1999-06-14
16:15 at HG D 1.1

Optical Microscopy in the Micron and Submicron Range for (Bio)material Testing

Gaudenz Danuser

ETHZ Biomechanics

Three examples from material testing in metallurgy and biology will be presented where light optical imaging in conjunction with sophisticated computer vision algorithms has unveiled (bio)mechanical phenomena down to the macro-molecular scale. Starting point will be the deformation analysis in micro-rods imaged by simple low resolution optics and reflected light under brightfield illumination. Despite the classic optical resolution of 2 um, deformations of less than 20 nm were quantified with this setup. The second example will deal with the imaging and tracking of micro-tubules in differential interference contrast microscopy. Micro-tubules have a diameter of 25 nm, which is 10 times less than the classic optical resolution of the microscope used. Thermal fluctuations at room temperature were measured with single nanometer sensitivity. From this we derived not only the mechanical rigidity of these biopolymers but analyzed structural weaknesses in the molecular chain. An extension of the measurement procedure to a novel version of polarized light microscopy enabled the visualization of actin traffic in motile cells. What is seen in the image data is actually neither the actin monomer nor the single filament but the intrinsic birefringence of protein bundles of 10 or more filaments. We used these bundles for in vivo probing of the rearward actin flow behind the leading edge of a growing neuron. The common thread between the three examples is the effective fusion of prior knowledge about the imaging procedure and the test sample in order to boost the resolution of light optical microscopy. Structures and processes which are at first sight invisible become quantifiable by using adequate data analysis. This paradigm of model-based light optical microscopy is particularly relevant for the field of material testing since accurate descriptions of the sample characteristics and the imaging properties are inherently available from the testing hypothesis. The talk will conclude in suggesting a revised terminology of resolution. Obviously, the classic bounds which are defined solely by optical diffraction and aberration do not serve in the context of model-based sensing. First attempts in analyzing resolution from a more integral viewpoint will be reported and some novel values for the practical limits in modern microscopy will be given.