Generalization of Deep Convolutional Neural Networks -- A Case-study on Open-source Chest Radiographs
This work addresses the challenge of model generalization in medical image analysis, which is crucial for reliable clinical diagnosis, though it is incremental in nature.
The study investigated the generalization of deep convolutional neural networks for predicting five common chest pathologies across three public datasets, finding that internal performance consistently exceeded external performance but that training on a mix of heterogeneous datasets improved external generalization.
Deep Convolutional Neural Networks (DCNNs) have attracted extensive attention and been applied in many areas, including medical image analysis and clinical diagnosis. One major challenge is to conceive a DCNN model with remarkable performance on both internal and external data. We demonstrate that DCNNs may not generalize to new data, but increasing the quality and heterogeneity of the training data helps to improve the generalizibility factor. We use InceptionResNetV2 and DenseNet121 architectures to predict the risk of 5 common chest pathologies. The experiments were conducted on three publicly available databases: CheXpert, ChestX-ray14, and MIMIC Chest Xray JPG. The results show the internal performance of each of the 5 pathologies outperformed external performance on both of the models. Moreover, our strategy of exposing the models to a mix of different datasets during the training phase helps to improve model performance on the external dataset.