Super-resolution imaging of Moving Microbubbles

Friday, May 29, 15.00-17.00 on the ground floor in build 349, rooms 19, 25 and 34.

Purpose:

Through this exercise, students will learn how to generate a super-resolution image from amplitude-modulated data containing a series of diffraction-limited images of moving microbubbles. Students will apply various techniques such as data preparation, noise reduction, microbubble detection and localization, and tracking, and incorporate these results into a high-resolution image.

Preparation:

Read the section on super-resolution imaging in the course notes.

Go through the different exercise points and write down suggestions for your Matlab code.

Data and parameters:

Data location in Dropbox: /CFU_summer_school_2023/exercises/exercise_4/SRI_phantom/

The variables stored in the env_bf.mat file has following descriptions:

 - env: envelope of beamformed data in uint8 format (0~255)
	- [Nz, Nx, Nt] = size(env); % Nz: number of pixels in z-dir, 
	                            % Nx: number of pixels in x-dir, 
	                            % Nt: number of frames
 - metadata
	- metadata.x_axis: 	x values of the beamforming grid (m) - [Nx, 1] vector
	- metadata.z_axis: 	z values of the beamforming grid (m) - [1, Nz] vector
	- metadata.frame_rate: frame rate of the imaging system (Hz)
	- metadata.f0:         center frequency (MHz)
	- metadata.dynamic_range: actual dynamic range of env

% If your system has a low memory try the batch compressed data and load a fraction of frames

Exercise:

1. Preparation of the Data, Noise Reduction by Filtering and Smoothing
   
   Using the provided dataset, perform the following steps:
   
   
   1. Import the dataset and visualize the initial raw data.
   2. Apply a noise reduction technique of your choice (e.g., a median filter or Gaussian smoothing). Compare the images before and after noise reduction.
   3. How do you incorporate the wavelength of the system for noise reduction?


2. Detection of the Microbubbles
   1. Determine an appropriate threshold for detecting the microbubbles in the filtered data. Justify your choice of threshold.
   2. Apply the threshold to the data and visualize the detected microbubbles.


3. Localizing Microbubbles
   1. Choose a method for localizing the microbubbles - either peak detection or centroid. Explain why you chose this method.
   2. Implement the localization technique and visualize the localized microbubbles.


4. Insertion of Localized Positions into a High-Resolution Image
   1. Generate a high-resolution image based on the localized positions of the microbubbles. Consider a pixel size of at least 25 um for the final image.
   2. Discuss any challenges you encountered in this step and how you addressed them.


5. Performing a Simple Tracking of the Detected and Localized Positions (optional)
   1. Implement a basic tracking algorithm for the detected and localized positions (You can use the publicly available code: SimpleTracker)
   2. Visualize the tracking of the microbubbles. Discuss the effectiveness of your tracking algorithm.


6. Insertion of the Tracks into a High-Resolution Image (optional)
   1. Insert the tracks from part 5 into a high-resolution image.
   2. Discuss the final output and any improvements that could be made to your approach.