MetaMed: Few-shot medical image classification using gradient-based meta-learning

The occurrence of long-tailed distributions and unavailability of high-quality annotated images is a com-mon phenomenon in medical datasets. The use of conventional Deep Learning techniques to obtain an unbiased model with high generalization accuracy for such datasets is a challenging task. Thus, we for-mulated a few-shot learning problem and presented a meta-learning-based MetaMed approach. The model presented here can adapt to rare disease classes with the availability of few images, and less com-pute. MetaMed is validated on three publicly accessible medical datasets - Pap smear, BreakHis, and ISIC 2018. We used advanced image augmentation techniques like CutOut, MixUp, and CutMix to overcome the problem of over-fitting. Our approach has shown promising results on all the three datasets with an accuracy of more than 70%. Inclusion of advanced augmentation techniques regularizes the model and in-creases the generalization capability by 2-5%. Comparative analysis of MetaMed against transfer learning demonstrated that MetaMed classifies images with a higher confidence score and on average outperforms transfer learning for 3, 5, and 10-shot tasks for both 2-way and 3-way classification. (c) 2021 Published by Elsevier Ltd.

Automated summary

The study addresses the challenges of long-tailed distributions and lack of high-quality annotated images in medical datasets. By formulating a few-shot learning problem and proposing a meta-learning-based MetaMed approach, the model achieved promising results with an accuracy of over 70% on three medical datasets, showcasing improved generalization capability.