| 
  • If you are citizen of an European Union member nation, you may not use this service unless you are at least 16 years old.

  • You already know Dokkio is an AI-powered assistant to organize & manage your digital files & messages. Very soon, Dokkio will support Outlook as well as One Drive. Check it out today!

View
 

lsgui Applications

Page history last edited by Max Seligman 3 years, 10 months ago

Home  *  Installation  *  Using lsgui  *  Under the hood  *  Applications  *  Discussion  *  Contact Us

 


Tiling

The flexibility of MATLAB’s environment not only allows for ease of creating applications, but also the ability to create features not seen in other software for lab equipment. One feature that we have added in addition to the basic function of our Lumascope controller is the ability to tile images together. This is a great advantage because, especially at high magnification, the field of view on a microscope is small and moving across a slide can be difficult. This feature allows the user to slowly pan across a slide and create a composite image that gives the user a greater field of view not possible in one image before. It can also be used to generate high resolution images of a subject by allowing the user to use a high magnification to use the camera sensor and pan across the desired field of view to give a greater resolution not possible with a lower magnification objective.  

 

Pseudocolor

Pseudocolor or false color allows us to associate various illumination schemes (or computed images) to the red, green, and blue color planes of displayed images. The first example shows plankton (sampled from Ventura Harbor). A brightfield image is shown in gray (red=green=blue), superimposed in the green color plane is green autofluorescence measured with blue illumination. It is intriguing that what appear to be 3 similar organisms vary greatly in their expression of this fluorophore. This image demonstrates the importance of not only magnification, but contrast in microscopy, and how fluorescence gives a microscope molecular sensitivity independent of its spatial resolving power. 

In this second example, we superpose brightfield (mapped to the red color plane) and green fluorescence (in green color plane), at nearly fluid real-time frame rates. We hope to achieve much higher frame rates by hardware synchronizing flashing LEDs with the camera's shutter. 

 

Focus stacking

Focus stacking increases the low depth of field (DOF) inherent in high magnification imaging, by combining the sharpest pixels from a stack of images taken in adjacent focal planes. On a Lumascope without motorized focus, the user moves the focus knob to one end of the image stack, select Focus Stacking from the Advanced Features menu, and then slowly (to avoid blur) rotates the focus knob. The software records 30 frames in ~1/3 second intervals (10 seconds, in acquireFocusStack.m). It computes a focus metric for each image (infocus.m) and rejects images below a threshold. The remaining images are passed to fstack.m, that computes a focus metric for each pixel of each image, by squaring the pixel's intensity-the average intensity over a 9x9 pixel neighborhood. Pixels of the output image are those with maximum focus metric in the stack. There are several focus stacking implementations posted by MATLAB users on their website. 

To demonstrate focus stacking, we purposely put a commercial (flat) microscope slide (BPAE epithelial cells) askew on the stage, the left side higher than the right. The 4 images on the far left (A) are frames 1, 6, 11, and 16 of a 20 image manually focused z-stack. The center pseudo color image (B) shows the image stack index corresponding to the sharpest pixels -- the right side is sharpest at the beginning of the stack (blues), and the left side at the end (reds). The right frame (C) is the composite image formed by combining the sharpest pixels from each image, reconstructing high resolution across the field of view.

 

Frame averaging and simple math

Wouldn't it be nice if the longer you stare at a field of view, the sharper and less noisy the image (magically) becomes? The default algorithm for Lumaview and most other digital microscopes is to replace the previous image with the next image as soon as the next one is available. But why throw out information in the previous images instead of utilizing it? Especially if the subject is static, then utilizing prior images is easy -- average them and take advantage of the law of large numbers. We are exploring implementing this automatically by cross-correlating successive images, and if the correlations indicate no change in the subject, then replace the image displayed in the main GUI with an average. If done properly, there need be no user interface and no need for user intervention. The longer you stare at a static image, the more the Gaussian noise should attenuate.

 

Synchronizing Lumascope with an Arduino 

Structured illumination offers possibilities to beat the diffraction limit and enhance contrast. We've used an Arduino to control complex illumination sources and synchronize it with the Lumascope, as demonstrated in the following video. 

 

Deep learning

Deep learning allows users to create an interface that would be able to learn from events that it experiences and make decisions based on its learning. This has many useful applications, especially in a lab setting where scientists or researchers are looking for a specific outcome. An example of this is cancer screening. If a deep learning system was in place, it would observe the scientist viewing samples, see what areas are of interest to the scientist, understand the diagnosis received from the doctor, and be a able to view and interpret samples on its own and give the results to the doctor with a great level of confidence. Deep learning systems used to be expensive and difficult to set up but now, there are many options available that are relatively cheap but effective. NVIDIA has a line of products that are small, cheap, and reliable for developing deep learning systems.  One of these devices is small enough to even be placed inside a Lumascope, adding no additional size to the instrument, but allow the lumascope to learn on the job and offer great value to labs all over the world.  

 

Discussion

 

Comments (0)

You don't have permission to comment on this page.