Application of Re-Weight Method in Multiple Class-Imbalance Medical Images Detection.

In medical image detection, the dataset often has a situation od class-imbalanced, which causes the long-tail distribution problem in the dataset samples and the performance of the detection model becomes seriously. This paper aims to solve the over-fitting problem of the network when training with multiple class-imbalance dataset, improving the original loss function by re-weight method, and preprocessing the X-ray image by CLAHE algorithm to highlight the internal details of the image. Finally, ResNext50 is selected as the feature extraction network. By using covid-chestxray dataset as the dataset, this paper evaluates the Accuracy, Precision, Recall rate and F1 value of the model. The effectiveness of this method is verified in experiments. [ABSTRACT FROM AUTHOR]