# Flatness/Symmetry module documentation¶

## Overview¶

The Flatness & Symmetry module (pylinac.flatsym) allows a physicist to check their linac’s beam profile against well-known flatness/symmetry calculation standards, reporting absolute values. Film or EPID images can be loaded in and analyzed. There is one main class FlatSym. If you will be using custom algorithms there are two module dictionaries you will need: FLATNESS_EQUATIONS and SYMMETRY_EQUATIONS.

## Running the Demo¶

To run the demo, import the main class and run the demo method:

from pylinac import FlatSym

FlatSym.run_demo()


Which will also result in the following output:

Flatness & Symmetry
File: C:\Users\jkern\...inac\pylinac\demo_files\flatsym_demo.dcm

Flatness method: Varian
Vertical flatness: 1.931%
Horizontal flatness: 1.857%
Symmetry method: Varian
Vertical symmetry: 2.462%
Horizontal symmetry: 2.989%

Penumbra (80/20):
Horizontal: 2.6mm
Vertical: 2.8mm

Field Size:
Horizontal: 140.9mm
Vertical: 200.2mm

CAX to edge distances:
CAX -> Upper edge: 99.9mm
CAX -> Lower edge: 100.4mm
CAX -> Left edge: 60.4mm
CAX -> Right edge: 80.6mm


## Typical Use¶

In most instances, a physicist is interested in quickly calculating the flatness, symmetry, or both of the image in question. The flatsym module allows you to do this easily, using any of multiple definitions of flatness or symmetry.

To get started, import the FlatSym class:

from pylinac import FlatSym


# from a file
my_file = r"C:/my/QA/folder/img.dcm"
my_img = FlatSym(path=my_file)


If you don’t have an image you can load the demo image:

my_img = FlatSym.from_demo_image()


You can then calculate the flatness and symmetry with the analyze() method:

my_img.analyze(flatness_method='varian', symmetry_method='varian', vert_position=0.5, horiz_position=0.5)


After analysis, the results can be printed, plotted, or saved to a PDF:

print(my_img.results())  # print results
my_img.plot_analyzed_image()  # matplotlib image
my_img.publish_pdf(filename="flatsym.pdf")  # create PDF and save to file


## Raw Data¶

The raw data values of analysis are also available within the public attributes of the class:

my_img = FlatSym.from_demo_image()
my_img.analyze(flatness_method='varian', symmetry_method='varian')
my_img.symmetry['horizontal']['value']  # the actual symmetry value
my_img.flatness['vertical']['value']


## Analysis Options¶

The flatness/symmetry algorithms can be specified as well as the position of the analysis within the image and the width of the profile. See Analysis Definitions for the common algorithms.

my_img.analyze(flatness_method='elekta', symmetry_method='point difference', vert_position=0.4, horiz_position=0.6,
vert_width=0.05, horiz_width=0.05)


You can also create your own algorithms.

## Analysis Definitions¶

Warning

The following definitions are for photons only.

There are multiple definitions for both flatness and symmetry. Your machine vendor uses certain equations, or your clinic may use a specific definition. Pylinac has a number of built-in definitions which you can use.

Symmetry:

Name Parameter values Vendors Equation
Point Difference varian, point difference Varian $$100 * max(|L_{pt} - R_{pt}|)/ D_{CAX}$$ over 80%FW, where $$L_{pt}$$ and $$R_{pt}$$ are equidistant from the CAX.
Point Difference Quotient (IEC) elekta, pdq iec Elekta $$100 * max(|L_{pt}/R_{pt}|, |R_{pt}/L_{pt}|)$$ over 80%FW if 10<FW<30cm [1]
• Parameter value(s), Name, Vendors that use it – Equation
• varian, point difference, Point Difference, Varian$$100 * max(|L_{pt} - R_{pt}|)/ D_{CAX}$$ over 80%FW, where $$L_{pt}$$ and $$R_{pt}$$ are equidistant from the CAX.
• elekta, pdq iec, Point Difference Quotient (IEC), Elekta$$100 * max(|L_{pt}/R_{pt}|, |R_{pt}/L_{pt}|)$$ over 80%FW if 10<FW<30cm [1].

