WBCclassification

In the last decade, convolutional neural networks, a form of deep learning algorithms, have become the methodology of choice for automated medical image analysis. It is used for various tasks across a variety of fields within the field of medical image analysis. Despite being successfully applied in the field of medical image analysis, challenges of training convolutional neural networks for the purpose of medical image analysis do remain. Some of these challenges include but are not limited to a lack of large annotated datasets, datasets often being imbalanced and a large variability in medical images. These problems may be overcome by using synthetically generated medical images. Using peripheral blood smears as a test case, convolutional neural networks for the purpose of single white blood cell classification and segmentation were trained on procedurally generated white blood cell images. Real white blood cell images were used for evaluating the performance of the convolutional neural networks. In this repository, scripts used for this project are stored. The real cell images were extracted from images of the LISC database (available at: http://users.cecs.anu.edu.au/~hrezatofighi/Data/Leukocyte%20Data.htm)