While challenging, the dense segmentation of histology images is a necessary\nfirst step to assess changes in tissue architecture and cellular morphology.\nAlthough specific convolutional neural network architectures have been applied\nwith great success to the problem, few effectively incorporate visual context\ninformation from multiple scales. With this paper, we present a systematic\ncomparison of different architectures to assess how including multi-scale\ninformation affects segmentation performance. A publicly available breast\ncancer and a locally collected prostate cancer datasets are being utilised for\nthis study. The results support our hypothesis that visual context and scale\nplay a crucial role in histology image classification problems.\n