I will present a high-level overview of how automated image analysis approaches can be incorporated into pharmaceutical discovery pipelines. By taking a look at two GSK case studies I will demonstrate how to apply computer vision techniques to featurize imaging data, enabling the use of standard machine learning algorithms. I will highlight how these techniques benefit the drug discovery process.
I will present a high-level overview of how automated image analysis approaches can be incorporated into pharmaceutical discovery pipelines. I will explore the nature of imaging features such as Zernike moments, Haralick coefficients and parameter-free TAS.
I will then demonstrate how to use computer vision libraries (OpenCV & mahotas) to extract these features from microscopy images and how to use them as input to machine learning models implemented in sklearn. I will highlight how these techniques benefit the drug discovery process.
Prior knowledge of pandas and sklearn required.