Lakera releases robustness testing suite for digital pathology
Lakera now offers you the opportunity to easily test whether your algorithms are robust to histological artifacts and variations. It lets you stress test your computer vision models to gain confidence in their robustness properties prior to clinical validation and deployment.
We are thrilled to announce the release of our robustness testing suite for state-of-the-art computer vision in digital pathology. With the explosion of this new field, the medical sector is experiencing a tremendous flurry of activity, particularly in the space of medical imaging.
**Pro tip:💡 Want to join other pathology companies in automating model validation with MLTest? You can get started in minutes.**
A major headache for pathology imaging, however, is ensuring model robustness during operation. It's not only that histological slides are very heterogeneous, but the abundance of histological artifacts that commonly appear in operation can lead to severe underperformance as well. These include dust particles, oily spots, loose cells from tissue tearing, staining - the list goes on and on. Unfortunately, each one of these culprits can play havoc with your algorithms, leading to missed objects or too much debris masquerading as your target.
A few examples of Lakera’s robustness tests for pathology. From left to right: dark spots caused by e.g. dust, an oily spot caused by e.g. fingerprints, squamous epithelial cells, and synthetic threads causing local focus deterioration. These histological artifacts typically lead to severe model underperformance during operation.
Digital pathology teams are already using Lakera to speed up model validation, train better models, and expedite clinical validation and certification processes.
With this latest release, they bring their machine learning testing capabilities to a new level. This test suite includes new types of robustness tests, such as:
Dark spots, often caused by dust particles or glass scratches.
Oily spots, often caused by residue from, for example, fingerprints.
Squamous epithelia, one of the most common artifacts resulting from contamination of biopsy specimens during tissue processing.
Synthetic threads, a common artifact that causes focus deterioration.
Image qualitydifferences, often caused when data comes from multiple scanning devices.
Variations in focus, often caused by the presence of foreign objects.
Variations in lighting conditions (e.g. brightness and contrast) often caused by differences in tissue, processing, or scanning devices.
Lakera now offers you the opportunity to easily test whether your algorithms are robust to histological artifacts and variations. It lets you stress test your computer vision models to gain confidence in their robustness properties prior to clinical validation and deployment.
Does your tumor detector still work with dust on the slides? Can something as simple as extra tissue cells dramatically increase your number of false positives? If questions like these are on your mind, then head on over to MLTest to learn why leading medical companies trust Lakera.
Learn how to protect against the most common LLM vulnerabilities
Download this guide to delve into the most common LLM security risks and ways to mitigate them.
Great news for all AI radiology teams—you can now take your medical machine learning testing capabilities to a new level with MLTest. You can now easily test whether your algorithms are robust to radiological artifacts and variations.