Back

The Batch-Resourcing Angiogenesis Tool (BRAT) to enable high-content microscopy screening of microvascular networks.

Krek, H.; Murphy, A. R.; McKinnon, R.; Franco, R. A.; Allenby, M. C.

2025-01-26 bioengineering
10.1101/2025.01.24.634836 bioRxiv
Show abstract

Vessel forming assays are a valuable in vitro technology to evaluate the vasculogenic and angiogenic potential of different cell types, matrix proteins, and soluble factors. Recent advances in high-content microscopy allow for vascular morphogenesis assays to be captured in real-time and in high-throughput screening purposes. Unfortunately, existing microvascular network quantification algorithms are either inaccurate, not user-friendly, or manually analyse one image at a time, unfavourable to high-content screening applications. This manuscript introduces BRAT, the Batch-Resourcing Angiogenesis Tool, a computer algorithm with an open-source graphic user interface to efficiently segment, skeletonize, and analyse large batches of vascular network images with high accuracy. Benchmarked across diverse clinical and cultured microvascular network images, BRAT is the most sensitive vascular network image analysis tool (94.5%) and exhibits leading accuracy (93.3%). BRATs multi-threaded processing automatically analysed 886 microscopy images at a speed of 0.17 seconds/image (2:29 minutes) on a performance computer or 2.31 seconds/image (34:04) on a typical laptop. This is 10-to-100 fold more time-efficient than existing tools, which require 12 to 16 seconds of direct user input per image. BRAT is broadly useful to analyse vessel networks of different endothelial cell types cultured on 2D substrates and within 3D biomaterials. BRAT represents a powerful approach for the accurate and high-content screening of vessel forming assays for disease models, regenerative medicines, and therapeutic testing.

Matching journals

The top 11 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.