Non-invasive imaging techniques that can identify early structural changes due to fibrosis in vivo are of high clinical importance. In this work, a five-level wavelet decomposition of biopsy confirmed normal and fibrotic ex vivo liver tissues is performed and histogram-based features are extracted from the wavelet subbands. A linear classifier is trained using the top 10 features and applied to classify liver fibrosis in Gadolinium-enhanced delayed phase T1-weighted in vivo images. The results show that normal samples yield low posterior probabilities for fibrosis whereas these values are very high for fibrotic samples.
This abstract and the presentation materials are available to members only; a login is required.