The method could become central to future research that relies on higher accuracy analysis of the microscopic images produced by today's high-throughput biological screening methods.
In particular, high-throughput screening processes used for drug discovery could benefit the most as pharmaceutical industries look to curb the billion-dollar spending that currently characterises drug development.
The technique, discovered by researchers from the Carnegie Mellon University, comes as an improved form of an existing algorithm. The algorithm, known as the belief propagation algorithm is a popular method used for formulating conclusion about interconnected networks such as the structure of human tissues.
The belief propagation algorithm has become a popular method in the laboratory proving effective in establishing relationships between cells or bits of text. The computer essentially makes deductions about a set of data by drawing on multiple sources of information.
A problem many researchers found was these relationships are formed at the expense of massive amounts of data as well as computational time. What the Carnegie Mellon researchers achieved was to find shortcuts for generating these messages, which significantly improved the speed of the entire network.
Geoffrey Gordon, associate research professor in the School of Computer Science's Machine Learning Department and his fellow authors, biomedical engineering student Shann-Ching "Sam" Chen and computational biologist Robert Murphy, were also able to expand their focus from single to multiple cells by increasing the efficiency of the belief propagation algorithm.
In the case of biological specimens, the speed at which parts of an individual cell image can be established quicker as can the determination of abnormal distributions of particular proteins within each cell.
"Current automated screening systems for examining cell cultures look at individual cells and do not fully consider the relationships between neighbouring cells," said Gordon.
"This is in large part because simultaneously examining many cells with existing methods requires impractical amounts of computational time," he added.
For this paper, published online in the Journal of Machine Learning Research, the scientists applied their methods to look at protein patterns within HeLa cells. They found the technique speeded analysis by several orders of magnitude.
Many current high-throughput screening processes used for drug discovery and other research, tens of thousands of wells, each containing tens or hundreds of cells, need to be analysed each day, a task that is both labour and data intensive.
The researchers commented that improved accuracy could decrease the cost and the time needed for these screening techniques, make possible new experiments that before would've required a massive amount of resources. The prospect of uncovering interesting but subtle anomalies that would've normally gone undetected is entirely feasible too.
Murphy also said that this technique may improve the performance of belief propagation algorithms in various applications, including text analysis, Web analysis and medical diagnosis.