NMJ_Analyser

The code was created using python 3.7 and the following version of packages

Requirements

• scipy 1.5.3
• Pillow 7.2.0
• numpy 1.19.1
• pandas 1.1.0
• nibabel 3.1.1

In addition, the following modules are imported

• glob
• os
• argparse
• sys
• Tinker Performs analysis of NMJ data

Installation

Python3 is required for NMJ-Analyser installation and won't work with Python2

Before attempting to install NMJ-Analyser, do make sure the following packages are installed • SciPy, for multidimensional image processing: https://www.scipy.org/install.html • pillow, for reading images: https://pillow.readthedocs.io/en/stable/installation.html • glob, for finding the pathnames: https://docs.python.org/3/library/glob.html • os: https://pypi.org/project/os-sys/ • numpy: https://numpy.org/install/ • argparse: https://pypi.org/project/argparse/ • pandas: https://pandas.pydata.org/pandas-docs/stable/getting_started/install.html • sys: https://docs.python.org/3/install/ • Scikit-image:https://scikit-image.org/docs/dev/install.html#install-via-pip

With pip3 installed, type pip3 install nmjanalyzer to install the NMJ-Analyzer package

Once the other required packages are installed, type nmj_analyzer on the command line A window appears that will allow for the different parameters to be entered

Usage

NMJ-Analyser requires NMJs to be separated by at least 20um. Images which contain superimposed NMJs are not ideal for analysis. The NMJ Analyser takes as input directories where jpg or png files have been stored

For each subject, the input is presented as SUBJ/*jpg or png The ..... files must contain the keyword red or RED The ..... files must contain the keyword green or GREEN The slices should ordered numerically ex Mouse1_GREEN_0001.jpg.... Mouse1_GREEN_0010.jpg

The following parameters are given to the system:

• -p regular expression of the subject path
• -dx planar resolution
• -dz slice thickness
• -t threshold for voxels to be considered as positives

Output

For each subject the following parameters are calculated for each RED connected component and intersection of GREEN on RED component

RegionProperties

• 'centre of mass': (self.centre_of_mass, ['CoMx', 'CoMy', 'CoMz']),
• 'centre_abs': (self.centre_abs, ['Truex, Truey, Truez']),
• 'volume': (self.volume, ['NVoxels', 'NVolume']),
• 'fragmentation': (self.fragmentation, ['Fragmentation']),
• 'mean_intensity': (self.mean_int, ['MeanIntensity']),
• 'surface': (self.surface, ['NSurface', 'Nfaces_surf', 'NSurf_ext', 'Nfaces_ext']),
• 'surface_dil': (self.surface_dil, ['surf_dil', 'surf_ero']),
• 'surface volume ratio': (self.sav, ['sav_dil', 'sav_ero']),
• 'compactness': (self.compactness, ['CompactNumbDil' ]),
• 'eigen': (self.eigen, ['eigenvalues']),
• 'std': (self.std_values, ['std']),
• 'quantiles': (self.quantile_values, ['quantiles']),
• 'bounds': (self.bounds, ['bounds']),
• 'cc': (self.connect_cc, ['N_CC']),
• 'cc_dist': (self.dist_cc, ['MeanDistCC']),
• 'cc_size': (self.cc_size, ['MinSize', 'MaxSize', 'MeanSize']),
• 'max_extent': (self.max_extent, ['MaxExtent']),
• 'shape_factor': (self.shape_factor, ['ShapeFactor', 'shapefactor_surfcount']),
• 'skeleton_length': (self.skeleton_length, ['SkeletonLength'])

Comparison Properties

• 'green volume': (self.n_pos_ref, 'Volume_(Green)'),
• 'red volume': (self.n_pos_seg, 'Volume_(Red)'),
• 'n_intersection': (self.n_intersection, 'Intersection'),
• 'n_union': (self.n_union, 'Union'),
• 'IoU': Intersection of union
• 'coverage': Overlap
• 'vol_diff': Volume difference
• 'ave_dist': Average distance
• 'haus_dist': Hausdorff distance
• 'haus_dist95': 95% HD
• 'com_dist': distance between centre of mass
• 'com_ref': centre of mass RED
• 'com_seg': centre of mass GREEN