(Press-News.org) A new artificial intelligence (AI) tool could make it much easier—and cheaper—for doctors and researchers to train medical imaging software, even when only a small number of patient scans are available.
The AI tool improves upon a process called medical image segmentation, where every pixel in an image is labeled based on what it represents—cancerous or normal tissue, for example. This process is often performed by a highly trained expert, and deep learning has shown promise in automating this labor-intensive task.
The big challenge is that deep learning-based methods are data hungry—they require a large amount of pixel-by-pixel annotated images to learn, explained Li Zhang, a Ph.D. student in the Department of Electrical and Computer Engineering at the University of California San Diego. Creating such datasets demands expert labor, time and cost. And for many medical conditions and clinical settings, that level of data simply doesn’t exist.
To overcome this limitation, Zhang and a team of researchers led by UC San Diego electrical and computer engineering professor Pengtao Xie have developed an AI tool that can learn image segmentation from just a small number of expert-labeled samples. By doing so, it cuts down the amount of data usually required by up to 20 times. It could potentially lead to faster, more affordable diagnostic tools, especially in hospitals and clinics with limited resources.
The work was published in Nature Communications.
“This project was born from the need to break this bottleneck and make powerful segmentation tools more practical and accessible, especially for scenarios where data are scarce,” said Zhang, who is the first author of the study.
The AI tool was tested on a variety of medical image segmentation tasks. It learned to identify skin lesions in dermoscopy images; breast cancer in ultrasound scans; placental vessels in fetoscopic images; polyps in colonoscopy images; and foot ulcers in standard camera photos, just to list several examples. The method was also extended to 3D images, such as those used to map the hippocampus or liver.
In settings where annotated data were extremely limited, the AI tool boosted model performance by 10 to 20% compared to existing approaches. It required 8 to 20 times less real-world training data than standard methods while often matching or outperforming them.
Zhang described how this AI tool could potentially be used to help dermatologists diagnose skin cancer. Instead of gathering and labeling thousands of images, a trained expert in the clinic might only need to annotate 40, for example. The AI tool could then use this small dataset to identify suspicious lesions from a patient’s dermoscopy images in real time. “It could help doctors make a faster, more accurate diagnosis,” Zhang said.
The system works in stages. First, it learns how to generate synthetic images from segmentation masks, which are essentially color-coded overlays that tell an algorithm which parts of an image are, say, healthy or diseased. Then, it uses that knowledge to create new, artificial image-mask pairs to augment a small dataset of real examples. A segmentation model is trained using both. Through a continuous feedback loop, the system refines the images it creates based on how well they improve the model’s learning.
The feedback loop is a big part of what makes this system work so well, noted Zhang. “Rather than treating data generation and segmentation model training as two separate tasks, this system is the first to integrate them together. The segmentation performance itself guides the data generation process. This ensures that the synthetic data are not just realistic, but also specifically tailored to improve the model’s segmentation capabilities.”
Looking ahead, the team plans to make their AI tool smarter and more versatile. The researchers also plan to incorporate feedback from clinicians directly into the training process to make the generated data more relevant for real-world medical use.
Full study: “Generative AI enables medical image segmentation in ultra low-data regimes”
This work was supported by the National Science Foundation (IIS2405974 and IIS2339216) and the National Institutes of Health (R35GM157217 and R21GM154171).
END
New AI tool learns to read medical images with far less data
2025-08-01
ELSE PRESS RELEASES FROM THIS DATE:
Announcing XPRIZE Healthspan as Tier 5 Sponsor of ARDD 2025
2025-08-01
The University of Copenhagen is excited to announce XPRIZE Healthspan as Tier 5 Sponsor of the 12th Aging Research & Drug Discovery Meeting, the world's largest conference on aging research in the biopharmaceutical industry that will transpire on August 25 - August 29, 2025 on-site at the Ceremonial Hall, University of Copenhagen, and online.
ARDD has grown to become the largest and most important conference in longevity biotechnology. Each year, ARDD brings together a unique mix of academic luminaries, biotech innovators, investors, top pharmaceutical executives, physicians, and related health care professionals to discuss ...
Announcing Immortal Dragons as Tier 4 Sponsor of ARDD 2025
2025-08-01
The University of Copenhagen is excited to announce Immortal Dragons as Tier 4 Sponsor of the 12th Aging Research & Drug Discovery Meeting, the world's largest conference on aging research in the biopharmaceutical industry that will transpire on August 25 - August 29, 2025 on-site at the Ceremonial Hall, University of Copenhagen, and online.
Immortal Dragons is a purpose-driven longevity fund headquartered in Singapore, supporting 15+ portfolio companies in longevity and radical life extension technologies. Immortal Dragons values impact rather than economic return, supporting moonshot longevity startups in radical life extension, key areas include: Wholebody replacement, ...
