Digitally Defining 2 mm² for Mitotic Assessment in Neuroendocrine Tumors: A Technical Comparison of Whole Slide Imaging Approaches
A pilot study comparing conventional manual mitotic counting with two digitally defined 2 mm² approaches for grading neuroendocrine tumors found that while digital methods yielded slightly higher counts and excellent inter-method agreement, they required significantly more time and did not alter tumor grades, highlighting the urgent need for standardized digital protocols and automated detection tools.
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 you are a detective trying to solve a mystery inside a tiny, bustling city made of cells. In the world of medicine, specifically when doctors look at a type of tumor called a neuroendocrine tumor, they need to count how many "construction workers" (dividing cells) are busy building new structures. The more workers they find in a specific area, the more aggressive the tumor is considered to be. This count helps decide if the patient needs a gentle watch-and-wait approach or a heavy-duty treatment.
For decades, detectives used a magnifying glass (a microscope) to look at these cities. They had a rule of thumb: "Count the workers in about ten of these little circular windows." But here's the catch: every microscope had slightly different-sized windows. One detective's "ten windows" might be a tiny neighborhood, while another's was a whole district. This made it hard to compare notes between different hospitals. To fix this, the medical world decided to switch to a fixed-size map: "Count the workers in exactly 2 square millimeters." It's like saying, "Don't count by windows; count by a specific-sized square of land."
Now, imagine the city has been scanned into a giant, high-definition digital photo. This is called a "Whole Slide Image." It's amazing because you can zoom in and out on a computer screen. But there's a new problem: a computer screen doesn't have "windows." The size of the "2 square millimeter" area changes depending on how big your monitor is or how sharp the picture is. If you aren't careful, you might accidentally count a tiny patch or a huge one, throwing off your entire diagnosis. This paper is about figuring out how to draw that perfect, unchanging square on a digital screen so that every doctor, no matter what computer they use, counts the exact same amount of land.
The Digital Detective's Dilemma
In this study, a researcher named Monika Vyas set out to test two different ways to draw that perfect "2 square millimeter" square on a digital photo of a tumor, comparing them to the old-school way of counting through a microscope. She took 20 real-life tumor samples (from the gut and pancreas) and put them under the microscope three times: once the old-fashioned way, once using a digital "grid" (like graph paper laid over the image), and once using a "freehand" digital pen to draw a custom square around the busiest area.
The Results: A Slight Bump, But No Panic
What did she find? Well, the digital methods were a bit more generous than the old microscope. On average, the manual microscope count was 1.75 dividing cells per 2 mm². When the digital grid and the freehand drawing were used, the count jumped slightly to 2.05. That's a 17% increase in the number of cells found.
However, here is the most important part: even though the numbers went up, no one's diagnosis changed. None of the 20 tumors were "upgraded" to a more dangerous category just because the digital method found a few extra cells. The study suggests that while digital counting tends to find a few more cells, it's not a game-changer for these specific cases.
The "Speed Trap" of Digital Pathology
But there was a trade-off. The digital methods were much slower. Counting by hand with a microscope took about 3.1 minutes per slide. The digital grid took 8.05 minutes, and the freehand drawing took 7.3 minutes. That means the digital detectives were working more than twice as long to do the same job! The paper notes that this slowness is because the doctors have to manually find the "hotspot" (the busiest area) and draw the square themselves, rather than the computer doing it for them.
The "Low Count" Problem
The study also found something interesting about the math. When there were very few dividing cells (low counts), the digital methods were a bit more "wobbly" in their agreement with the manual method. The difference between the methods was about 47% on average. This means that if a tumor is right on the edge of a grading threshold (the line between "low risk" and "medium risk"), a digital method might accidentally push it over the line just by finding one or two extra cells. The paper suggests we need to be very careful when the numbers are low and the stakes are high.
What This Means for the Future
The paper concludes that while drawing a digital square works and matches the old microscope method pretty well (with a high agreement score of 0.82 to 0.97), it's not perfect yet. It's too slow, and it finds slightly more cells. The author argues that we can't just switch to digital photos and expect everything to be the same. We need new, "digital-native" rules and better computer programs (Artificial Intelligence) that can automatically find the busy spots and count the cells without making the doctor wait over twice as long. Until those tools are ready and tested specifically for these tumors, we have to be extra vigilant when using digital screens to grade these cancers.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.