BME Frontiers / 2022 / Article / Fig 3

Research Article

Weakly- and Semisupervised Probabilistic Segmentation and Quantification of Reverberation Artifacts

Figure 3

Our proposed probabilistic network is similar to Hierarchical Probabilistic U-Net [19], but instead of learning the posterior with labeled segmentation masks, our method learns mean and variance of the posterior from known mean and variance separately. Therefore, two variational autoencoders are used in the posterior.