.png){kind=link}
BFuse-Net: Bonferroni Mean Operator-aided Fusion of Neural Networks for Medical Image Classification
This is the official implementation of "BFuse-Net: Bonferroni Mean Operator-aided Fusion of Neural Networks for Medical Image Classification".
Presented in '5th International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD)', 19 - 21 Nov 2024, University of Manchester, UK.
Triyas Ghosh
Soham Chakraborty
Dmitrii Kaplun
Vyacheslav Gulvanskii
Ram Sarkar
Please do cite our paper in case you find it useful for your research.
Citation -
@InProceedings{10.1007/978-981-96-3863-5_10,
author="Ghosh, Triyas
and Chakraborty, Soham
and Kaplun, Dmitrii
and Gulvanskii, Vyacheslav
and Sarkar, Ram",
editor="Su, Ruidan
and Frangi, Alejandro F.
and Zhang, Yudong",
title="BFuse-Net: Bonferroni Mean Operator-Aided Fusion of Neural Networks for Medical Image Classification",
booktitle="Proceedings of 2024 International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD 2024)",
year="2025",
publisher="Springer Nature Singapore",
address="Singapore",
pages="100--109",
abstract="In this paper, we have proposed a novel model, called Bonferroni Mean Operator-aided Fusion of Neural Networks (BFuse-Net). Here, we have taken advantage of the capabilities of four deep learning models as the base learners and then applied a customized attention method to prioritize different fine-grained features by passing their outputs via four parallel pipelines. Next, we have employed a novel aggregation method, a modified Bonferroni Mean operator, to merge the decision scores acquired from every pipeline, prioritizing their individual contributions above their interactions with others in the decision scores. Three medical image datasets--- LC25000 (colon cancer), Oral Squamous Cell Carcinoma, and Malaria Cell Images have been used for experimentation. The model yields state-of-the-art outcomes for the three datasets. The source codes and additional results can be found at GitHub repository.",
isbn="978-981-96-3863-5"
}Link to our paper - https://link.springer.com/chapter/10.1007/978-981-96-3863-5_10