Malignant and healthy cells display characteristic fractal patterns, which can be used to tell them apart
new approach has given rise to the hope for a faster and more reliable method for determining cancer cell types. Scientists from the Max Planck Institute for Intelligent Systems in Stuttgart and the University of Heidelberg found that cells can be very accurately characterised using fractal geometry. This theory describes objects whose minute structural details resemble their larger contours. Cancer cells are not able to regulate their growth and, as a consequence their shape, as effectively as healthy cells. The particular fractal geometry of a cell therefore becomes a marker of the cell type. Using this mathematical method in combination with sophisticated image recognition, it is possible to establish the progression of cancer in a cell. The researchers studied the statistical distribution of the occurrence of structural details on the surface of different tumour cells, and were thus able to identify cancer cells with more accuracy than when using the conventional immunohistological method. Moreover, they were able to distinguish between different tumours.
Preis für entscheidende Beiträge im Bereich der Neuronalen Netzwerke
Tübingen. Prof. Ph.D. Jan Peters, Leiter des „Robot-Learning-Labs“ am MPI für Intelligente Systeme, ist mit dem 2013 Young Investigator Award der International Neural Network Society ausgezeichnet worden. Peters hat den Preis erhalten für seine entscheidenden Beiträge im Bereich der Neuronalen Netzwerke, insbesondere zur Entwicklung neuer Lernmethoden, die es Robotern erlauben, neue Fähigkeiten zur Bewegung zu lernen.