A Case for the Score: Identifying Image Anomalies using Variational Autoencoder Gradients
This addresses anomaly detection for computer-aided diagnosis, offering a theoretically grounded method that outperforms prior approaches, though it is incremental as it builds on existing VAE frameworks.
The paper tackled unsupervised pixel-wise anomaly detection in medical images by proposing a gradient-based rating from a Variational Autoencoder, achieving a ROC-AUC of 0.94 on the BraTS-2017 dataset for tumor detection.
Through training on unlabeled data, anomaly detection has the potential to impact computer-aided diagnosis by outlining suspicious regions. Previous work on deep-learning-based anomaly detection has primarily focused on the reconstruction error. We argue instead, that pixel-wise anomaly ratings derived from a Variational Autoencoder based score approximation yield a theoretically better grounded and more faithful estimate. In our experiments, Variational Autoencoder gradient-based rating outperforms other approaches on unsupervised pixel-wise tumor detection on the BraTS-2017 dataset with a ROC-AUC of 0.94.