Home›Companies›Mecha Health
Mecha Health

Foundation models to automate x-ray analysis for radiologists

Mecha Health builds foundation models to automate x-ray analysis for radiologists. We take medical images and process them using proprietary models to produce accurate draft medical reports. Our first model was built in less than two months, and beat Microsoft, Google, and OpenAI on clinical accuracy metrics. On top of that, it’s two orders of magnitude smaller and trained with a quarter of the data. We are partnering with the largest privately owned radiology practice in the US and a multinational tele-radiology company to provide them with their own foundation model, enabling their radiologists to go from reading 1 scan per hour to 1 scan every 5 minutes. By charging on a per scan basis, x-ray report generation represents a 40B+ market opportunity.
Active Founders
Ahmed Abdulaal
Ahmed Abdulaal
Founder/CEO
Medical doctor @Imperial, where I gained entry to the most competitive deanery in the country. Microsoft PhD scholar in machine learning @UCL - I have more than 20 peer reviewed publications in medicine, digital healthcare, and machine learning - including first author publications in top AI venues such as ICLR. 360+ citations on google scholar. Ex Research Scientist at the Center for Artificial Intelligence @AstraZeneca. Building the future of automated radiology @Mecha Health.
Hugo Fry
Hugo Fry
Founder
Mathematics (BA) and Physics (MSc) graduate from the University of Cambridge, ranking top 30 in the university. British Physics Olympiad participant, ranking in the top 11 in the UK. Former ML interpretability research scholar (MATS), where I was the first in the world to use Sparse Autoencoders on vision models. My research has been cited by Anthropic and Google DeepMind. Currently working on building state of the art radiology report generation AI in a way that is interpretable and safe.
I am a Machine Learning & Medical Imaging PhD candidate from University College London (UCL). During my PhD I published work at tier-one Machine Learning venues (NeurIPS, ICML & ICLR), and was invited to present my work at international medical imaging conferences, Digitas, & Harvard Medical School. Before this startup I was poached by my Professors from UCL including the Head of Computer Science to be CEO for their spin-out. Currently building Radiology 3.0 by Automating reporting.
Solving the interpretive moment in radiology. ex-ML Engineer across multiple medical device startups, with experience in FDA and CE marked medical devices, as well as LLM safety in healthcare. My PhD in Medical Imaging (UCL) focussed on surgical vision; my work has been patented, cited 100+ times, and published across the world's highest impact scientific venues (Nature, NeurIPS, ICLR).
Company Launches
Mecha Wayfinder: locating any finding, anywhere, in any medical image.
See original launch post

Mecha Wayfinder

One of the most important challenges for next generation foundation models in radiology is the ability to locate any type of pathology in across anatomies and modalities.

We’ve recently had a bit of research breakthrough on this, and today we’re introducing Mecha Wayfinder, a new tool for locating findings in medical images.

Traditional systems like segmentation models are too restrictive for the task we’re trying to solve. Grad-CAM weights and upsamples a late convolutional feature map, which can leave localization spatially coarse and obscure boundaries.

Wayfinder takes a different approach. Think of giving the underlying AI an “MRI”: a way to inspect where the model looks, requiring zero human annotations.

The part we’re so excited about is that this approach scales with only images and reports. As we train better underlying systems, we expect Wayfinder to improve too.

You can learn all about it in our latest blog post linked below:
https://www.mecha-health.ai/blog/Locating-Any-Pathology-Anywhere-in-All-Types-of-Medical-Images

https://youtu.be/CDkv4uWWpjw

Previous Launches
Today Mecha is launching a new set of capabilities, vision-language foundation models that interpret studies the way radiologists do.
Automated analysis of medical images that is accurate and interpretable.
Mecha Health
Founded:2024
Batch:Winter 2025
Team Size:4
Status:
Active
Location:San Francisco