112,000 X-rays to train the algorithm. A panel of three radiologists then reviewed a different set of 420 X-rays, one by one, for the 14 pathologies. Their conclusions served as a “ground truth”— a diagnosis that experts agree is the most accurate assessment — for each scan. This set would eventually be used to test how well the algorithm had learned the telltale signs of disease in an X-ray. It also allowed the team of researchers to see how well the algorithm performed compared to the radiologists.

“We treated the algorithm like it was a student; the NIH data set was the material we used to teach the student, and the 420 images were like the final exam,” Lungren said. To further evaluate the performance of the algorithm compared with human experts, the scientists asked an additional nine radiologists from multiple institutions to also take the same “final exam.”

“That’s another factor that elevates this research,” Lungren said. “We weren’t just comparing this against other algorithms out there; we were comparing this model against practicing radiologists.”

What’s more, to read all 420 X-rays, the radiologists took about three hours on average, while the algorithm scanned and diagnosed all pathologies in about 90 seconds.

Next stop: the clinic

Now, Lungren said, his team is working on a subsequent version of CheXNeXt that will bring the researchers even closer to in-clinic testing. The algorithm isn’t ready for that just yet, but Lungren hopes that it will eventually help expedite the X-ray-reading process for doctors diagnosing urgent care or emergency patients who come in with a cough.

“I could see this working in a few ways. The algorithm could triage the X-rays, sorting them into prioritized categories for doctors to review, like normal, abnormal or emergent,” Lungren said. Or the algorithm could sit bedside with primary care doctors for on-demand consultation, he said. In this case, Lungren said, the algorithm could step in to help confirm or cast doubt on a diagnosis. For example, if a patient’s physical exam and lab results were consistent with pneumonia, and the algorithm diagnosed pneumonia on the patient’s X-ray, then that’s a pretty high-confidence diagnosis and the physician could provide care right away for the condition. Importantly, in this scenario, there would be no need to wait for a radiologist. But if the algorithm came up with a different diagnosis, the primary care doctor could take a closer look at the X-ray or consult with a radiologist to make the final call.

“We should be building AI algorithms to be as good or better than the gold standard of human, expert physicians. Now, I’m not expecting AI to replace radiologists any time soon, but we are not truly pushing the limits of this technology if we’re just aiming to enhance existing radiologist workflows,” Lungren said. “Instead, we need to be thinking about how far we can push these AI models to improve the lives of patients anywhere in the world.”

Other Stanford authors of the study are biostatistician Robyn Ball, PhD; undergraduate student Kaylie Zhu; former research assistant Brandon Yang; data scientist Hershel Mehta; research assistants Tony Duan and Daisy Ding; former research assistant Aarti Bagul; professor of radiology and of medicine Curtis Langlotz, PhD; assistant professor of radiology Bhavik Patel, MD; associate professor of radiology Kristen Yeom, MD; research associate Katie Shpanskaya; associate professor of radiology Francis Blackenberg, MD; clinical assistant professor of radiology Jayne Seekins, MD; clinical associate professor of radiology Safwan Halabi, MD; and clinical assistant professor of radiology Evan Zucker, MD.

Researchers from Duke University and from the University of Colorado also contributed to the study.

Lungren is a member of Stanford Bio-X, the Stanford Child Health Research Institute and the Stanford Cancer Institute.

Stanford’s departments of Radiology and of Computer Science along with the Stanford Center for Artificial Intelligence in Medicine & Imaging supported the work.

Tagged with:

About author

Related Articles