Deep learning techniques for biomedical image classification

• Main Author: Nguyen, Duc Long
• Other Authors: Lin Zhiping
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering; DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/75368
• Institution: Nanyang Technological University

Deep convolutional neural networks (deep CNNs) is currently the state-of-the-art methods for image classification. In biomedical image field, deep learning can be used with transfer learning to address the problem of small training datasets. For deep CNNs that are pretrained on large image datasets, their layers are general enough to be capable of generating useful features for an unseen biomedical dataset. In this research, we propose various deep CNNs based on transfer learning technique. We come out with a feature concatenation scheme that effectively exploit the power of deep neural networks. We also experiment new technique such as ensemble learning for accuracy improvement. The experiments on the 2D-Hela and the PAP-smear microscopic datasets show that our proposed structures have significant performance improvements over several traditional classification methods. Part of this project have been accepted as a poster session in the IEEE International Symposium on Circuits and Systems (ISCAS) 2018 conference.