nucleAIzer: A Parameter-Free Deep Learning Framework for Nucleus Segmentation Using Image Style Transfer

Hollandi, Réka and Szkalisity, Ábel and Tóth, Tímea and Tasnádi, Ervin and Molnár, Csaba and Máthé, Botond and Grexa, István and Molnár, József and Bálind, Árpád and Görbe, Máté and Kovács, Mária and Migh, Ede and Balassa, Tamás and Koós, Krisztián and Bara, Norbert and Kovács, Ferenc and Danka, Tivadar and Kriston, András and Horváth, Péter (2020) nucleAIzer: A Parameter-Free Deep Learning Framework for Nucleus Segmentation Using Image Style Transfer. CELL SYSTEMS, 10 (5). pp. 453-458. ISSN 2405-4720

Preview

Text 353535.pdf Available under License Creative Commons Attribution. Download (1MB) | Preview

Abstract

Single-cell segmentation is typically a crucial task of image-based cellular analysis. We present nucleAIzer, a deep-learning approach aiming toward a truly general method for localizing 2D cell nuclei across a diverse range of assays and light microscopy modalities. We outperform the 739 methods submitted to the 2018 Data Science Bowl on images representing a variety of realistic conditions, some of which were not represented in the training data. The key to our approach is that during training nucleAIzer automatically adapts its nucleus-style model to unseen and unlabeled data using image style transfer to automatically generate augmented training samples. This allows the model to recognize nuclei in new and different experiments efficiently without requiring expert annotations, making deep learning for nucleus segmentation fairly simple and labor free for most biological light microscopy experiments.

Actions (login required)

Edit Item