The digitization of whole-slide imaging within digital pathology has propelled the advancement of computer-assisted tissue examination utilizing machine learning methodologies, particularly convolutional neural networks (CNNs). Numerous CNN-based approaches have been proposed to effectively analyze histopathological images for tasks such as cancer detection, risk prediction, and cancer subtype classification. While many existing methods have relied on patch-based examination due to the immense size of histopathological images, such small window patches often lack sufficient information or patterns for the tasks at hand. Pathologists routinely inspect tissues at various magnification levels to scrutinize complex morphological patterns through microscopes. In response to these challenges, we propose a novel deep-learning model for histopathology, named Deep-Hipo, which concurrently utilizes multi-scale patches for precise histopathological image analysis. Deep-Hipo simultaneously extracts two patches of identical size from both high and low magnification levels, enabling the capture of intricate morphological patterns within both large and small receptive fields of a whole-slide image.