Liver elastography gives information on the viscoelastic properties of hepatic tissue-liver stiffness (LS). In contrast, serum markers and imaging techniques provide different information. Information held by circulating biomarkers relate to hepatocellular or biliary damage, inflammation, extracellular matrix remodelling, liver function, or portal hypertension, whereas imaging provides information on structural changes and hepatic hemodynamics. Consequently, the combination of different types of information from elastography, blood tests, and imaging methods may provide knowledge and decision-making power that an individual test cannot provide alone. Algorithms are the mathematical combination of different types of information, either in parallel (at the same time) or sequential (one index test, followed by another test depending on the result of the index test). So far, combining elastography with other biomarkers in multivariable models has only discrete impact on the overall diagnostic accuracy for detecting liver fibrosis and cirrhosis across etiologies. A better approach is to address whether elastography and a serum test are in concordance or discordance. However, while this approach may decrease false positives, there is a risk of placing more patients in a “grey zone” of discordant results, than if using elastography alone. The most promising approach is in primary care, where we may greatly improve case finding while keeping screening costs down, by the systematic use of sequential algorithms with an initial, inexpensive serum test, followed by elastography only in case of a positive screening result. New algorithms will likely be developed using machine-learning techniques and will take advantage of the widespread availability of smartphones and software, with calculating power that is vastly superior to the human brain.
|Title of host publication||Liver Elastography : Clinical Use and Interpretation|
|Publisher||Springer Publishing Company|
|Publication date||1. Jan 2020|
|Publication status||Published - 1. Jan 2020|
- Artificial intelligence, ai
- Machine learning
- Portal hypertension