Abstract PDF Code BibTeX

	Class imbalance in the medical image dataset is almost inherent due to the limited availability of clinical data for certain diseases and patient populations. Under-represented classes in the training set affect the classification task because the classifier tends to learn more from the majority classes, which are more common in the dataset and ignore data from the minority classes. To mitigate this issue, we propose a method to learn using two different convolutional neural network-based experts; such experts try to learn boundaries within the head classes, between the head and tail classes, and within the tail classes. During expert training, we integrate the MixUp regularization method to augment imbalanced data, employing distinct data sampling strategies for more effective mixing compared to random selection in traditional MixUp. During the inference phase, we combine the logits of the different experts based on their expertise in the corresponding classes. This way, we can improve the accuracy of the head and tail classes. Experiments using highly imbalanced and long-tailed datasets demonstrate the effectiveness of the suggested framework.

	@inproceedings{ansari2024mo2e,
  		title={Mo2E: Mixture of Two Experts for Class-Imbalanced Learning from Medical Images},
  		author={Ansari, Faizanuddin and Bhattacharya, Agnish and Saha, Biswajit and Das, Swagatam},
  		booktitle={2024 IEEE International Symposium on Biomedical Imaging (ISBI)},
  		pages={1--5},
  		year={2024},
  		organization={IEEE}
	  }
	

Abstract PDF Code BibTeX

	One of the most frightening and talked-about diseases in the modern world is cancer. Huge amounts of research are conducted worldwide to make this ailment less fearsome, be it by finding its cure, discovering ways to detect it in much earlier stages to reduce the mortality rate, or identifying precautions for humans to avoid it. The availability of large collections of biomedical and clinical data has ushered in the use of computer vision for cancer detection, especially for two of its most common types, lung and colon carcinomas. In this work, we present a framework wherein both deep learning and meta-heuristic approaches have been used for the prediction of colon or lung cancer, or both, from histopathological images with near-perfect precision. Initially, deep learning models, namely ResNet-18 for 2-class classification and EfficientNet-b4-widese for 3-class and 5-class classification, have been trained on the LC25000 dataset, followed by the extraction of deep features. The feature vector obtained from a deep learning model may have some redundancy. Hence, the selection of the most useful features has been done with the application of our proposed hybrid meta-heuristic optimization algorithm, AdBet-WOA (Whale optimization algorithm with integrated Adaptive β-Hill Climbing local search), utilizing which the Support Vector Machine (SVM) classifier classifies the colon cancer test data, lung cancer test data, and both combined with an accuracy of 99.99%, 99.97%, and 99.96%, respectively, matching the benchmark results comprehensively. For comparison, we have used a few independent as well as hybrid optimization algorithms. Our proposed approach succeeds greatly in reducing the number of features and also leads to better classification performance, as indicated by the obtained results. The relevant codes for our proposed approach are publicly available at: https://github.com/raj-1411/AdBet-WOA

	@article{bhattacharya2023deep,
		title={Deep feature selection using adaptive $\beta$-Hill Climbing aided whale optimization algorithm for lung and colon cancer detection},
		author={Bhattacharya, Agnish and Saha, Biswajit and Chattopadhyay, Soham and Sarkar, Ram},
		journal={Biomedical Signal Processing and Control},
		volume={83},
		pages={104692},
		year={2023},
		publisher={Elsevier}
	  }