🩻 Teaching machines to see disease: AI in diagnostic radiology

In partnership with

Welcome back to Healthy Innovations! 👋

Healthy Innovations is the newsletter for forward-looking clinicians and healthcare business leaders who want to get to grips with the latest advances in this fast-paced industry. From AI-powered diagnostics to revolutionary gene therapies, I will highlight the fascinating breakthroughs reshaping healthcare and what this means for you, your business and the wider community.

In this issue of Healthy Innovations, we are deep diving into the world of clinical radiology and the transformational impact that AI and generative AI have on the field.

When I started exploring AI in healthcare a few years ago, diagnostic radiology stood out as one of the most promising applications. Among all medical specialties, AI-aided tools seemed perfectly suited to this demanding field.

Think about it: radiologists analyze hundreds of medical images daily, making critical decisions that directly impact patient outcomes. The sheer volume of data, combined with increasingly complex imaging techniques, creates an enormous cognitive load. It's precisely this combination of high-stakes decision-making and data intensity that makes radiology such a perfect testbed for AI integration.

Foundational AI applications in radiology

AI in radiology is more than image enhancement.

These systems have become invaluable allies in the diagnostic process, showing remarkable capabilities in detecting early-stage cancers, subtle vertebral fractures, and even pathological changes before they become clinically apparent.

Take Medtronic's GI Genius, for example - it's changing how doctors perform colonoscopies. Studies from Northwestern University show that physicians using this AI-assisted tool find significantly more polyps than traditional methods alone. According to a study in Gastroenterology, these AI tools have slashed polyp miss rates by 50%, with over 3 million patients worldwide now benefiting from AI-enhanced colonoscopies annually.

More than improving detection

The impact of AI in radiology extends beyond improving detection rates. It can also improve safety and tackle health disparities.

Deepmeds, a Sydney-based startup, has developed AI-based MRI technology that reduces potentially toxic gadolinium dye use in brain scans by up to 80%.

Meanwhile, Thermalytix, a radiation-free, AI-based mammography tool deployed across 200 hospitals in India, is making critical diagnostic capabilities accessible to underserved communities through its affordability and portability.

GenAI takes it further

Today's advanced language models don't just analyze images - they create standardized radiology reports, translate complex findings into patient-friendly language, and predict prognoses. Harvard's Clinical Histopathology Imaging Evaluation Foundation (CHIEF) platform showcases this progress, achieving 94% accuracy in early-stage tumor detection and accurately predicting patient survival based on initial diagnostic imaging.

Multi-modal opportunities

The latest innovation in AI-assisted radiology is the ability to analyze multiple types of medical images simultaneously. One study found that combining PET and MRI scans with AI analysis revealed previously undetectable biomarkers for early-stage Alzheimer's, identifying at-risk patients years before traditional methods.

However, technical hurdles remain. These challenges stem from the variety of imaging equipment used across healthcare facilities, each potentially producing data in different formats. Startups like Neuroharmony are addressing these standardization challenges by harmonizing volume measurements from different MRI scanners to improve diagnostic consistency.

Building appropriate trust

Let's talk about trust - it's critical when implementing AI technologies in clinical settings. Both healthcare providers and patients need to have confidence in AI systems while maintaining appropriate skepticism.

A recent study in Radiology found that radiologists sometimes rely too heavily on AI when it specifically calls out an area of interest in an X-ray. As the researchers noted, we need to "be aware of these pitfalls and stay mindful of our diagnostic patterns and training."

Future directions

The next frontier in medical AI is moving away from complex black boxes toward more transparent systems seamlessly integrated into clinical workflows. For healthcare facilities considering AI implementation, key focus areas include comprehensive staff training, robust validation processes, clear workflow integration protocols, and continuous monitoring of clinical outcomes.

The future of radiology lies in the powerful partnership between human expertise and AI - each enhancing the other's capabilities. While AI's role will continue to grow, it's here to amplify, not replace, radiologists' critical thinking and clinical judgment.

