Scientific Motivation

Light microscopy is a routine imaging technique in bio­logical and medical research and diagnosis. Al­though nowadays instrumentation has made sub­stan­tial progress concerning imaging quality and speed, there is still a gap in resolution between light microscopy (~200 nm) and electron micro­scopy (~10 nm). This so far missing scale range would however open new insights into the nano­cosm of a cell and its sub-cellular structures [NANO1].

Localization nanoscopy, being a candidate to fill this gap, is a novel technique overcoming resolution limits due to diffraction. During the last decade several setups have been developed and used to answer interesting and challenging questions in the field of cellular biology and molecular biomedicine [NANO2].



Figure 1: Histone H2B distribution in HeLa cell nuclei and (pro-)metaphase chromosomes [NANO1].


For localization nanoscopy standard microscopic optics and fast imaging systems are required. The prin­­ciple of the so far developed techniques de­pends on optical isolation and separation of indi­vidual dye molecules by their spectral signature. The embodiment (SPDM = Spectral Position Deter­mination Microscopy) used in our collaboration makes use of dye molecules for specific labeling of cellular sub-structures that are able to undergo so called reversible photo-bleaching which results in stochastic molecular blinking. Taking a huge time stack of images (~1000 frames) the switch off/on of each molecule can be detected and the molecular coordinates can be determined precisely (in the range of nm). Hence, distances between dye molecules can be calculated in the ten nm range and thus sub-cellular structures can be visualized and measured also in 3D conserved cells or even under vital conditions.


In the following two typical examples will be explained: In Figure 1 an example of a cell nucleus and (pro-)metaphase chromosomes are shown. a) – c) show the wide field microscopic images; d) – f) present the merged images from the time stack of SPDM displaying thousands of individual molecules by a color dot. In g) – i) these images are enlarged and coded according to the numbers of next neighbors so that structural infor­mation can be elucidated [NANO3]. Such chromatin 2D/3D nano-structures are of impor­tance to understand chromatin rearrange­ments du­ring repair processes of DNA after exposure to ionizing radiation. This information is used to create and validate a consistent architectural model in the field of radiobiology.

The images of Figure 2 show on the left an overlay of a standard wide-field image (green) and the result of localization imaging (red) of a membrane sec­tion of a breast cancer cell where the Human Epi­­dermal growth factor Receptor 2 (Her2/neu, a typical breast cancer marker) is specifically labeled. The right image shows the result of loca­li­zation imaging which is obtained from a time series of 1000 image frames (979 x 816 pixels, 150 ms per image). Here, each point represents a single fluorochrome respectively antibody attached to a receptor molecule. In contrast to the wide-field image that does not allow to identify any detailed nano-structural information about the spatial arrangement of the anti­bodies/receptors, the resulting localization image reveals details of the formation of receptor clusters or linear arrangements of receptors (inserts) which can be correlated to dimerization induced functional activity [NANO4].

These examples indicate the huge progress going along with localization nanoscopy. However the volume of the data is drastically increasing by orders of magnitudes requiring novel approaches of managing, archiving and analyzing.


From the examples shown above we assume the digital volume of one cell nucleus of about 20 μm diameter with a resolution of approximately 10 nm is about 32 GB per channel of color. In larger screening experiments the limit of one PB data volume is thus reached easily. For the highly sensitive analyses and structure elucidation, very complex and highly variable algorithms have to be used to avoid artifacts and to find out structural re-arrangements. This includes iterative variation based denoising and deblurring techniques. Still the data is saved and worked on in an ad hoc manner, which with serial computation systems leads to extremely long processing times and a limitation of the selectable volume size due to limitations in the computer memory. The data rate created by a nanoscope is in the range of up to GB/s depending on the size of the detected region of interest and the dimensionality (2D/3D) required for scientific investigations.

Actual algorithms and techniques have been developed for a PC basis without usage of techniques for parallelization. This strongly limits the handling of large data sets as being necessary in biological research and medical diagnosis especially if a serious significance of statistics is required (i.e. if a large series of cells have to be evaluated). Here, we develop a pipeline for parallel data analysis. Variation based methods need, with parallel analysis of the data, a synchronous update of all analyzed regions, which will be realized with message passing. The access to the data has to be self-explaining for the user and has to fulfill the rules for storage of the DFG for several years.


Due to the increasing data volume and the complexity of the analyses of nanoscopic image data, new and more advanced techniques for image analysis and storage are needed. The goal for the image acquisition is an isotropic resolution of 10 nm of point patterns, which, in case of a nuclear or cellular diameter of 20 µm results in the magnitude of 32 GB per color channel. The major challenge is to reconstruct the point pattern free of any bias, which is very demanding of computational power with this amount of data. Moreover from such point patterns continuous structures have to be calculated and interpreted.  Screening experiments can easily need data volumes greater than one PB.

The data can be analyzed in parallel, since the information in pictures after the calculation of the positions of the fluorescence molecules, is pro­vided in a grid structure. It is planned to use vari­ation based techniques, which put into effect a-priori-reconstruction, as on the one hand it follows the physical image, and on the other hand also the sparseness as a priori knowledge about the structures is used (images of these structures will reach high compression rates). In this case, an algorithm in parallel will be developed, which yields good performance on computer clusters. The synchronization between the nodes will be realized by message-passing.


Figure 2: Image section of the membrane of a breast cancer cell after specific labeling of the Her2-receptors by means of fluorescence labeled antibodies. (courtesy J. Neumann, Kirchhoff-Institute for Physics, University of Heidelberg).

[NANO1]    Müller, P.; Weiland, Y.; Kaufmann, R.; Gunkel, M.; Hillebrandt, S.; Cremer, C. & Hausmann, M., Analysis of fluorescent nanostructures in biological systems by means of Spectral Position Determination Microscopy (SPDM), In: “Current microscopy contributions to advances in science and technology” (Méndez-Vilas A, ed.), 2012, 1, 3 – 12

[NANO2]   Cremer, C.; Kaufmann, R.; Gunkel, M.; Pres, S.; Weiland, Y.; Müller, P.; Ruckelshausen, T.; Lemmer, P.; Geiger, F.; Degenhard, S. & others, Superresolution imaging of biological nanostructures by spectral precision distance microscopy Biotechnology Journal, Wiley Online Library, 2011, 6, 1037-1051

[NANO3]   Bohn, M.; Diesinger, P.; Kaufmann, R.; Weiland, Y.; Müller, P.; Gunkel, M.; Von Ketteler, A.; Lemmer, P.; Hausmann, M.; Heermann, D. W. & others, Localization microscopy reveals expression-dependent parameters of chromatin nanostructure, Biophysical Journal, Elsevier, 2010, 99, 1358-1367

[NANO4]    Kaufmann, R.; Müller, P.; Hildenbrand, G.; Hausmann, M. & Cremer, C., Analysis of Her2/neu membrane protein clusters in different types of breast cancer cells using localization microscopy, Journal of Microscopy, Wiley Online Library, 2011, 242, 46-54


Uni-HD, KIP: Nick Kepper, Michael Hausmann  

Uni-HD, ERO: Jürgen Hesser      

Copyright by SWM, KIT – Universität des Landes Baden-Württemberg und nationales Forschungszentrum in der Helmholtz-Gemeinschaft
Templates Joomla 1.7 by Wordpress themes free