← Latest papers
📄 medicine

Digital quantification of vascular β-amyloid and perivascular plaques in cerebral amyloid angiopathy

This study demonstrates that digital pathology classifiers can reliably automate the quantification of vascular β-amyloid and distinguish it from plaques in cerebral amyloid angiopathy, revealing a strong correlation with manual scores and a specific association between higher CAA severity and increased perivascular plaque proximity.

Original authors: Paniz Mojdeganlou, Jennifer Poirier, Lisa L Barnes, David A Bennett, Sue E Leurgans, Julie A Schneider, Alifiya Kapasi

Published 2026-08-13
📖 6 min read🧠 Deep dive

Original authors: Paniz Mojdeganlou, Jennifer Poirier, Lisa L Barnes, David A Bennett, Sue E Leurgans, Julie A Schneider, Alifiya Kapasi

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the human brain as a bustling, high-tech city. In this city, the streets are lined with tiny pipes called blood vessels, which deliver oxygen and nutrients to keep the lights on and the traffic moving. But sometimes, a sticky, gummy substance called beta-amyloid starts to leak out of the pipes and clog them up. This condition is known as Cerebral Amyloid Angiopathy, or CAA for short. It's like a slow-motion traffic jam where the pipes get stiff and leaky, and the sticky goo also piles up on the sidewalks (the brain tissue) next to the pipes.

For a long time, doctors and scientists have tried to measure how bad this clogging is. They usually look at brain slices under a microscope and give a "grade" from zero to four, kind of like rating a movie from "boring" to "masterpiece." But this manual grading is a bit like trying to count every single grain of sand on a beach by hand—it's slow, and two different people might give the same beach a slightly different rating. Plus, it doesn't tell us exactly how the sticky goo on the sidewalks relates to the clogged pipes. To solve this, a team of researchers decided to teach computers to do the counting. They used a super-smart digital eye to scan brain tissue and automatically find the clogged pipes and the sticky piles, hoping to get a more precise picture of what's happening in the aging brain.


The Digital Detective: Counting the Clogs

In this study, researchers from Rush University Medical Center built a digital detective using a platform called HALO AI. Their goal was to automate the job of spotting two specific things in the brain's "calcarine cortex" (a specific neighborhood in the back of the brain): the blood vessels clogged with beta-amyloid (CAA-positive vessels) and the separate piles of beta-amyloid sitting in the brain tissue (plaques).

Think of the manual grading system as a human looking at a map and saying, "Hmm, there's a lot of traffic here, I'll give it a 3 out of 4." The new digital system, however, is like a drone flying over the city, counting every single car, measuring the exact width of the traffic jam, and calculating the percentage of the road covered in goo. The researchers trained their AI on 100 brain slides that had already been graded by human experts. They taught the computer to distinguish between a vessel that was clogged and a plaque that was just sitting nearby, ensuring it didn't get confused by the background noise.

What the Numbers Say

The results were impressive. When the computer's measurements were compared to the human experts' grades, they matched up incredibly well.

  • The computer's count of the "CAA classified area" (how much space the clogged vessels took up) had a correlation of 0.96 with the human scores. That's almost a perfect match.
  • The computer's measurement of "vascular beta-amyloid percent positivity" (how much of the vessel wall was covered in sticky goo) had a correlation of 0.79.

The study looked at 100 participants who passed away at an average age of 91.1 years (with 76% being women). The computer found that the number of clogged vessels varied wildly depending on the human score:

  • In a "Score 1" case (mild), the computer found a median of 3 clogged vessels.
  • In a "Score 2" case, it found 12.
  • In a "Score 3" case, it jumped to 153.
  • In a "Score 4" case (severe), it found a median of 399 clogged vessels, with some slides having as many as 1,017!

This shows that while humans might just say "Score 4," the digital eye can see that one "Score 4" patient might have twice as many clogged vessels as another, revealing details that the old grading system missed.

The "Sidewalk" Mystery: Where the Piles Are

Here is where the story gets really interesting. The researchers wanted to know: Do the piles of sticky goo (plaques) on the sidewalks cluster right next to the clogged pipes, or are they scattered randomly?

They defined "perivascular plaques" as any pile of goo within 150 microns (a tiny distance, about the width of a few human hairs) of a clogged vessel. They divided this zone into three rings: 0–50 microns, 50–100 microns, and 100–150 microns.

The findings were surprising:

  1. Total Amount: The total amount of sticky goo in the whole brain neighborhood didn't change much between patients with mild CAA and those with severe CAA.
  2. The Clustering: However, the location of the goo changed dramatically. In patients with higher CAA scores (more clogged pipes), there was a massive spike in the number of goo piles sitting right next to the pipes.
    • In the innermost ring (0–50 microns), the density of plaques jumped from almost nothing in low-score cases to 87.22 plaques per unit area in high-score cases.
    • In the middle ring (50–100 microns), it went from 0.85 to 54.75.
    • In the outer ring (100–150 microns), it went from 0.00 to 48.77.

This suggests that as the pipes get clogged, the sticky goo doesn't just spread out evenly; it piles up right next to the trouble spots. The researchers suggest this might happen because the clogged pipes block the brain's natural drainage system, causing the goo to back up and settle right next to the vessel walls.

What the Study Didn't Find (and Why That Matters)

It's important to note what this study didn't find. The researchers checked if the total amount of goo in the whole brain changed based on how bad the pipe clogging was. It didn't. The total area of plaques remained relatively stable across all scores, with a median of 1.36 mm² across the group. This means the severity of the disease isn't just about having more goo overall; it's about where that goo is sitting.

The study also looked at other factors. They found that people with the APOE-e4 gene (a known risk factor for brain issues) had significantly more clogged vessels (0.51 mm² on average) compared to those without it (0.12 mm²). However, they found no link between the clogging and the patient's age at death, years of education, or sex.

The Bottom Line

This paper doesn't claim to have cured the problem or found a new drug. Instead, it proves that a digital, automated way of looking at brain tissue is reliable and can see details that human eyes might miss. It suggests that the relationship between clogged blood vessels and the sticky plaques next to them is a key part of the story. By using these digital tools, scientists can now measure the "traffic jams" in the brain's city with much greater precision, potentially helping us understand how these clogs and piles interact to cause memory loss and dementia. The authors suggest this method could help refine how we study these diseases, making future research more consistent and detailed.

Drowning in papers in your field?

Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.

Try Digest →