Autologous stem cells show great promise for tissue engineering as they can regenerate diseased or damaged tissue, without requiring an organ donor or causing immune rejection. A primary goal of regenerative medicine is to identify methods for controlling stem cell fate so that tissue regeneration can be induced. Recent work has determined that the physical structure of a tissue engineering scaffold can direct stem cell function by driving stem cells into morphologies that induce their differentiation. To support this work we have developed a high-throughput method that used 3D cell shape to detect whether a given cell was grown on a differentiating scaffold or a non-differentiating scaffold. We intersect lines with a 3D confocal image of a stem cell and form a histogram of the intersection lengths. We use these histograms to train a Support Vector Machine (SVM) to recognize differentiating scaffold versus non-differentiating scaffold cells, and then use the trained SVM to classify new query cells. The algorithm is successful in properly classifying the query cells 80\% of the time. The algorithm is easily parallelizable and we demonstrate its implementation on a commodity Graphics Processing Unit (GPU). Use of a GPU to run the algorithm increases throughput by over 100-fold as compared to use of a CPU. Our algorithm is also progressive, providing an approximate answer quickly and refining the answer over time. This allows further increase in the throughput of the algorithm by allowing the SVM classification scheme to terminate early if it becomes sure enough of the class of the cell being analyzed. These results demonstrate a rapid method for classifying stem cells based on their 3D shape that can be used by tissue engineers to identify 3D tissue scaffold structures that induce differentiation and by stem cell biologists to recognize differentiating stem cells.
Citation: Computational Science & Discovery
Pub Type: Journals
Stem cells, parallelization, GPU, geometry, morphology