SpotCaliper

Fast Spot Detection with Accurate Size Detection • ImageJ/Fiji plugin

Written by Zsuzsanna Püspöki, John Paul Ward and Daniel Sage at the Biomedical Image Group (BIG), EPFL, Switzerland

Outline

SpotCaliper is a wavelet-based image-analysis software providing a fast automatic detection scheme for circular patterns (spots), combined with the precise estimation of their size. It is implemented as an ImageJ plugin with a complete user interface. The user is allowed to edit the results by modifying the measurements (in a semi-automated way), extract or load data, or do further analysis. The fine tuning of the detections includes the possibility of adjusting or removing the original detections, as well as adding further spots.

Reference

Zsuzsanna Puspoki et al., "SpotCaliper: Fast Wavelet-based Spot Detection with Accurate Size Estimation," Submitted to Bioinformatics Oxford, July 2015.

Video Tutorial

Description

Main features

SpotCaliper is able to automatically find spots of varying size, and gives a precise measurement on their radius. The detections are collected in a table, displaying the following parameters: unique identifier (ID) of the spot, its location (x and y coordinates), radius, confidence, contrast, SNR and type.

1. ID of the spot: Each spot has its own unique identification number. It helps the user to find the corresponding junction on the image.
2. Confidence level: The detection algorithm is based on the strength of wavelet coefficients. When the coefficients have a high value, one can be more certain that the corresponding detections are correct. The confidence level reflects this property.
3. Contrast: The contrast measures the distinction between the object and its background: the higher the difference between the average intensity of the spot and the surrounding concentric neighbourhood, the higher the probability that the detection is correct.
4. SNR: Signal-to-noise ratio. Similarly to the contrast, the SNR characterises the "correctness" of the spots. It gives the ratio between the average of the intensity of the spot (inside the surrounding circle) and the standard deviation of the background (surrounding concentric neighbourhood).
5. Type of the spot: it can be either "Auto" or "Manual + timestamp". "Auto" means that the detections were generated automatically by the software. "Manual" means that the object was added or edited manually by the user, at time "timestamp". We note that the radius of the spot is always determined automatically.

The detected objects are visualized by their outline, covered with a variable opacity disk, or both of them. There is a possibility to label the spots either by their ID, their radius, their confidence level, their contrast or their SNR. The center of the spots are optionally shown as well with a cross.

There is a direct connection between the table and the visualization of the spots. Once the row of the spot is highlighted in the table, the corresponding spot appears automatically with an inverted color. The same way, once a spot is selected on the image, the corresponding row is highlighted too.

There is a possibility to fine tune the given measurements, and also to postprocess the data.

User interaction

1. Edit the range of the detections, changing the lower and upper bounds for the possible radii, location (coordinates in x and y), confidence, contrast and SNR.
2. Define a measure of "correctness" for the detections, and visualise the best N detections with respect to that. The possible measures are confidence, contrast and SNR.
3. Add further spots (double-click or shift + click). The radius of the spot is automatically computed by the software.
4. Remove spots (select spot and "delete" or alt + click).
5. Drag spots. The algorithm helps the user to find the optimal center and the corresponding radius by overlaying the spot on that image.
6. Save the results as a comma-separated text file (.cvs), or load previous measurements.
7. Create a ROI (region of interest) from the measurements, and use the automatic measurement tools of ImageJ to obtain further results.

Software

The software is a plugin running on ImageJ or Fiji. ImageJ is a general purpose image-processing package under public domain licence; it runs on several platforms: Unix, Linux, Windows, Mac OSX.

Installation

1. Get a copy of ImageJ.
2. Download the plugin: Spot_Caliper.jar.
3. Place the file Spot_Caliper.jar.jar in the "plugins" folder of ImageJ. That's it!

Usage

1. Open a grayscale image on ImageJ
2. Start the plugin by choosing SpotCaliper in the menu plugins of ImageJ.
3. Follow the instruction of the graphical user interface

Distribution

• Plugin ready for ImageJ and Fiji: Spot_Caliper.jar
• Java source code: Spot_Caliper-src.zip
• Java classes: Spot_Caliper-cls.zip
• Example of the parameters file: SpotCaliper.txt

Conditions of use

The software is freely available for research purposes. However, it should not be redistributed without the consent of the authors. We expect the user to include a citation of this publication whenever presenting or publishing results that are based on the ImageJ plugin SpotCaliper.

Test images

Real images
Fluorescence microscopy image Click to enlarge	Result of the SpotCaliper Click to enlarge
EliSpot image Click to enlarge	Result of the SpotCaliper Click to enlarge
Control images
For these experiments on control images, the settings of SpotCaliper are the same for the 3 images. No additional manual edition was performed.
Synthetic image, no noise Click to enlarge	Result of the SpotCaliper: 80 detected spots Click to enlarge
Synthetic image with additive gaussian noise Click to enlarge	Result of the SpotCaliper: 80 detected spots Click to enlarge
Synthetic image with a Brownian motion background Click to enlarge	Result of the SpotCaliper: 79 detected spots Click to enlarge