Neutrophils Identification by Deep Learning and Voronoi Diagram of Clusters
Neutrophils are a primary type of immune cells, and their identification is critical in clinical diagnosis of active inflammation. However, in H&E histology tissue slides, the appearances of neutrophils are highly variable due to morphology, staining
- PDF / 1,298,042 Bytes
- 8 Pages / 439.363 x 666.131 pts Page_size
- 10 Downloads / 236 Views
Department of Computer Science and Engineering, University of Notre Dame, USA 2 Department of Radiology and Biomedical Imaging, UCSF, USA 3 Department of Pathology, UCSF, USA
Abstract. Neutrophils are a primary type of immune cells, and their identification is critical in clinical diagnosis of active inflammation. However, in H&E histology tissue slides, the appearances of neutrophils are highly variable due to morphology, staining and locations. Further, the noisy and complex tissue environment causes artifacts resembling neutrophils. Thus, it is challenging to design, in a hand-crafted manner, computerized features that help identify neutrophils effectively. To better characterize neutrophils, we propose to extract their features in a learning manner, by constructing a deep convolutional neural network (CNN). In addition, in clinical practice, neutrophils are identified not only based on their individual appearance, but also on the context formed by multiple related cells. It is not quite straightforward for deep learning to capture precisely the rather complex cell context. Hence, we further propose to combine deep learning with Voronoi diagram of clusters (VDC), to extract needed context. Experiments on clinical data show that (1) the learned hierarchical representation of features by CNN outperforms hand-crafted features on characterizing neutrophils, and (2) the combination of CNN and VDC significantly improves over the state-of-the-art methods for neutrophil identification on H&E histology tissue images.
1
Introduction
Identification of a primary type of immune cells, neutrophils, is of great medical importance. Because the number and the locations are key features for acute inflammation diagnosis [6]; further, quantitative analysis of the distribution patterns of neutrophils may help provide deep insight into acute inflammation. In H&E histology tissue images (Fig. 1(a)), neutrophils are characterized as having multiple lobes in their nuclei, and almost invisible cytoplasms (Fig. 1(b)). But, in practice, it is highly challenging to identify them for the following reasons. First, there are lots of variations in neutrophil appearances due to, e.g., staining, shape, and size. In fact, large portions of neutrophils do not show common
This research was supported in part by NSF Grant CCF-1217906, a grant of the National Academies Keck Futures Initiative (NAKFI), and NIH grant K08-AR06141202 Molecular Imaging for Detection and Treatment Monitoring of Arthritis.
c Springer International Publishing Switzerland 2015 N. Navab et al. (Eds.): MICCAI 2015, Part III, LNCS 9351, pp. 226–233, 2015. DOI: 10.1007/978-3-319-24574-4_27
Neutrophils Identification by Deep Learning and Voronoi Diagram
227
“textbook” characteristics (Fig. 1(c)). Second, the background is quite noisy and complex with a mixture of different biological structures, such as other types of immune cells (e.g., lymphocytes, eosinophils, and plasma cells, see Fig. 1(d)) and some tissue layers (e.g., glands and villi). These structures greatly complicate
Data Loading...
