When people using a microscope to image fluorescent biological samples, they need to shine light onto the sample and see its fluorescence intensity. But due to the turbidity of a sample, light traveling inside the sample gets scattered and absorbed, causing the light coming out of the sample to be mixed with noise. This produces a very low contrast between the desired signal and the noisy background, which prevents the identification of the target fluorescent structure. Researchers have tried to solve this problem by using a better fluorescent dye or a complicated post-processing algorithm, including machine learning, to retrieve the desired structure. Those methods require huge effort and often fail.
Technology Overview
Northeastern University researchers have developed a method to solve the low contrast problem that needs no modification to the sample and can be performed with no modification to the microscope, with much easier/faster computation. This technology address the low contrast problem by measuring responsivity instead of intensity of the image.
The samples are illuminated with three different intensities of light. A pixel’s responsivity is the change in the fluorescence of the pixel to the different emission intensities. Instead of an intensity, the pixel’s responsivity is used to convey the image. 
This responsivity image can produce eight times better signals to the background contrast than intensity image, which becomes a much clear image with a much lower noise background.
For scanning microscopes, all that is required is software. If it does not have scanning capability, it can essentially take an ordinary microscope and give it the capability of a scanning microscope. 
- Improves signal to background contrast 
- Can be applied to any scanning microscopy without any add‑on or change to their previous setup 
- Can make any microscope behave like a scanning microscope
- The hardware operation and software computation combination method requires a lower computation resource than most existing image enhancement algorithms
- Achieve better neuron image for more precise neuron segmentation and tracing; 
- Image deeper structure in scattering tissues; 
- Improve scanning microscope resolution; 
- Can apply to any signal detection processing, e.g., microwave or x-ray imaging, to get an improved signal to noise contrast
- License
- Partnering
- Research collaboration
Patent Information:
Lab tech/equipment
For Information, Contact:
Mark Saulich
Associate Director of Commercialization
Northeastern University
Samuel Chung
Yao Wang
contrast enhancement
neuron reconstruction
neuron visualization