The ImageQLSI class

ImageQLSI is a handle class.

Properties

Public properties
name	type	description
`Microscope`	Microscope	Microscope object
`Illumination`	Illumination	Illumination object
`comment`	char	Any comment on the image

Read-only properties
name	type	description
`T`	double array	Intensity image
`OPD`	double array	OPD image
`DWx`	double array	OPD gradient along x
`DWy`	double array	OPD gradient along y

Dependent properties
name	type	description
`OPDnm`	double array	OPD in nm
`Ph`	double array	Phase image
`Nx`	double	Number of columns
`Ny`	double	Number of rows

Constructor

ImageQLSI methods

alpha_ImageProfile
Return the polarisability, optical volume and dry mass of small objects.

Synthax

% prototype
params = obj.alpha_ImageProfile(Name,Value);

% examples
params = obj.alpha_ImageProfile();
params = obj.alpha_ImageProfile('nmax', 100, 'nBkg', 4);

Description

This method of the ImageEM and ImageQLSI classes returns the polarisability, OV and DM of small objects, using a radial profile method. The algorithm involves a sum of the pixels on a circular area. The method opens a window with the image. First, click on the OPD image to zoom in on the particle of interest, and press ‘z’ when the zoom is correct. Second, click on the center of the particle. A new figure will show up, plotting the pixel summation as a function of the radius of the circular area, from 0 to 100 px. Finally, click two times on the graph to define the range of values corresponding to a proper convergence of the integration. In practice, the line shape should feature a plateau, and the user should click at the beginning and at the end of the plateau. For instance, in this example, the user could click on \(x=18\) and \(x=40\):

Finally, the values of polarisability, and optical volume are returned as a structure, containing the fields alpha, OV and OVw. OVw is the weighted optical volume as defined in Ref. 1.

1: Biomass measurements of single neurites in vitro using optical wavefront microscopy, L. Durdevic, A. Resano Gines, A. Roueff, G. Blivet, G. Baffou, Biomedical Optics Express 13, 6550-6560 (2022)

Name-Value inputs

Several Name-Value inputs can be used to adjust the way the procedure works:

'nmax', default value: 40

Maximum radius of the integration area.
'nBkg', default value: 3

Width of the boundary considered to calculate the zero value of the background.
'NNP', default value: 1

Number of particles to be clicked on the image. The procedure stops after N particles are processed, and the returned data is an array of values.
'zoom', default value: true

Enables the user to first zoom before clicking on the particle
'step', default value: 1

The integration as a function of the radius will be calculated only every N pixels, where N is the step value. Specifying a value larger than 1 can make the processing faster.
'keepPoint', default value: false

Keeps the same clicking point from one image to another. It can save time if many images need to be processed, and if the NP does not move from one image to another.
'display', default value: false

Displays the results of all the measurements in a single graph. Makes sense only if multiple measurements are made within a single call of the function.

flatten
Flatten the background of the OPD image.

The flatten method corrects any artefactual image distorsion, which comes for instance from a mismatch between the reference image and the object image. To flatten the image, this method either removes a blurred image to the image ('Gaussian' mode) or computes the low-order moments of the images and subtract them.

Synthax

% general form
obj.flatten()
obj.flatten(method)
obj.flatten(___, Name, Value)
objList.flatten(___)
obj2 = obj.flatten(___);

% examples
obj.flatten('Zernike')
obj.flatten('Legendre', 'mnax', 3, 'threshold', 1.2);
obj.flatten('Chebyshev', 'kind', 1, 'mnax', 3, 'display', true)
obj.flatten('Gaussian', 'nGauss', 100);
obj.flatten('Legendre', 'nmax', 4, 'mask', booleanMatrix);
obj.flatten('Legendre', 'nmax', 4,  'threshold', 1.2, 'nGauss', 10, 'cut', 0.5, 'invert', true);
obj.flatten('Legendre', 'nmax', 4,  'params', params);

Description

obj.flatten() removes the tilt and coma aberration of the image, by a smoothed image subtraction, with a smoothing parameter nGauss = 100.