Reporting guideline for chatbot health advice studies
2025-08-01
About The Article: The rise in chatbot health advice studies is accompanied by heterogeneity in reporting standards, impacting their interpretability. This article provides reporting recommendations for studies evaluating the performance of generative artificial intelligence (AI)–driven chatbots when summarizing clinical evidence and providing health advice. This article is being published jointly in Artificial Intelligence in Medicine, Annals of Family Medicine, BJS, BMC Medicine, BMJ Medicine, JAMA Network Open, The Lancet, NEJM-AI, and Surgical Endoscopy.
Corresponding Author: To contact the corresponding ...
Announcing Mitra Bio as Tier 3 Sponsor of ARDD 2025
2025-08-01
The University of Copenhagen is excited to announce Mitra Bio as Tier 3 Sponsor of the 12th Aging Research & Drug Discovery Meeting, the world's largest conference on aging research in the biopharmaceutical industry that will transpire on August 25 - August 29, 2025 on-site at the Ceremonial Hall, University of Copenhagen, and online.
“ARDD sits at the intersection of frontier longevity science and real-world impact—exactly where Mitra Bio wants to be,” said Shakiba Kaveh, PhD, co-founder & CEO of Mitra Bio. “By sponsoring this year’s meeting, we hope to support the longevity science community and share our insights ...
Study identifies global upswing in photosynthesis driven by land, offset by oceans
2025-08-01
Terrestrial plants drove an increase in global photosynthesis between 2003 and 2021, a trend partially offset by a weak decline in photosynthesis — the process of using sunlight to make food — among marine algae, according to a new study published in Nature Climate Change on Aug. 1. The findings could inform planetary health assessments, enhance ecosystem management, and guide climate change projections and mitigation strategies.
Photosynthetic organisms — also known as primary producers — form the base of the food chain, making most life on Earth possible. Using ...
Study reports final clinical trial data for advanced kidney cancer treatment
2025-08-01
A two-drug combination for treating advanced kidney cancer had sustained and durable clinical benefit in more than five years of follow-up, according to a study published Aug. 1 in Nature Medicine.
The study reports final clinical data and biomarker analyses from the Phase 3 KEYNOTE-426 trial, which compared the drug combination pembrolizumab plus axitinib versus the single drug sunitinib for patients with previously untreated advanced clear cell renal cell carcinoma, the most common type of kidney cancer.
“KEYNOTE-426 was the first trial to combine a PD-1 ...
Antibiotic resistant bacteria found in malnourished children under five years old
2025-08-01
A new study led by researchers at the Ineos Oxford Institute for antimicrobial research (IOI) has found that antimicrobial resistant bacteria is spreading rapidly among children being treated for severe malnutrition in a hospital facility in Niger. The findings have been published today (1 August) in Nature Communications
Globally 45 million children under the age of five are estimated to be severely malnourished. These children are also at a higher risk of developing life-threatening infections such as tuberculosis or sepsis due to their weakened immune systems.
Working with Médecins Sans Frontières (Doctors Without Borders), researchers analysed over 3,000 ...
Study: Most US homes can save money and affordably weather blackouts with solar plus storage
2025-08-01
Most U.S. households could reduce their electricity costs and endure power outages by installing rooftop solar panels and battery packs, according to a new Stanford University study, though people may need to buy the equipment by Dec. 31.
About 60% of families could reduce their electricity costs by 15% on average by installing a solar-battery system. That’s after accounting for annualized capital and operating costs of the equipment. Some 63% of U.S households could also weather local or regional blackouts with such systems, able to meet about half their electricity needs on average. These households would either save money on electricity or ...
The human touch of doctors will still be needed in the AI healthcare revolution, technology expert suggests
2025-08-01
AI-based medicine will revolutionise care including for Alzheimer’s and diabetes, predicts a technology expert, but it must be accessible to all patients.
Healing with Artificial Intelligence, written by technology expert Daniele Caligiore, uses the latest science research to highlight key innovations assisted by AI such as diagnostic imaging and surgical robots.
From exoskeletons that help spinal injury patients walk to algorithms that can predict the onset of dementia years in advance, Caligiore explores what he describes as a ‘revolution’ ...
Helping me, inhibiting you: Analysis of interactions between intestinal microbiota
2025-08-01
Intestinal bacteria are important for human health as they help digest food and regulate immune function. Thus, in recent years, it has become clear that gut microbiota dysbiosis, an imbalance in types and composition of intestinal bacteria, is associated with digestive disorders and other diseases. However, the mechanism by which bacteria interact in the intestine to maintain the microbiota balance remains unclear.
A research group led by Associate Professor Koji Hosomi at Osaka Metropolitan University’s Graduate School of Veterinary Science studied two bacteria: Fusobacterium varium (F. varium), an oral and intestinal bacterium linked to inflammation and colon ...