Innovation highlights

🦴 Bone healing breakthrough: Scientists have discovered that combining intermittent fasting with a specialized protein-delivering bandage can restore bone healing abilities in older mice to match those of younger animals. The treatment works by rejuvenating special bone-forming cells through improved cellular energy production and beneficial gut bacteria changes. While still in the early stages, this dual approach could change how we treat fractures and bone injuries in older adults, potentially extending to other aging tissues as well.

🔄 Molecular jackhammers vs cancer: Scientists have engineered a cancer treatment using vibrating molecules that act like microscopic jackhammers. When stimulated with near-infrared light, these aminocyanine molecules vibrate in sync to physically break apart cancer cell membranes, destroying 99% of cancer cells in lab tests. Unlike traditional treatments, this mechanical approach could potentially reach deep-seated tumors without surgery, and early trials in mice with melanoma showed promising results with half the test subjects becoming cancer-free.

🐕 Anti-aging for our furry friends: Scientists are on the verge of a breakthrough in extending dogs' lifespans through two promising approaches. A beef-flavored pill aims to give dogs an extra year of healthy life by targeting age-related metabolic changes, while the Dog Aging Project is testing rapamycin to potentially add three healthy years to canine lives. What makes this especially exciting is that dogs share our living environments and develop similar age-related diseases, making them perfect test subjects for treatments that could eventually help humans live longer, healthier lives too.

🫀 AI spots hidden heart risks: UK doctors have developed an algorithm that could prevent thousands of strokes by scanning millions of GP records for signs of undiagnosed atrial fibrillation. Using machine learning to analyze factors like age, medical history, and ethnicity, the tool identifies high-risk patients who are then given portable ECG devices for monitoring. Already showing success in early trials, this AI-powered approach could transform early detection of this dangerous heart condition that contributes to 20,000 strokes annually in the UK.

Cool tool

🌐 Lovable is an AI-powered platform that turns your words into working web apps. Simply describe what you want to build in plain English, and watch as it generates a fully functional application complete with layouts, components, and even a database backend. Perfect for quickly testing business ideas or learning web development by observing AI-generated code. While the platform offers a free tier for basic projects, you'll need a paid subscription to access more advanced features.

Company to watch

🤖 Belong.Life is innovating digital health support through AI-powered social platforms designed for specific patient communities. Their flagship app, "Belong - Beating Cancer Together," has grown into the world's largest social network for cancer patients, while sister platforms serve those with MS, IBD, and other chronic conditions. What sets them apart is their suite of AI health mentors - Dave, Sophie, Tara, and Fred - who provide 24/7 personalized guidance validated by medical professionals. Serving over a million patients across 100+ countries, the company is not just building apps but creating global communities that combine peer support with cutting-edge AI assistance, while gathering valuable real-world data to advance medical research.

Weird and wonderful

💀 Dead memories? Scientists are exploring a fascinating question: could we ever retrieve memories from someone after they've died? While we can now identify where memories physically exist in the brain as 'engrams', actually extracting these memories remains in the realm of science fiction. Unlike a computer's hard drive, our memories aren't simple recordings but complex, dynamic networks spread across different brain regions. While full memory retrieval may be impossible, researchers suggest that with enough brain scans of someone repeatedly remembering events throughout their life, we might one day be able to reconstruct fragments of their memories - though the technology remains far in the future.

Start learning AI in 2025

Everyone talks about AI, but no one has the time to learn it. So, we found the easiest way to learn AI in as little time as possible: The Rundown AI.

It's a free AI newsletter that keeps you up-to-date on the latest AI news, and teaches you how to apply it in just 5 minutes a day.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses – tailored to your needs.

Sign up to start learning.

Thank you for reading the Healthy Innovations newsletter!

Keep an eye out for next week’s issue, where I will highlight the healthcare innovations you need to know about.

Have a great week! Alison ✨

P.S. If you enjoyed reading the Healthy Innovations newsletter, please subscribe so I know the content is valuable to you!

P.P.S. Check out beehiiv (affiliate link) if one of your 2025 resolutions is to start your very own newsletter.