obj.flatten(method) removes the tilt and coma aberration of the image, where the image moments belong to a specific class of polynomials. The possible values of the method input are 'Waves', 'Zernike', 'Chebyshev', 'Hermite', 'Legendre'. This is also the option 'Gaussian'. In this latter case, not moment is calculated, and the background correction is obtained by a subtration of a blurred image.

obj.flatten(___, Name, Value) enables the use of optional inputs, defined by keywords 'Name'. The possibles Names are:

'nmax'
Tells until which order the image moments are calculated (inactive if method = 'Gaussian'). For instance,
obj.flatten('Legendre', 'nmax', 2)
removes all the \((n,m)\) Legendre moments from the image such that \(n+m\le n_\mathrm{max}=2\), i.e., (0, 0), (1, 0), (0, 1), (1, 1), (2, 0), (0, 2). Here is a repesentation of the Legendre polynomials. Alternatively, nmax canbe a 2-vector representing the maximum values of \(n\) and \(m\):
obj.flatten('Legendre', 'nmax', [2,2])
removes all the \((n,m)\) Legendre moments from the image such that \(n\le n_\mathrm{max}=2\) and \(m\le m_\mathrm{max}=2\), i.e., (0, 0), (1, 0), (0, 1), (1, 1), (2, 0), (0, 2), (2, 1), (1, 2), (2, 2).
obj.flatten('Zernike', 'nmax', 2)
removes all the \((n,m)\) Zernike moments from the image up to \(n=n_mathrm{max}=2\), i.e., (0, 0), (1, -1), (1, 1), (2, -2), (2, 0), (2, 2). Here is a representation of the Zernike polynomials.
'kind'

Used when using the Chebyshev method. Tells whether Chebyshev polynoms of the 1st or 2nd kind should be used.
'nGauss'

Used when using the Gaussian method. The value is the parameter used with the imgaussfilt function that makes the blurred image to be subtracted. The larger this value, the lesser the flatten effect. 'nGauss = 100' is the default value, and a good starting value.
'threshold'

It is common to observe thick objects within the field of view of the microscope, like eukaryotic cells. Such objects would contribute to the moment computation while they should not. The moment should only consider distorsion of the background. The flatten function can be used to compute the moments only stemming from the background of the image. For this purpose, the code has to determine which part of the image is the background (or reciprocally, which part of the image is the object). For this purpose, a segmentation procedure is used, optimized for the study of eukaryotic cells. The procedure involves a free parameter that is specified using this 'threshold' parameter. A value a 1 is a good starting value.
'mask'

For some samples, the automatic segmentation procudure that is used when setting a value to threshold may be uneffective. In this case, one can specify a mask, as a matrix of the same size as the image, with 0 and 1 values, where the 1 values define to the background area. This mask is to be built by the user itself in the main PhaseLAB file before calling the function flatten. See the Background flattening tutorial for more information on how to create a mask, and use the threshold, nGauss, cut, and invert parameters.
'display'

Equals true or false. If true, the segmentation of the background is displayed, as a means to visually check it is effective.

level0
Offset the OPD image to adjust the background value to zero.

Synthax

obj.level0()
obj.level0(Name, Value)
objList.level0(___)
obj2 = obj.level0(___);
[obj2, params] = obj.level0(___);

Description

obj.level0() adjusts the offset of the image. More precisely. This method asks for an area on the image, the average value of which should be zero. The method applies an offset to the whole image to satifies this condition. Many options can be specified to tell how this area is selected. See below.

Object vectors can also be used with this method. The transformation applies then to all the objects of the vector, with the same reference area.

If an output is used, obj2, then the object is not modified, but duplicated.

