| dc.description.abstract | Histopathology is the diagnosis and study of the different tissues. For diagnosis of different diseases, there are parameters like nuclei count, anatomy of nuclei, etc. However, identifying these parameters manually is a tedious and time- consuming task. Computeraided diagnosis (CAD) helps smoothen the process. In CAD, nuclei segmentation is the crucial task of identifying the nuclei�s anatomy that helps identify the diseases more efficiently. Many deep learning methodologies are present to do the task of nuclei segmentation, but all the methods work on the dataset provided at the time of training. However, there are high chances of the data variabilities present in Histopathology tissue slides due to the different scanners, variations in stain, storage conditions, etc. This kind of data variation affects the performance of any Deep Learning architecture. For that, color normalization techniques help to remove data variations between the histopathological images. So there are many techniques available to do the color normalization, but they have different reactions to the different datasets. So there is a problem with the selection of the normalization techniques. In this study, we have proposed a model that can help to use multiple normalization techniques simultaneously and feed them to the Deep Learning model to get a more robust architecture. We also performed cross data and self data analysis for the all viewed analysis. | |
