Computer-aided-diagnosis of liver fibrosis using non-linear optics microscopy

Abstract

Excessive accumulation of extracellular matrix results in fibrosis, which is the hallmark of chronic liver diseases. The role of liver biopsy as the gold standard for liver fibrosis assessment has recently been challenged due to inter- and intra-observer variation and sampling error. We have developed qFibrosis - a fully-automated classification of liver fibrosis through quantitative extraction of pathology-relevant features using non-linear optics microscopy, trained and tested in both animal and human studies. qFibrosis faithfully recapitulates the liver fibrosis staging performed by pathologists, and is robust with reference to sampling size. It can significantly predict staging underestimation in short biopsy cores, thus aiding in the correction of sampling error-mediated intra-observer variation. qFibrosis can predict the staging underestimation of the non-expert pathologist, thus further aiding in the correction of inter-observer variation. qFibrosis can also significantly differentiate intra-stage cirrhosis changes that can be monitored for making informed clinical decisions, and for predicting possible prognostic outcomes. qFibrosis has the potential to expedite the re-establishment of liver biopsy as the gold standard for assessment of fibrosis in chronic liver diseases. Furthermore, we have hypothesized that the less invasive liver surface imaging could serve as a favourable alternative to biopsy. We established a Capsule Index based on significant parameters extracted from the non-linear optics microscopy images of liver capsule from two fibrosis rat models. The Capsule Index is capable of differentiating different fibrosis stages in both animal models, making it possible to quantitatively stage liver fibrosis via liver surface imaging without biopsy.