pygempick 1.1.3

Open Source Batch Gold Particle Picking & Procesing for Immunogold Diagnostics

“””

PyGemPick is the cummulation of Joseph Marsilla’s research project under Dr. Avi Chakrabartty.

This module contains functions that enable:

• TEM Image filtering

• immunogold particle detection

• Spatial Statistical modeling

The main project goal was to greate an open source batch gold particle picking module built in python that could detect gold particles regardless of the amount of counterstaining present in the IGEM (Immunogold Electron Microscopy) micrograph.

This module has three main dependencies that are needed before usage:

1. OpenCV (cv2) for image processing.

2. pandas (pd) for Data analysis

3. numpy (np)

The project will be updated in the upcoming weeks with tutorials on how to use the functions given within pygempick. This package was built to help researchers diagnose pateints with rare protein misfolding diseases like ATTR, AD, FTD and ALS using novel Immunogold diagnostic techniques.

NEW: This update contains supplementary 11 supplementary documents that will help you use the module. We cover image compression, image picking with singular and duplicate filtering, statistical analysis, separation & efficiency tests to test the algorithm’s fairness.

For more information visit the github!

---

Functions:

• ‘compress(orig-img)’: a function that takes an original large scale electron

  micrograph image and compresses it such that 1px = aproximately one nanometer. the exact pixle dimentions for a 3.1x compression are given below - All micrographs in each set were taken at random with a 80Kv Joel 1200 microscope at 150,000x magnification.

• ‘IGEM_filter(p,image, noise)’: is a function that takes a scaling factor value

  and applies the HCLAP filter to the compressed grayscale image produced by compress(orig-img). Output is the filtered binary image. If image was drawn by our IGEM_draw function, add ‘yes’ to the noise section. This applies median filter to neutralize all the speckled noise that the filter doesnt take care of

• ‘DOG_filter(tau,image, noise)’: is a function that takes a scaling factor tau

  and outputs an image from the difference of gaussian method ie output = I*DOG - I*DOG*sqrt(2)*tau

• ‘BIN_filter(p, image)’: is a function that takes a regular grayscale image and

  runs binary thresholding from the difference of the average gray pixel intensity of the micrograph and 60 + 1.5*p , which is the scaling parameter that will allow us to modulate the sepatation power of the binary filtering method

• New: ‘key_filt(keypoints1, keypoints2)’ : Allows you to scan detected keypoints and eliminate duplicates! Allows you to detect partciles with more than one filter. Returns updated keypoints 1 and duplicate(s) detected.

• ‘IGEM_pick(image, minAREA, minCIRC, minCONV, minINER)’: is the function that takes a filtered binary image from BIN_filter(p, image), DOG_filter(tau,image, noise), or IGEM_filter(p,image, noise) and four minfiltering parameters used to detect and filter out keypoints of interest. Values will varry depending on counter staining conditions that are present in a micrograph set.

• ‘IGEM_draw(n, test_number, noise, images, mu1, sig1)’: is a function to draw

  test micrograph sets that will be used in subsequent efficiency tests.

  1. Test number 1 is draw only circles, 2 is draw both circles and ellipses.

  2. Noise if == ‘yes’ then, randomly distibuted gaussian noise will be drawn

     according to mu1, sig1.

  3. images are the number of images in the set - used with n which is number of particles detected in the actual set to calulate the particle density of model set.

• ‘Gamma(r)’: is the window covariogram to solve the boundary problem experienced when

  finding the original Ripley’s K function… Takes radial components as well as the width (a) and height (b) in pixels of the image.

• ‘bin2csv(images)’: - function takes a list of filelocations from glob.glob (asks for the

  filtering parameters),then it outputs a csv of the x and y coordinates of keypoints for every image in images. For example row 1 contains the x coordinate of the keypoints in image 1 and row 2 contains the y coordinates in image 2.

• ‘bin2df(images)’: takes the images and instead of outputing ans saving a csv, it returns

  a pandas dataframe.

• ‘csv2pcf(data, dr)’: - takes the filename data from a csv produced by bin2csv() and outputs

  non-normalized scale invarient k (cross-corelation) and pcf (pair-correlation) statisticaldata from the spatial distribution of the paticles on each micrograph. (determines wheter the nul-hypothesis of CSR [Complete Spatial Randomness] is upheld or voided…)

• ‘imgclass(inv_img)’: - uses a compressed grayscale image from cvt_color(RGB2GRAY)

  returns the im_class and the mean of the pixel intensity histogram of the grayscale image being analyzed…

• ‘septest(p,image)’: - test the separation power of particles detected while changing the

  scaling factor of set filtering method, will return the particles picked and detected for each scaling factor input…if if p = range(1,31,2) it will return two arrays for particles detected at each filter scale. First position of picked lap outlines the number of particles detected when that the first was used on the laplacian filter.

  1 = LAP (Modified Laplacian Filter), 2 = DOG (Normal Difference of Gaussian Filter)

  Note: Binary filter is more of a morphological filtering technique outlined first in 2003 with the algorithm goldfinder (that approach doesn’t use scales but intensity thresholding)

• ‘keypoints2pcf(data_set, dr)’: is a function that takes the output from csv2pcf , propogates

  the error and normalize the resulting pcf data which will be plotted by fitpcf()

• ‘pcf(r, N, p0, p1)’: is the probability distribution of a CSR related process

  that we will used to fit our normalized version of Philmoneko’s PCF diostributtion for calculating colocolization of immunogold particles on microgrpahs

• ‘fitpcf(data)’: is a function that takes the output of keypoints2pcf (CSV file)

  and plots resulting normalized PCF for V30M and CD1 positive controls. needs modification if looking to test distributions for more than one set, but note resolution of calculation and plot will increase with number of images and detected keypoints. That’s why out positive control sets were used because they have an average particle density of 10 - 21 particles per image. Therefore the distribution of the colocolization can be easily calculated.

“””

__docformat__ = ‘restructuredtext’