Since the start of the Athero-Express Biobank Study in 2002 we have collected atherosclerotic plaque samples of over 3,500 patients. Each of these samples have been stained for various plaque characteristics, including smooth muscle cells based on smooth-muscle cell actin (SMA), macrophages based on CD68, intraplaque hemorrhage, and intraplaque microvessels based on CD34 among others.

The community standard is to analyze histological slides manually, with the downside of inter- and intra-observer variability leading to a “noisy” phenotype. To improve on this we have recently started to scan all slides (some 35,000!) at high-resolution using standard pathology scanners, alongside we have developed slideToolkit (Nelissen BGL et al. PLoS One 2014 | PMID: 25372389 and our GitHub-page) to process the large image files efficiently on a high-performance computer cluster. After checking, renaming (if needed), and masking non-tissue material, slideToolkit parses the images into tiles which accommodates parallel analyses on the HPC; these analyses are carried out using CellProfiler (developed by the BROAD Institute) which is reliant on standardized protocols based on training datasets.
This has proven to result in highly reproducible data (Nelissen et al. PLoS One 2014 | PMID: 25541691). Preliminary genetic analyses have also exceeded our expectations, as we are entering the final stages of scanning. A first milestone is the recent preprint reporting on the observational analyses, but there is more to come!