Tips and Tricks to Improve CNN-based Chest X-ray Diagnosis: A Survey
It provides incremental improvements for medical imaging practitioners by compiling known tricks to enhance CNN performance in data-scarce chest X-ray analysis.
This paper addresses the challenge of data scarcity in CNN-based chest X-ray diagnosis by surveying optimization techniques to improve generalization, and demonstrates their application in a solution that increased nodule detection sensitivity by 0.100 for radiologists and 0.131 for non-radiologists while maintaining specificity.
Convolutional Neural Networks (CNNs) intrinsically requires large-scale data whereas Chest X-Ray (CXR) images tend to be data/annotation-scarce, leading to over-fitting. Therefore, based on our development experience and related work, this paper thoroughly introduces tricks to improve generalization in the CXR diagnosis: how to (i) leverage additional data, (ii) augment/distillate data, (iii) regularize training, and (iv) conduct efficient segmentation. As a development example based on such optimization techniques, we also feature LPIXEL's CNN-based CXR solution, EIRL Chest Nodule, which improved radiologists/non-radiologists' nodule detection sensitivity by 0.100/0.131, respectively, while maintaining specificity.