Flatness:

• Parameter value(s), Name, Vendors that use it – Equation
• varian, vom80, siemens, Variation over mean (80%), Varian$$100 * |D_{max} - D_{min}| / (D_{max} + D_{min})$$ within 80%FW.
• elekta, iec, Dmax/Dmin (IEC), Elekta$$100 * D_{max}/D_{min}$$ within 80%FW for 10<FW<30cm [1].

Note

Siemens and other definitions (e.g. Area, Area/2) will be added if the community asks for it.

 [1] (1, 2, 3) The region calculated over actually varies by the following: for 5

## Creating & Using Custom Algorithms¶

### Custom Algorithm Structure¶

The flatness/symmetry algorithms can easily be extended. All algorithms are in a module dictionary, so the required steps are to 1) create the custom analysis function and 2) add it to the equation dictionary.

The custom algorithms must be functions and follow the below structure. All names can be changed, but the structure of one input and correct number and order of outputs is fixed:

def custom_flatness(profile: SingleProfile) -> Sequence[float]:
...
return flatness_value, max_value, min_value, left_edge_index, right_edge_index

def custom_symmetry(profile: SingleProfile) -> Tuple[float, Sequence[float], float, float]:
...
return symmetry_value, symmetry_array, left_edge, right_edge


While most values are easily understood the symmetry_array should be a list of floats or numpy array of the symmetry value at each point, assuming to start at the CAX and move outward.

Note

When creating custom algorithms, use the above for guidance. It 1) must be a function and 2) must have one input argument, which will be a SingleProfile. Depending on whether it is a flatness or symmetry calculation the return values must follow the above structure. For further examples, see the source code for the built-in equations; e.g. flatness_varian().

Note

The SingleProfile given to the function is very powerful and can calculate numerous helpful data for you such as the field edges, minimum/maximum value within the field, and much more. Read the documentation before creating a custom algorithm.

Note

For flatness equations, if there is no max_value or min_value (e.g. area calculations) then just set those values to 0. Additionally, if you do not want or have a symmetry_array to plot, then pass 0 and it will not be plotted.

### Using Custom Algorithms¶

To use the custom algorithm we must now add it to the equation dictionary using a custom name. The name can be anything you like that is a valid dictionary key:

from pylinac import FlatSym
from pylinac.flatsym import SYMMETRY_EQUATIONS, FLATNESS_EQUATIONS

def custom_flatness(profile: SingleProfile):
...

FLATNESS_EQUATIONS['my custom flatness'] = custom_flatness  # no parentheses; we don't want to call it, just assign it

my_img = FlatSym.from_demo_image()
my_img.analyze(flatness_method='my custom flatness', ...)  # use whatever key you defined in the equation dict
...


## Algorithm¶

There is little of a true “algorithm” in flatsym other analyzing profiles. Thus, this section is more terminology and notekeeping.

Allowances

• The image can be any size.
• The image can be digitized film or EPID (most image formats and DICOM).
• The image can be either inversion (Radiation is dark or light).

Restrictions

• The module is only meant for photon analysis at the moment (there are sometimes different equations for electrons for the same definition name).
• Analysis is limited to normal/parallel directions. Thus if the image is rotated there is no way to account for it other than rotating the image before analysis.

Analysis

• Extract profiles - With the positions given, profiles are extracted and analyzed according to the method specified (see Analysis Definitions). For symmetry calculations that operate around the CAX, the CAX must first be determined, which is the center of the FWHM of the profile.

## API Documentation¶

class pylinac.flatsym.FlatSym(path: str, filter: Optional[int] = None)[source]

Class for analyzing the flatness and symmetry of a radiation image, most commonly an open image from a linac.

symmetry

Contains the method of calculation and the vertical and horizontal symmetry data including the value.

Type: dict
flatness

Contains the method of calculation and the vertical and horizontal flatness data including the value.

