Limits...
High-speed microscopy of continuously moving cell culture vessels

View Article: PubMed Central - PubMed

ABSTRACT

We report a method of high-speed phase contrast and bright field microscopy which permits large cell culture vessels to be scanned at much higher speed (up to 30 times faster) than when conventional methods are used without compromising image quality. The object under investigation moves continuously and is captured using a flash illumination which creates an exposure time short enough to prevent motion blur. During the scan the object always stays in focus due to a novel hardware-autofocus system.

No MeSH data available.


Software-based stitching generating seamless scanning result.To illustrate the benefit of using our stitching approach, we imaged a tilted calibration grid with a line distance of 100 μm at a scanning speed of 60 mm/s using a 4x objective. The result on the left (a) without overlap in the image acquisition and without a post image alignment exhibits visible transitions whereas application of the stitching method ensures an exact fit of all image data (b).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5036042&req=5

f11: Software-based stitching generating seamless scanning result.To illustrate the benefit of using our stitching approach, we imaged a tilted calibration grid with a line distance of 100 μm at a scanning speed of 60 mm/s using a 4x objective. The result on the left (a) without overlap in the image acquisition and without a post image alignment exhibits visible transitions whereas application of the stitching method ensures an exact fit of all image data (b).

Mentions: For performance reasons we compute the feature detection on a graphics card (Nvidia GTX Titan X). With calibrated image data22 there is only a varying translation between neighbouring images that has to be detected in order to correctly align each tile in the image mosaic. Occasional outliers can be filtered out by comparing each translation with the expected range based on the trigger and positioning uncertainty (40 μs*scan velocity). Instead of merely aligning each image with its predecessor we use global optimization inspired by Steckhan et al.23 which takes into consideration all surrounding images in a 4-connected neighbourhood. In a weighted least-squares approach, a sparse overdetermined linear equation system is formed with the goal of minimizing the global alignment error (Fig. 11).


High-speed microscopy of continuously moving cell culture vessels
Software-based stitching generating seamless scanning result.To illustrate the benefit of using our stitching approach, we imaged a tilted calibration grid with a line distance of 100 μm at a scanning speed of 60 mm/s using a 4x objective. The result on the left (a) without overlap in the image acquisition and without a post image alignment exhibits visible transitions whereas application of the stitching method ensures an exact fit of all image data (b).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5036042&req=5

f11: Software-based stitching generating seamless scanning result.To illustrate the benefit of using our stitching approach, we imaged a tilted calibration grid with a line distance of 100 μm at a scanning speed of 60 mm/s using a 4x objective. The result on the left (a) without overlap in the image acquisition and without a post image alignment exhibits visible transitions whereas application of the stitching method ensures an exact fit of all image data (b).
Mentions: For performance reasons we compute the feature detection on a graphics card (Nvidia GTX Titan X). With calibrated image data22 there is only a varying translation between neighbouring images that has to be detected in order to correctly align each tile in the image mosaic. Occasional outliers can be filtered out by comparing each translation with the expected range based on the trigger and positioning uncertainty (40 μs*scan velocity). Instead of merely aligning each image with its predecessor we use global optimization inspired by Steckhan et al.23 which takes into consideration all surrounding images in a 4-connected neighbourhood. In a weighted least-squares approach, a sparse overdetermined linear equation system is formed with the goal of minimizing the global alignment error (Fig. 11).

View Article: PubMed Central - PubMed

ABSTRACT

We report a method of high-speed phase contrast and bright field microscopy which permits large cell culture vessels to be scanned at much higher speed (up to 30 times faster) than when conventional methods are used without compromising image quality. The object under investigation moves continuously and is captured using a flash illumination which creates an exposure time short enough to prevent motion blur. During the scan the object always stays in focus due to a novel hardware-autofocus system.

No MeSH data available.