• FILES (0)

The software solution developed is an automated tool using principles of machine-learning, and requiring only 2xmagnification of images. The solution has been validated against experts’ annotation with excellent concordance.

The software enables:

  • Accuracy of analysis
  • Reduction on time for analysis (2-5 mins)
  • Faster and more appropriate care


The tool could be used to supplement traditional histology to achieve more accurate diagnoses. It has a number of advantages over traditional histology:


The technique engages machine-learning techniques using algorithms with classic image processing methods. The methodology for steatosis and fibrosis measurement includes three stages:


The solution developed and validated is a highly novel technique for high-throughput, objective quantitation of steatosis and fibrosis in liver biopsies from those with NAFLD. Uniquely, this technique is fully-automated employing machine-learning software and requires no highly specialized equipment, making it directly applicable to clinical practice.

The automated image analysis operates with low sophisticated equipment, such as common microscopes and low-magnification JPEG images extracted from NDPI. Each image of liver biopsy is analyzed in a 2-5 minutes.

For those Centers interested in medical research and clinical trials, the use of such a solution leads to quantifiable measures, which expand the field of statistical analysis.

The solution can also be modified for non-liver pathology, i.e. colon, pancreas.



NAFLD is an increasing cause of chronic liver disease worldwide. Histology remains the reference standard for diagnosis and to assess the degree of steatosis, inflammation and fibrosis in patients with NAFLD. However, liver biopsies are currently interpreted using semi-quantitative scores, which show low agreement between pathologists and unreliable results. These limitations affect not only clinical practice, but also the interpretation of clinical trials. Also existing image analysis techniques for the quantitation of steatosis and fibrosis in liver biopsies require manual threshold detection, introducing a risk of bias and are often slow and cumbersome, and require specialised equipment.


To licence this tool in a clinical or clinical trial setting, please contact manish.patel@imperialinnovations.co.uk

Automated Assessment for Non-Alcoholic Fatty Liver Disease

Machine-learning based software, validated against experts' annotations

Please check carefully that the terms you select correspond to your intended use of the product.

General POA Licence 0.20

View Terms

Further details
Term: perpetual (does not require renewal)
Seats: n/a

Price excl. VAT: P.O.A. (price on application)

Please contact us before adding this licence to your basket.