A semi‐automated method to assess intraepidermal nerve fibre density in human skin biopsies. Issue 5 (21st September 2015)

• Record Type: Journal Article
• Authors: Seger, Shanon; Stritt, Manuel; Doppler, Kathrin; Frank, Stephan; Panaite, Adrian; Kuntzer, Thierry; Steck, Andreas; Pagenstecher, Axel; Sommer, Claudia; Stalder, Anna K
• Is Part Of: Histopathology. Volume 68:Issue 5(2016)
• Page Start: 657
• Page End: 665
• Publication Date: 2015-09-21
• Subjects: automated quantification -- IENF -- image analysis -- peripheral neuropathies -- skin biopsy; Histology, Pathological -- Periodicals; 611.018
• DOI: 10.1111/his.12794
• Languages: English
• ISSNs: 0309-0167

Abstract

Abstract : Aims: Evaluation of intraepidermal nerve fibres (IENFs) in skin biopsies is used in the diagnosis of small‐fibre neuropathies. The number of IENFs is assessed manually under a microscope, with an inter‐rater variability of ~25%. Unless the images are digitized, there is no documentation. Our aim was to develop a method for standardized semi‐automated quantification (SAQ) and documentation of IENF density. Methods and results: We analysed samples from four different university centres that were immunostained according to local protocols. Images were acquired through the Z ‐plane with a whole slide scanner.orbit image analysis software was used to create an analysable image and develop a reliable algorithm for IENF detection. Rebuilt images revealed well‐contrasted nerves, allowing detection of IENFs (automated). The software presented the detected nerves for confirmation by the operator (manual). As compared with the conventional microscopy count, the SAQ achieved correlation coefficients of 0.99 and 0.96 and interfacility variabilities of 19% and 23%, respectively. We found better reproducibility with fluorescence‐stained specimens than with bright‐field images. Conclusions: The new semi‐automated method has high experimenter‐independent reproducibility when based on nerve detection by fluorescence and is easy to perform, even by untrained users. The IENF counting is electronically well documented.