A method for the automated processing and analysis of images of ULVWF–platelet strings. (May 2013)

• Record Type: Journal Article
• Authors: Reeve, Scott R.; Abbitt, Katherine B.; Cruise, Thomas D.; Hose, D. Rodney; Lawford, Patricia V.
• Is Part Of: Platelets. Volume 24:Number 3(2013:May)
• Page Start: 226
• Page End: 234
• Publication Date: 2013-05
• Subjects: Blood platelets -- Periodicals; Blood Platelets -- Periodicals; 615.39
• DOI: 10.3109/09537104.2012.684732
• Languages: English
• ISSNs: 0953-7104

Abstract

<abstract> <title> <x xml:space="preserve">Abstract</x> </title> <p>We present a method for identifying and analysing unusually large von Willebrand factor (ULVWF)–platelet strings in noisy low-quality images. The method requires relatively inexpensive, non-specialist equipment and allows multiple users to be employed in the capture of images. Images are subsequently enhanced and analysed, using custom-written software to perform the processing tasks. The formation and properties of ULVWF–platelet strings released in <italic>in vitro</italic> flow-based assays have recently become a popular research area. Endothelial cells are incorporated into a flow chamber, chemically stimulated to induce ULVWF release and perfused with isolated platelets which are able to bind to the ULVWF to form strings. The numbers and lengths of the strings released are related to characteristics of the flow. ULVWF–platelet strings are routinely identified by eye from video recordings captured during experiments and analysed manually using basic NIH image software to determine the number of strings and their lengths. This is a laborious, time-consuming task and a single experiment, often consisting of data from four to six dishes of endothelial cells, can take 2 or more days to analyse. The method described here allows analysis of the strings to provide data such as the number and length of strings, number of platelets per string and the distance between each platelet to be found. The software reduces analysis time, and more importantly removes user subjectivity, producing highly reproducible results with an error of less than 2% when compared with detailed manual analysis.</p> </abstract>