Supplementary material
A fast and robust method for automated analysis of axonal transport
Oliver Welzel*,†, Jutta Knörr*,‡, Armin M. Stroebel, Johannes Kornhuber and Teja W. Groemer
Department of Psychiatry and Psychotherapy, Friedrich-Alexander University of Erlangen-Nuremberg, Schwabachanlage 6, 91054 Erlangen, Germany
‡ Department of Obstetrics and Gynaecology, Sozialstiftung Bamberg, Buger Straße 80
96049 Bamberg, Germany
* These authors contributed equally to this work
† Correspondence should be addressed to:
Oliver Welzel
Phone: +49 9131 85 44296
Fax: +49 9131 85 36381
E-mail:
Fig. S1. Principle of line detection using the Hough transform. (A) Simulated kymograph image (SNR = 8) showing one particle moving with a speed of -0.5 pixel/frame. The length of the trajectory is 54.12 pixels. (B) Corresponding binary image, which is prerequisite for the later calculation of the Hough transform and was generated using Canny edge detector. (C) Parameter space obtained from the binary image using the Hough transform: ρ = x·cos(θ) + y·sin(θ). The accumulation in this space (green circle) corresponds to the angle θ and Euclidean distance ρ of the line in the binary image. The angle can be used to determine the slope and thus velocity v = tan(-θ·π/180). (D) Overlay of the detected line (green) with its start and end point (red crosses) and the binary image. The reconstructed velocity is -0.49 and the detected length is 53.23 pixels.
Fig. S2. Principle of velocity determination using the Hough transform. (A) Flowchart of the algorithm. (B) Illustrations for the steps in the flowchart.
Fig. S3. Example of creating a binary kymograph image from real data using Canny edge detector. (A) Kymograph image from time-lapse movie (0.5 Hz) of hippocampal axon, synaptophysin labelled with mRFP. (B) Corresponding binary image. After detection of edges using Canny algorithm the center between the two detected edges was determined and resulted in the binary image.
Fig. S4. Examples of binary kymograph image, kymograph image with Gaussian profile and kymograph image with Gaussian profile and Gaussian white noise. Binary image (A), image with Gaussian profile (B) and image with Gaussian profile and Gaussian white noise (C). Corresponding exemplary lineprofiles are shown at the bottom.