9-point control list to improve an image analysis based on artificial intelligence in pathology

A new article in the introduction of a 9-point checklist designed to improve the quality of tests that use automatic image analysis based on artificial intelligence (AI) (AI). Because AI tools become wider in pathology -based research, fears of playback and transparency of published results have arisen.

Developed by an interdisciplinary team of veterinary pathologists, machine learning experts and magazine editors, the control list presents key methodological details that should be included in the manuscripts. They include creating a data set, model training, performance assessment and interaction with AI. The goal is to support clear communication of methods and reduce cognitive and algorithmic prejudices.

“Transparent reporting is crucial for the playback and translation of AI tools into routine flows of pathology,” write the authors. They emphasize that the availability of data support, like training sessions, source code and model mass, is necessary for significant validation and wider application.

The guidelines are designed to help authors, reviewers and editors and will be particularly useful for applications for the upcoming special edition in artificial intelligence.