The Breast Histopathology Image Retrieval

The Breast Histopathology Image Retrieval Using Graph based Variational Auto Encoder with RetCCL Feature Extractor

We introduced an effective framework to improve histopathological image retrieval performance. This model can effectively capture the high-dimensional features and complex relationships among histopathological images, enabling more meaningful and effective retrieval. Our unsupervised approach incorporates an autoencoder-based graph structure, with graph construction relying on approximate nearest neighbors. This method surpasses the baseline, demonstrating superior results in histopathology image retrieval.

In developing the system, the next steps involve various aspects. Moving on to graph transformers, the emphasis lies on exploring advanced GNN architectures. Furthermore, advancing towards radiology can also contribute significantly to the further development of this project. Lastly, collaboration and knowledge sharing can play a pivotal role, involving active collaboration with healthcare professionals to gain insights and consistent participation in knowledge sharing within the broader research community.

The Breast Histopathology Image Retrieval Using Graph based Variational Auto Encoder with .

Download • 697KB