NNCompare: a framework for dataset selection, data augmentation and comparison of different neural networks for medical image analysis

In our institute we capture a variety of medical image data - in particular, microscopy data for the use case of bronchoconstriction. In order to alleviate the manual intervention for the image analysis, we developed a comprehensive data analysis pipeline that automates the image preprocessing, data augmentation, as well as neural network training and deployment with a web-based user interface. We evaluate the pipeline on a real-world medical image dataset and comparatively analyze the performance of four different neural network architectures.