Computer Vision News - July 2019

Another complexity of this task emerges from the multitude of techniques for tissue staining . One of the most advanced techniques used in the study of cancer microenvironment is multiplex , which allows dying the same tissue with different specific fluorescents. The stained slice is photographed multiple times using different wavelengths and the different images are overlaid to produce a merged picture. This allows a classification of different cell types within the same slide, while conserving its spatial structure. Using automatic analysis of these images, opens new opportunities to shed more light on multiple open research questions. The main aspect of our solution was the decision to separate the project in two parts, enabling us to be specifically precise in each one of them. The other valuable choice that we made was the use of non- classic computer vision algorithms and techniques, in particular the use of deep learning. In contrast to classic algorithms, which require a detailed definition of conditions and heuristics which are specific to tissue, staining technique and even the microscope resolution, deep learning is independent with all of these and can fit different data types. Thanks to deep learning, when you train your model on many examples it learns by itself to identify and classify the different cells emerging from different tissues and dyed by different staining techniques. Read about more projects in medical segmentation Project 5 Computer Vision News A project by RSIP Vision Take us along for your next Deep Learning project! Request a call here