TGMIL: A hybrid multi-instance learning model based on the Transformer and the Graph Attention Network for whole-slide images classification of renal cell carcinoma.
Xinhuan SunWuchao LiBangkang FuYunsong PengJunjie HeLihui WangTongyin YangXue MengJin LiJinjing WangPing HuangRongpin WangPublished in: Comput. Methods Programs Biomed. (2023)
Keyphrases
- multi instance learning
- image classification
- image recognition
- image annotation
- image data
- image database
- multi instance
- cell nuclei
- input image
- image features
- image retrieval
- cancer cells
- object recognition
- image analysis
- classification accuracy
- text classification
- feature selection
- decision trees
- feature extraction
- support vector
- feature vectors
- model selection
- microscopy images
- pattern recognition
- machine learning
- cost sensitive
- image understanding
- visual features
- pairwise
- semi supervised