Evaluation of pseudo-healthy reconstruction for anomaly detection in brain FDG PET

Ravi Hassanaly, Camille Brianceau, Maëlys Solal, Olivier Colliot, Ninon Burgos

Keywords: Deep learning, Pseudo-healthy reconstruction, Unsupervised anomaly detection, Variational autoencoder, 3D PET, Alzheimer's disease

Abstract: We propose an evaluation procedure based on the simulation of realistic abnormal images to validate pseudo-healthy reconstruction methods when no ground truth is available. We apply this framework to the reconstruction of 3D brain FDG PET using a convolutional variational autoencoder. This work has recently been published at MELBA.

Submission Number: 167