Type: dict
positions

The position ratio used for analysis for vertical and horizontal.

Type: dict
widths

The width ratios used for analysis for vertical and horizontal.

Type: dict
Parameters: path (str) – The path to the image. filter (None or int) – If None, no filter is applied. If an int, a median filter of size n pixels is applied. Generally, a good idea. Default is None for backwards compatibility.
classmethod from_demo_image()[source]

Load the demo image into an instance

static run_demo()[source]

Run the Flat/Sym demo by loading the demo image, print results, and plot the profiles.

analyze(flatness_method: str, symmetry_method: str, vert_position: float = 0.5, horiz_position: float = 0.5, vert_width: Union[float, int] = 0, horiz_width: Union[float, int] = 0, invert: bool = False)[source]

Analyze the image to determine flatness & symmetry.

Parameters: flatness_method ({'varian', 'elekta', 'vom80', 'siemens', 'iec'}) – The flatness algorithm. See Analysis Definitions for equations. symmetry_method ({'varian', 'elekta', 'point difference', 'pdq iec'}) – The symmetry algorithm. See Analysis Definitions for equations. vert_position (float (0.0-1.0)) – The distance ratio of the image to sample. E.g. at the default of 0.5 the profile is extracted in the middle of the image. 0.0 is at the left edge of the image and 1.0 is at the right edge of the image. horiz_position (float (0.0-1.0)) – The distance ratio of the image to sample. E.g. at the default of 0.5 the profile is extracted in the middle of the image. 0.0 is at the top edge of the image and 1.0 is at the bottom edge of the image. vert_width (float (0.0-1.0)) – The width ratio of the image to sample. E.g. at the default of 0.0 a 1 pixel wide profile is extracted. 0.0 would be 1 pixel wide and 1.0 would be the vertical image width. horiz_width (float (0.0-1.0)) – The width ratio of the image to sample. E.g. at the default of 0.0 a 1 pixel wide profile is extracted. 0.0 would be 1 pixel wide and 1.0 would be the horizontal image width. invert (bool) – Whether to invert the image. Setting this to True will override the default inversion. This is useful if pylinac’s automatic inversion is incorrect.
results(as_str: bool = True) → Union[str, list][source]

Get the results of the analysis.

Parameters: as_str (bool) – If True, return a single string. If False, return a list. Useful for PDF publishing. results str or list
publish_pdf(filename: str, notes: Union[str, list] = None, open_file: bool = False, metadata: dict = None)[source]

Publish (print) a PDF containing the analysis, images, and quantitative results.

Parameters: filename ((str, file-like object}) – The file to write the results to. notes (str, list of strings) – Text; if str, prints single line. If list of strings, each list item is printed on its own line. open_file (bool) – Whether to open the file using the default program after creation. metadata (dict) – Extra data to be passed and shown in the PDF. The key and value will be shown with a colon. E.g. passing {‘Author’: ‘James’, ‘Unit’: ‘TrueBeam’} would result in text in the PDF like: ————– Author: James Unit: TrueBeam ————–
plot_analyzed_image(show: bool = True) → None[source]

Plot the analyzed image. Shows both flatness & symmetry.

Parameters: show (bool) – Whether to show the plot when called.
pylinac.flatsym.flatness_varian(profile: pylinac.core.profile.SingleProfile) → Sequence[float][source]

The Varian specification for calculating flatness

pylinac.flatsym.flatness_elekta(profile: pylinac.core.profile.SingleProfile) → Sequence[float][source]

The Elekta specification for calculating flatness

pylinac.flatsym.symmetry_point_difference(profile: pylinac.core.profile.SingleProfile) → Tuple[float, Sequence[float], float, float][source]

Calculation of symmetry by way of point difference equidistant from the CAX

pylinac.flatsym.symmetry_pdq_iec(profile: pylinac.core.profile.SingleProfile) → Tuple[float, Sequence[float], float, float][source]

Symmetry calculation by way of PDQ IEC