Pseudo Healthy Reconstruction

Pseudo-healthy reconstruction is an unsupervised anomaly detection technique in medical imaging that uses deep generative models, such as variational autoencoders and reversed autoencoders, to learn the characteristics of healthy tissue from a dataset of healthy images. This learned representation is then used to reconstruct a "pseudo-healthy" version of a new image; deviations from this reconstruction highlight potential anomalies, enabling the detection of various pathologies without requiring labeled data. Current research focuses on improving the accuracy and robustness of these reconstructions, particularly by incorporating information about healthy population variability and exploring multi-contrast image fusion techniques to enhance diagnostic capabilities across different imaging modalities. This approach holds significant promise for improving the efficiency and accuracy of medical image analysis, particularly in identifying rare or subtle anomalies.