If a second output is specified, then the area parameters are returned as a 4-vector params = [x1, x2, y1, y2];

Name-value arguments

Note

Specify optional pairs of arguments as Name1 = Value1, ..., NameN = ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

Example:

obj.level0(___,'Center','Manual','Size',300)

'Center'

With level0(___,'Center','Manual'), the user has to first click on the center of the reference area. If the argument is set to 'Auto', then this step is skipped, and the center is automatically set to the center of the image. Also, the user can indicate the coordinates of the center using level0(___,'Center',[x_c, y_c]).
'Size'

With level0(___,'Size','Manual'), once the center is defined (either manually or automatically), the user has to click on the figure to define the shape of the area, around the center point. The user can also indicate the dimensions of the reference area using level0(___,'Size',Npx) for a square area, or level0(___,'Size',[Nx, Ny]) for a rectangular area.
'twoPoints'

Instead of using the 'Center' and 'Size' keywords, one can also click on two opposite corners of the reference area, using level0(___,'twoPoints',true).
'params'

One can also direclty write the coordinates of the area, using level0(___,'params', [x1, x2, y1, y2]). In this case, no figure opens.

Here is an example of a code that manually ajusts the offset of a first object, and applies automatically the same adjustment to a second object:

[IMc(1) params] = IM(1).level0('Center', 'Manual', 'Size', 'Manual');
IMc(2) = IM(2).level0('params', params);

Note that this code is equivalent to:

IMc = IM.level0('Center', 'Manual', 'Size', 'Manual');

TMPprocess
Computes a temperature image from a wavefront image.

Synthax

objT = obj.TMPprocess(Med)
objT = obj.TMPprocess(Med, Name, Value)
objListT = objList.TMPprocess(___)
[objT, GreenFunction, GreenT_z0] = objList.TMPprocess(___)

Description

obj.TMPprocess() computes the 2D temperature map associated with a wavefront distorsion. See Ref. 3 for more detail.

Objects vectors can also be used with this method. The transformation applies then to all the objects of the vector.

The output objT is an object from the class ImageT.

This method accepts one input parameter, the variable Med from the class MediumT, defining the thermal properties of the surrounding medium (thermal conductivities and dn/dT values). It also accepts Name-value arguments, as listed below.

Optional outputs are the OPD and temperature Green’s functions. They can be reused in subsequent calls of the TMPprocess function to save computation time.

Name-value arguments

Note

Specify optional pairs of arguments as Name1 = Value1, ..., NameN = ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

'g' and 'loop' (default values: 1 and 1)
For large temperature increases, a non-linear algorithm is required to obtain accurate temperature increase maps, as explained in Ref.5. In that case, the 'g' and 'loop' parameters must be specified. Typical values are 0.4 and 10:
objT = TMPprocess(Med, 'g', 0.4, 'loop', 10);
Note that g=1; loop=1 corresponds to the linear algorithm.
'alpha' (default value: 1e-5)

This is the Tikhonov parameter.Keeping it at 1e-5 is usually fine. Higher values tend to smooth the image, and underestimate the temperature increase. Smaller values tend to increase the noise on the temperature image.
'smoothing' (default value: 0, that is no smoothing)

When the OPD image is very noisy, one may want to smooth it. This can be done using the smooth method of the ImageQLSI class, but also using this Name-Value argument here. If different from zero (the default value), this smoothing option applies the imgaussfilt function to the OPD image, the value being the parameter of the imgaussfilt function. The smaller the parameter and the stronger the smoothing.
'imExpander' (default value: true)

This parameter is a boolean. If set to true, it extrapolates the image over a double-size area to avoid artefacts on the boundaries of the reconstructed temperature image. This parameter is true by default, and we recommend to leave it like that, unless the temperature increase is really located at the center of the image.
'T0'

Ambient temperature value, set at 22°C by default.
'zT'

Height at which the temperature should be computed. 0 by default, i.e. the interface between the two media.
'GreenOPD'

Specifies the OPD Green’s function, in case it has aleady been computed before, just to save computation time.
'GreenT_z0'

Specifies the temperature Green’s function, in case it has aleady been computed before, just to save computation time.
'GreenT_3D'

Specifies the 3D temperature Green’s function, in case it has aleady been computed before, just to save computation time.

4: Thermal Imaging of Nanostructures by Quantitative Optical Phase Analysis, G. Baffou et al., ACS Nano 6, 2452 (2012)
5: Three-dimensional temperature imaging around a gold microwire, Applied Physics Letters 102, 244103 (2013)