The rapid development in health care and research is responsible for the data explosion from diverse modalities such as clinical data, imaging data and molecular data. It is necessary to extract salient features from these multimodal datasets. Multimodal data fusion is the process that integrates the data acquired from different modalities to improve the diagnostic accuracy, management, and implementing personalized medicine. In addition, multimodal data fusion provides a complete understanding of the system. The level of data fusion may be in the early, intermediate and late stages. Autoencoders, Multi-Channel Neural Networks and Generative Adversarial Networks are used for data fusion. The present chapter discusses the technique of multimodal data fusion, challenges and its application in cytology.