Deep Learning Approach to Cell Classificatio in Human Peripheral Blood

Abstract

In the health-care sector, things are distinct, unlike other industrial platforms. The priority of the sector is very high and both the players of it and patients need the highest level of services in terms of medical care and diagnostic tools. Almost always, the diagnosis of the medical data is being processed by healthcare experts. When it comes to image analysis, it is thoroughly a challenging process because of the subjective points of view, and possible complexities in the image. Deep learning promises good performance for medical image processing with its success on other real field applications. In this study, a deep learning-based method is proposed for the classification of human peripheral blood cells (PBC). Most of the hematological diseases can be diagnosed with the help of analysis on PBC whilst the morphological interpretation of the abnormalities in PBC is still a complicated process. Hence, this study presents an automated system for the classification of eight types of PBC with a deep learning approach. The proposed model uses a ShuffleNet structure for automatic feature extraction and decision-making with a tremendous dataset. The performance values of the proposed model are high and also the model outperforms the main study that provides the dataset.