Thrombotherm: Analyzing Blood Platelets with ArrayFire

John MelonakosArrayFire, Case Studies, Image Processing Leave a Comment

The Thrombotherm project by Catalysts is developing a method to analyze blood platelets by means of cell microscopy in real time and to classify them according to their activation state. ArrayFire enabled faster overall research project times and real-time analysis on video data.

This project represents an enormous extension of thrombocyte diagnotics, especially through significantly accelerated analysis times. Faster analyses enabled university research collaborators from the University of Applied Sciences OÖ and the Ludwig Boltzmann Institute to shorten research project times.

The project has three main parts:

  1. Detect cell morphology in real-time

Thombotherm makes it possible to mathematically determine and categorize the cell boundaries by means of transmitted light microscopy. The software distinguishes between “fried-egg”-shaped cells and “spider”-shaped cells. This is used to assess the quality of blood platelets in the blink of an eye.

Fig1

  1. Real-time cluster analysis

By means of a fluorescence micrograph, Thrombotherm identifies cell clusters taking into account the diffraction phenomena that occur. This allows a quantitative real-time representation of the platelets in the present sample.

Fig2

  1. Real-time analysis of the mobility of receptors

Thrombotherm also analyzes real-time image sequences. This allows following single molecules and their motion traces. Pixel resolution is not sufficiently accurate, so exact values are calculated with a point spread model. Complex modeling on real-time image sequences is computationally intensive, and ArrayFire enabled realtime computational performance results.

Complex calculation of moving images in real time enabled with ArrayFire

With the help of ArrayFire, engineers were able to attain dramatic performance improvements on two of the core algorithmic steps for analyzing real-time image sequences.

Engineers achieved a 2451X and a 529X speedup on both algorithms, respectively, when using ArrayFire over their standard CPU implementation. Significant performance improvements were also observed on both algorithms when using ArrayFire over their standard CUDA implementation using custom kernels – 368X and 95X speedups, respectively.

Through this completely new method of analyzing thrombocytes, laboratories can identify the mobility of cells, the compression of surface area markers, the morphology of cells, and their activation condition in real-time.

Leave a Reply

Your email address will not be published. Required fields are marked *