We’ve recently seen quite some large-scale data sets popping up, allowing to train deep learning models to automate a variety of tasks, such as plant cell assessment, cellular nucleus segmentation, or mitosis detection, all highly accurate and sometimes even exceeding human experts.

So, while some people already talk about another AI winter, and promises that are not fulfilled, in the biomedical field we just see deep learning models really delivering on the promise and achieving superb results.

Yet — they have a major problem. And most papers are not open about it:

The fine performance of many of these AI…

Microscopy Image Processing on Whole Slide Images

Is less variation in the data really better?

This post is part three of a series. If you don’t know what microscopy stains are, or how they affect the use of images in computerized detection, you can find that in the first part. There we also discussed a popular method of how to estimate the stains from a microscopy image. If you want to get some hands-on experience how to estimate those stains using python, you can find all about that in the second post.

Since we now know how to estimate stains of microscopy slides, we have the complete toolset at our fingertips to get a feeling…

Microscopy Image Processing on Whole Slide Images

Getting a stable stain estimate on gigapixel images.

This post is part of a series. If you don’t know what a microscopy stain is, how much and why it varies so strongly, or how to estimate it on small image patches, you can find all of that in the first part: “Microscopy stain variations and how to estimate them”.

This time, we want to get a little bit more hands-on and solve a problem that you might have if you deal with whole slide microscopy images: Those images are really big. And by that, I mean that they can easily exceed 100,000 pixels in width and height. …

Microscopy Image Processing on Whole Slide Images

How Macenko’s method works

If you have seen a couple of microscopy images of tissue (histopathology images), you must have noticed that they come in all variants of colors. Even when the same dying chemicals (stains) are used, the visual appearance is influenced by so many factors that it can easily become a big problem if you work with those images algorithmically.

The default stain in histopathology is a combination of two chemicals: hematoxylin and eosin. The first is responsible for the dark violet (or blueish) color of all acid components (like, e.g., the DNA, residing in the cellular nucleus), while the second is…

Prof. Marc Aubreville

I do teaching and research in medical image recognition at THI, Germany, primarily focused tumor diagnostics.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store