Cancer cells detection using Neural Networks
Vladimir Mityushev, Tatjana Gric, R. A. Kycia, Natalia Ryłko · 2025
In this chapter, we construct a simple artificial neural network (NN) to classify cancer in histopathologic scans. It is more technical and demanding since we expect the reader to be familiar with the Digital Image Processing methods described in the previous chapter and have at least basic knowledge of the construction of Artificial Neural Networks described, e.g., in [?,?]. In the previous chapter, we used unsupervised learning that allows an algorithm for clustering similar pixels. We obtained some groups of cells, which experts should examine since the algorithm https://www.w3.org/1998/Math/MathML" display="inline"> k https://www.w3.org/1999/xlink" xlink:href=" https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9781032646633/1d05ecdf-4880-4695-83f9-bd851d783628/content/mathvi_28.tif "/> -Means, which was used, has no internal knowledge of detecting cancer cells. To make the classification more automatic, we must invest additional knowledge in preparing data and then use supervised learning algorithms to learn these examples. The NN approach is one of the most universal approaches to supervised learning that can detect complex relations in the data. We will not focus on the broad theory and practice of specific NN architectures since many resources cover these topics in detail, e.g., [?,?]. In the chapter, we show an example of how to construct and train a model and then use it for classification. This chapter contains two sections. In the first one, we describe the preparation of the classifier; in the second section, we use it to make the classification. Complete examples of the code from this chapter can be downloaded from