Embedded Files
2019 – Now, Dynamic Model for Disease Forecast
- Established a sequential fundus image database for glaucoma forecast
- Proposed a variable time interval CNN+LSTM model based on the sequential samples
- Introduced an active convergence training strategy for the imbalanced distribution problem
2018 – 2019, Attention Mechanism-based Medical Image Detection and Visualization
- Proposed an attention-based CNN model for glaucoma detection, with hierarchical features
- Visualized the pathological regions on the fundus images by guided-bp method
- Embedded weakly supervised learning method motivated by the rotation invariance
2017 – 2018, Generalized DNN System for Disease Classification
- Constructed and standardized a large-scale medical image database (200k fundus images)
- Developed a robust CNN method and achieved high accuracy (96%) in glaucoma detection
- Visualized the CNN temporally and spatially by feature embedding and feature occlusion
- Introduced online learning method for real application scenario
2018 – 2019, Pathology-aware Deep Network Visualization
- Proposed a pathology-aware visualization approach for DNN-based glaucoma classification
- Synthesized glaucoma fundus images with GANs based on the visualization maps
2016 – 2017, Study on Multi-face Alignment Algorithm in Video
- Proposed a method of face alignment in video, based on the correlations between frames
Page updated
Google Sites
Report abuse