Expert-annotated datasets for deep-learning based classification methods for morphologic Leukaemia diagnostics

Christian Matek, Carsten Marr

Keywords: Microscopy imaging, cytopathology, deep learning, single-cell data, multiple-instance learning.

Abstract:

In recent years, image analysis and classification methods have become increasingly popular for segmentation and classification tasks in pathology imaging, specifically for the evaluation of tissues sections and single cells.
Hematological cytomorphology, with single cell-classification at its base and lacking some of the complexities of tissue architecture, has been at the forefront of this development. Recent additions of large-scale datasets to the public domain open up new possibilities for developing and benchmarking algorithms for morphologic diagnostics, prognostication and subgrouping.

Submission Number: 164