(This is an excerpt of the Health Rounds newsletter, where we present latest medical studies on Tuesdays and Thursdays.)
By Nancy Lapid
March 25 (Reuters) – Fake X-ray images created by artificial intelligence to resemble true results from human patients can fool not only experienced radiologists but also the AI tools themselves, according to a study that illustrates the potential for manipulation by bad actors.
Seventeen radiologists from 12 hospitals in six countries reviewed 264 X-ray images, half of which had been generated by the AI tools ChatGPT or RoentGen.
When radiologist readers were unaware of the study’s true purpose, only 41% spontaneously identified AI-generated images, according to a report published in Radiology.
After being informed that the dataset contained synthetic images, the radiologists’ mean accuracy in differentiating the real and synthetic X-rays rose to 75%.
Having deepfake X-rays realistic enough to deceive radiologists “creates a high-stakes vulnerability for fraudulent litigation if, for example, a fabricated fracture could be indistinguishable from a real one,” study leader Dr. Mickael Tordjman of the Icahn School of Medicine at Mount Sinai in New York said in a statement.
“There is also a significant cybersecurity risk if hackers were to gain access to a hospital’s network and inject synthetic images to manipulate patient diagnoses or cause widespread clinical chaos by undermining the fundamental reliability of the digital medical record,” Tordjman said.
The accuracy of four large language models – GPT-4o (OpenAI), GPT-5 (OpenAI), Gemini 2.5 Pro (Google), and Llama 4 Maverick (Meta Platforms) – at detecting the fake images ranged from 57% to 85%.
Even ChatGPT-4o, the model that created the deepfakes, failed to detect all of them, though it identified more than the other LLMs, the researchers reported.
Potential digital safeguards are needed to help distinguish real and fake images and prevent tampering such as use of invisible watermarks that embed ownership, researchers said.
“We are potentially only seeing the tip of the iceberg,” said Tordjman of the eventual possibility of fake CT and MRI scans. “Establishing educational datasets and detection tools now is critical.”
BIOMARKER IMPROVES DIAGNOSIS OF LEWY BODY DEMENTIA
Researchers have discovered a biomarker in cerebrospinal fluid that should help improve diagnosis of Parkinson’s disease and dementia with Lewy bodies and distinguish that from other forms of dementia.
An enzyme called DOPA decarboxylase, which plays a crucial role in the production of dopamine in the brain, is present in cerebrospinal fluid at significantly higher concentrations in patients with Parkinson’s disease and Lewy body dementia, the researchers found.
This difference is clearly measurable compared with patients suffering from more common Alzheimer’s disease, making the test highly specific, the researchers said.
“The importance of this discovery for clinical practice is considerable, as dementia with Lewy bodies is often difficult to diagnose correctly at present,” study leader Dr. Sebastiaan Engelborghs of Vrije Universiteit Brussel said in a statement.
“Because of the strong overlap of symptoms with other forms of dementia, patients are regularly misdiagnosed,” he added. “The new measurement method provides doctors with an objective tool for determining the right course of action at an early stage.”
The researchers have developed two highly sensitive but still experimental laboratory tests for DOPA decarboxylase. The results of their testing correlated directly with the degree of pathological changes in autopsy samples from patients’ brains, they reported in Nature Medicine.
“This publication brings a crucial biomarker closer to the patient, precisely in cases where diagnosis is still too often associated with uncertainty,” Engelborghs said.
WEEDKILLER CONTRIBUTES TO ANTIBIOTIC RESISTANCE IN HOSPITALS
Drug-resistant bacteria from hospitals can thrive in soil treated with glyphosate, and bacteria carrying antibiotic-resistance genes can spread from soil treated with the widely used weedkiller to hospitals, researchers in Argentina found.
In 2018 and 2020, the researchers collected 68 bacterial strains from sediments in the Paraná delta wetlands of Argentina north of Buenos Aires, near where glyphosate is often used in agricultural areas. They also collected 15 strains isolated from feedlots and herbicide-impacted agricultural soils in the region.
They tested each strain’s degree of resistance to 16 common antibiotics as well as to glyphosate and glyphosate-based herbicides. They then compared the results with those from 19 multidrug-resistant strains from local hospitals.
All of the hospital strains were highly resistant to glyphosate and glyphosate-based weedkillers, researchers reported in Frontiers in Microbiology. Most of the glyphosate-resistant strains found in the environment were genetically related to the multidrug-resistant strains from the hospitals, they found.
“Our findings indicate that glyphosate exposure could favor the prevalence of bacteria associated with (hospital-acquired) infections and the rise of multidrug-resistant clinical strains,” the researchers wrote.
Pesticide labels should include a warning that genes for antibiotic resistance can spread from glyphosate-contaminated soils to hospitals, study leader Dr Daniela Centrón of the Institute of Medical Microbiology and Parasitology in Buenos Aires said in a statement.
(To receive the full newsletter in your inbox for free sign up here)
(Reporting by Nancy Lapid; Editing by Bill Berkrot)
Comments