Automating Morphological Profiling with Generic Deep Convolutional Networks

By Nick Pawlowski, Juan C. Caicedo, Shantanu Singh, Anne E Carpenter, Amos J. Storkey. 2016

90 citations recorded by the source corpus.

Morphological profiling aims to create signatures of genes, chemicals and diseases from microscopy images. Current approaches use classical computer vision-based segmentation and feature extraction. Deep learning models achieve state-of-the-art performance in many computer vision tasks such as classification and segmentation. We propose to transfer activation features of generic deep convolutional networks to extract features for morphological profiling. Our approach surpasses currently used methods in terms of accuracy and processing speed. Furthermore, it enables fully automated processing of microscopy images without need for single cell identification.

Subfields: Biophysics, Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biophysics

1. Shantanu Singh
2. Juan C. Caicedo
3. Anne E Carpenter
4. Casey S. Greene
5. Nick Pawlowski
6. G. Way
7. Peter Goldsborough
8. Brett K. Beaulieu-Jones

In Computer Vision and Pattern Recognition

1. Amr M. Alexandari
2. Anshul B Kundaje
3. Avanti Shrikumar
4. Brett K. Beaulieu-Jones
5. Zhiyong Lu
6. Juan C. Caicedo
7. Shantanu Singh
8. Anne E Carpenter

In Media Technology

1. Shantanu Singh
2. Juan C. Caicedo
3. Nick Pawlowski
4. Peter Goldsborough
5. Zhiyong Lu
6. Anne E Carpenter
7. Yanjun Qi
8. Yifan Peng

Citations in the corpus

• Opportunities and obstacles for deep learning in biology and medicine (2017, background): “In image analysis, previous examples of deep transfer learning applications proved large-scale natural image sets [42] to be useful for pre-training models that serve as generic feature extractors for various types of biological images [14,206,414,415].”
• CytoGAN: Generative Modeling of Cell Images (2017, methodology): “[16] for the first time reported a representation-learning method based on deep learning that is competitive with hand-engineered features at the task of predicting mechanism-ofaction (MOA) of chemicals.”
• Deep learning for cellular image analysis (2019, methodology): “Because of this similarity, and because of the relative lack of annotated training data for biological images, transfer learning has featured prominently in the creation of image classifiers that perform well on biological data 50,76 .”