df2dicom Module

This module contains functions used to transform a pandas DataFrame into DICOM and PNG files. It can build a valid DICOM file by obtaining the pixels and DICOM attributes from the DataFrame.

`df2dicom(df, outdir, do_image_deidentification=False, test=False, output_file_formats=None)`

Build DICOM and/or PNG files from a pandas DataFrame obtained with dicom2df.

Parameters:

Name	Type	Description	Default
`df`	`DataFrame`	The pandas DataFrame obtained with `dicom2df` that you want to convert back to DICOM/PNG files.	required
`outdir`	`str`	Path of the directory that will contain your files at the end of the process.	required
`do_image_deidentification`	`bool`	Whether or not this process will trigger the OCR deidentification for pixels.	`False`
`test`	`bool`	This option should not be left to False anytime.	`False`
`output_file_formats`	`list`	A list of formats. Currently supported formats are ["dcm", "png"]. Both can be used alone if you need a single output format. If you select "png", the process will produce a PNG for each line of `df`	`None`

Example

In the following example, we have a directory containing a single DICOM file.

deidentify_attributes_and_pixels.py
from deidcm.dicom.df2dicom import df2dicom
from deidcm.dicom.deid_mammogram import deidentify_attributes

# First part of the deidentification process (Attribute deidentification)
df = deidentify_attributes(indir='/path/to/indir', outdir='/path/to/outdir', org_root='15681.1344456.1444', erase_outdir=False)
df.to_csv(os.path.join(outdir, 'meta.csv'))

# Second part of the deidentification process (OCR deidentification)
df2dicom(df, outdir='/path/to/outdir_folder', do_image_deidentification=True)

Here is an overview of the resulting files in the output directory:

outdir/
├── meta.csv
└── 15681.1344456.1444.150859203650428010901213750509.png

Example

It is also possible to obtain only DICOM files:

only_dcm_files.py
...
df = deidentify_attributes(indir='/path/to/indir', outdir='/path/to/outdir', org_root='15681.1344456.1444', erase_outdir=False, output_file_formats = ["dcm"])
...

Output directory:

outdir/
└── 15681.1344456.1444.150859203650428010901213750509.dcm

it is also possible to obtain DICOM and PNG files:

only_dcm_files.py
...
df = deidentify_attributes(indir='/path/to/indir', outdir='/path/to/outdir', org_root='15681.1344456.1444', erase_outdir=False, output_file_formats = ["dcm", "png"])
df.to_csv(os.path.join(outdir, 'meta.csv'))
df2dicom(df, outdir='/path/to/outdir_folder', do_image_deidentification=True)

Output directory:

outdir/
├── meta.csv
├── 15681.1344456.1444.150859203650428010901213750509.png
└── 15681.1344456.1444.150859203650428010901213750509.dcm