Back

aDISCO: A Broadly Applicable Method for 3D Microscopy of Archival Paraffin-Embedded Human Tissues

Reuss, A. M.; Rupprecht, P.; Groos, D.; Cerisoli, M.; Nordberg, L.; Frick, L.; Voigt, F.; Vladimirov, N.; Bethge, P.; Reimann, R.; Helmchen, F.; Aguzzi, A.

2025-05-26 bioengineering
10.1101/2025.05.21.655358 bioRxiv
Show abstract

Human surgery and autopsy specimens are routinely stored as formalin-fixed paraffin-embedded (FFPE) tissue blocks. Diagnoses are based on microscopic examination of two-dimensional sections. Good clinical practice requires that samples be retained for decades, thus giving rise to vast archives of healthy and diseased tissues. While tissue clearing and whole-mount microscopy enable 3D analysis, FFPE human tissue blocks are often unsuitable for clearing and immunolabeling due to their large size, extensive cross-linking, and wax embedding. Here, we introduce archival DISCO (aDISCO), a versatile and robust clearing method designed to overcome these challenges. aDISCO achieves complete clearing and immunolabeling of large samples stored for 15 years or more. We applied aDISCO to human brain, spinal cord, peripheral nerve, skin, muscle, heart, kidney, liver, spleen, colon, and lung, using a broad range of antibodies. To investigate the clinical usefulness of quantitative 3D histology by aDISCO, we performed a case study of focal cortical dysplasia, a neurodevelopmental disease causing epilepsy. Using deep-learning-based analysis, we found disrupted cortical layering and both focal and global neuronal density variations in FCD, features that are likely to be overlooked by conventional histology. In summary, aDISCO delivers large datasets suitable for deep-learning-based processing, enabling the detection of subtle and sparse pathologies in large archival human tissue specimens.

Published in Science Advances (predicted rank #1) · training set

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.