PHENOMICL: detecting bleeding in atherosclerotic plaques

Atherosclerosis Unravelled by AI: More Accurate Research with New Insights 

Imagine limescale buildup in a water pipe. It narrows the passage and makes the pipe vulnerable. Just as limescale can reduce water pressure and cause leaks, deposits in your blood vessels can lead to serious problems. These deposits, known as plaques, are essentially fatty buildups that form inside the arteries. They narrow the passage for blood and can weaken the vessel walls. One particularly risky feature of these plaques is intraplaque haemorrhage (IPH), bleeding within the plaque itself. These haemorrhages are a major indicator of an increased risk of severe cardiovascular problems. Until now, detecting and measuring these haemorrhages has been a challenge because traditional methods are time-consuming and subjective. 

 

An Innovative Approach: AI as a Research Tool 

To tackle this issue, researchers from UMC Utrecht, the University of Virginia, and Human Technopole have developed an advanced AI system, called PHENOMICL, that can automatically detect and measure these haemorrhages. This system analyses images of plaques from the Athero-Express biobank, a comprehensive dataset of 2,595 patients, providing us with valuable insights. It enables us to more accurately predict who is at a higher risk of future cardiovascular problems. The AI thus serves as a powerful tool for identifying risks and may contribute to the development of better diagnostic methods in the future. 

 

Key Findings: Accuracy and New Insights 

Our research shows that AI is not only faster but also significantly more accurate than traditional manual methods in detecting intraplaque haemorrhage. By analysing stained tissue sections and using advanced techniques that visualize different cell types and molecular processes, we have mapped out the processes behind these haemorrhages in greater detail. We discovered that specific inflammatory responses and changes in the structural tissue of plaques play a crucial role. These insights are invaluable because they help us understand how these haemorrhages contribute to plaque instability and, ultimately, to severe cardiovascular problems. 

 

The Impact of This Research: Improved Risk Assessment and Future Treatments 

This research marks a significant step forward in the fight against cardiovascular diseases and represents the largest digital pathology analysis of intraplaque haemorrhage to date. By leveraging AI, we can analyse plaques not only faster but also with unprecedented accuracy. By integrating histological findings with clinical data and transcriptomic data, we have demonstrated that IPH is a critical factor in plaque vulnerability, with important implications for risk assessment and personalized cardiovascular care. We have not only developed an accurate method for quantifying IPH but also mapped the molecular basis of plaque vulnerability. This research lays the foundation for future studies that combine spatial transcriptomics with histological and clinical data to further understand the biology of plaques. 

 

Curious to Learn More? Dive Into Our Research! 

Are you interested in the details of our study? Read our full article at https://doi.org/10.1101/2025.03.04.25323316. And as always, our code and data are #FAIR available: 

 We would also like to thank our dear colleagues from the UMC Utrecht, but especially from the University of Virginia (Clint L. Miller and Yipei Song), and Human Technopole (Craig Glastonbury and Francesco Cisternino) for their invaluable contributions to and the great collaboration during this project. Together, we strive for healthier blood vessels for everyone! 

PHENOMICL

Detecting bleeding in atherosclerotic plaques.

Tim S. Peters

Tim Peters is a PhD student working at vdLaan & Science.

Next
Next

dr. Joost Mekke