In biomedical research, microscopic image analysis is a fast-growing research field that imparts quantitative assistance for enhancing the characterizations of many diseases, such as breast cancer, lung cancer, brain tumor, and many more.
Due massive amount of microscopic image data, deep learning is applied for microscopic image analysis, which is one of the popular computerized methods. Deep learning has accomplished outstanding performance in various microscopy image analysis applications.
Exciting advancements in microscopy technologies have recently influenced tremendous growth in the quality and quantity of image data collected in biomedical research. In microscopy image analysis, deep learning models are often utilized for many tasks, including object detection, tracking, reconstruction, segmentation, and classification. High image dimension, image artifacts and batch effects, object crowding and overlapping, and insufficient, imbalanced, and inconsistent data annotations are the unique challenges in microscopy image analysis.
Convolutional neural networks, fully convolutional networks, recurrent neural networks, stacked auto-encoders, and deep belief networks are common deep learning architectures utilized for microscopic image analysis. The modern development of augmented intelligent microscopy is based on deep learning technology, bringing a revolution in biomedical research. Many deep learning-based microscopic image analysis surveys and reviews are published, presenting deep learning-based methods, challenges, evaluations, future trends, limitations, and potential solutions.