Do a pull from XiO for as many shapes as are possible -- just do it for the shapes that have been written in as ellipses.
Save them as individual yaml files.
Make a script that converts the current format into yaml. Each cutout having its own file. Do a parametrisation except if it is labelled circle or ellipse. Store x and y coords for all shapes. In the file call width, equivalent_width, and length equivalent_lenght.
Have a script that pulls equivalent widths and equivalent lenghts out of all the individual cutout yaml files, create all the models based on just that.
Write a stripped down version of the electron factors module. Have it just contain the parameterization and the modelling? What about the bokeh plots?
Really, electronfactors should just contain two public functions, parameterise and model. Then there should be a series of workflow scripts that work with the electronfactors code that can be edited to suit the centre.
These workflow scripts can create paramateried yaml files based off dicom, create yaml files based off generic shapes, create a yaml file from XiO/Monaco. Set them up in the "main" method so they are called from bash/cmd with inputs such as "rectangle" etc. And take all yaml files and produce a CSV print overview, another can make a matplotlib print, another a CSV interpolation table, another html interactive report with either bokeh or plotly.
Should have all factors from the model be according to aapm so as to avoid coding confusion.
Refresh the git repo. Archive the old one under another repo. Build it using the python cookbook code.
Create a workflow/scripts directory which contains these scripts. Create a workflow/documentation directory detailing what each workflow script does and how we use them clinically.
Include parameters such as baseline R50, depth, measurement material, electrometer, voltage, chamber, linac model, measurement date.
Do a pull from XiO for as many shapes as are possible -- just do it for the shapes that have been written in as ellipses.
Save them as individual yaml files.
Make a script that converts the current format into yaml. Each cutout having its own file. Do a parametrisation except if it is labelled circle or ellipse. Store x and y coords for all shapes. In the file call width, equivalent_width, and length equivalent_lenght.
Have a script that pulls equivalent widths and equivalent lenghts out of all the individual cutout yaml files, create all the models based on just that.
Write a stripped down version of the electron factors module. Have it just contain the parameterization and the modelling? What about the bokeh plots?
Really, electronfactors should just contain two public functions, parameterise and model. Then there should be a series of workflow scripts that work with the electronfactors code that can be edited to suit the centre.
These workflow scripts can create paramateried yaml files based off dicom, create yaml files based off generic shapes, create a yaml file from XiO/Monaco. Set them up in the "main" method so they are called from bash/cmd with inputs such as "rectangle" etc. And take all yaml files and produce a CSV print overview, another can make a matplotlib print, another a CSV interpolation table, another html interactive report with either bokeh or plotly.
Should have all factors from the model be according to aapm so as to avoid coding confusion.
Refresh the git repo. Archive the old one under another repo. Build it using the python cookbook code.
Create a workflow/scripts directory which contains these scripts. Create a workflow/documentation directory detailing what each workflow script does and how we use them clinically.
Include parameters such as baseline R50, depth, measurement material, electrometer, voltage, chamber, linac model, measurement date.
Remove data that was parameterised by hand.