Assessing liver health through morphometry
The aim of this study is to determine if the amount of fibrosis in the liver can be determined by looking at the shape and size of the liver, the spleen and the hepatic (liver) blood vessels.
The liver can be divided into three sections - the right lobe, the left medial lobe and the left lateral lobe. The two left lobes can be further divided into a top and bottom (or superior and inferior) section. The right lobe can be divided into 4 sections by dividing it into a top and bottom section, and a front and back (anterior and posterior). Many researchers have looked at how these lobes and sections change in size and shape when there is fibrosis present in the liver. Generally, the left lateral lobe gets larger and the right lobe gets smaller as the amount of fibrosis increases. The spleen also tends to enlarge with increasing fibrosis in the liver. Additionally, the surface of the liver gets rougher, the hepatic veins tend to get narrower, and a notch sometimes appears in the bottom back section of the right lobe of the liver. Most of this information can be extracted from the MR images obtained in this study. Combining this morphometry information with the measured fat and the corrected-T1 should provide a relatively complete picture of the overall liver health without having to take a biopsy.
A multi-stage deep learning approach will be used to identify the spleen, the liver, and the different segments within the liver. For the liver and spleen identification, the first stage will identify which images contain the organ of interest. From those images, the second stage will identify the boundary of the organ based on learnings from manual identification of the organ.
For the individual segments in the liver, the first stage will identify images that contain anatomical landmarks that can be used to divide the liver into the different segments. A second stage will take these identified images and attempt to draw boundary lines between the segments. The final stage will divide the liver into the different segments based upon the boundary lines that were identified.
Once these regions are identified, the volumes and relative sizes can be calculated as described in the literature to assess fibrosis/cirrhosis risk.
It is anticipated that this research project will take approximately 24 months.