Adding ability to specify a filter with REMOVE

CHANGELOG.md

@@ -14,6 +14,7 @@ and **Merged pull requests**. Critical items to know are:
 Referenced versions in headers are tagged on Github, in parentheses are for pypi.
 
 ## [vxx](https://github.com/pydicom/deid/tree/master) (master)
+ - adding filters (contains through missing) for REMOVE (0.1.41)
  - adding support for tag groups (values, fields) (0.1.4)
  - Adding option to provide function to remove (must return boolean) (0.1.38)
  - removing matplotlib version requirement (0.1.37)

deid/config/standards.py

@@ -35,7 +35,7 @@
 groups = ["values", "fields"]
 group_actions = ("FIELD", "SPLIT")
 
-# Valid actions for a filter action
+# Valid actions for a field filter action
 filters = (
     "contains",
     "notcontains",
@@ -45,3 +45,11 @@
     "present",
     "empty",
 )
+
+# valid actions for a value filter
+value_filters = (
+    "contains",
+    "notcontains",
+    "equals",
+    "notequals",
+)

deid/dicom/actions.py

@@ -23,15 +23,16 @@
 """
 
 from deid.logger import bot
-from deid.config.standards import actions as valid_actions
+from deid.config.standards import actions as valid_actions, value_filters
 
-from .fields import expand_field_expression, find_by_values
+from deid.dicom.fields import expand_field_expression, find_by_values
+from deid.dicom.filter import apply_filter
+from deid.dicom.tags import add_tag, update_tag, blank_tag, remove_tag
 
 from deid.utils import get_timestamp, parse_value
 
 from pydicom.dataset import Dataset
 from pydicom.sequence import Sequence
-from .tags import add_tag, update_tag, blank_tag, remove_tag
 
 import re
 
@@ -234,7 +235,7 @@ def _remove_tag(dicom, item, field, value=None):
 
     # The user can optionally provide a function to return a boolean
     if re.search("[:]", value):
-        value_type, value_option = value.split(":")
+        value_type, value_option = value.split(":", 1)
         if value_type.lower() == "func":
 
             # An item must be provided
@@ -255,6 +256,18 @@ def _remove_tag(dicom, item, field, value=None):
                     "function %s returned an invalid type %s. Must be bool."
                     % (value_option, type(do_removal))
                 )
+
+        # A filter such as contains, notcontains, equals, etc.
+        elif value_type.lower() in value_filters:
+
+            # These functions are known to return boolean
+            do_removal = apply_filter(
+                dicom=dicom,
+                field=field,
+                filter_name=value_type,
+                value=value_option or None,
+            )
+
         else:
             bot.exit("%s is an invalid variable type for REMOVE." % value_type)
 

deid/tests/test_dicom_utils.py

@@ -175,6 +175,25 @@ def generate_uid(item, value, field):
         updated = perform_action(dicom=dicom, action=ACTION, item=item)
         self.assertEqual(updated.PatientID, "pancakes")
 
+        # Test each of filters for contains, not contains, equals, etc.
+        dicom = get_dicom(self.dataset)
+
+        print("Testing contains, equals, and empty action with REMOVE")
+        self.assertTrue("ReferringPhysicianName" in dicom)
+        REMOVE = {"action": "REMOVE", "field": "ALL", "value": "contains:Dr."}
+        dicom = perform_action(dicom=dicom, action=REMOVE)
+        self.assertTrue("ReferringPhysicianName" not in dicom)
+
+        self.assertTrue("InstitutionName" in dicom)
+        REMOVE = {"action": "REMOVE", "field": "ALL", "value": "equals:STANFORD"}
+        dicom = perform_action(dicom=dicom, action=REMOVE)
+        self.assertTrue("InstitutionName" not in dicom)
+
+        self.assertTrue("StudyID" in dicom)
+        REMOVE = {"action": "REMOVE", "field": "ALL", "value": "empty"}
+        dicom = perform_action(dicom=dicom, action=REMOVE)
+        self.assertTrue("StudyID" not in dicom)
+
     def test_jitter_timestamp(self):
 
         from deid.dicom.actions import jitter_timestamp

deid/version.py

@@ -22,7 +22,7 @@
 
 """
 
-__version__ = "0.1.4"
+__version__ = "0.1.41"
 AUTHOR = "Vanessa Sochat"
 AUTHOR_EMAIL = "vsochat@stanford.edu"
 NAME = "deid"

docs/_docs/user-docs/recipe-headers.md

@@ -173,6 +173,7 @@ True if this is the case. The dicom is the dicom file (read in with Pydicom) tha
 with (in the example above we grab the `PatientName`).
 
 #### Header
+
 We know that we are dealing with functions relevant to the header of the image 
 by way of the `%header` section. This section can have a series of commands called 
 actions that tell the software how to deal with different fields. For the header 
@@ -224,6 +225,7 @@ JITTER PatientBirthDate -31
 ```
 
 ##### Field Expansion
+
 In some cases, it might be extremely tenuous to list every field ending in the same thing, 
 to perform the same action for. For example:
 
@@ -361,6 +363,34 @@ REPLACE PatientID var:id
 REPLACE InstanceSOPUID var:source_id
 ```
 
+
+##### Value Expansion
+
+These same filters can also be used with any action that is considered a boolean,
+for example, the `REMOVE` tag. As we showed previously, you can remove using
+a filter like "contains" to select some subset of fields:
+
+```
+REMOVE contains:Patient
+```
+
+which would remove all fields that contain "Patient." What if we want to perform
+this same kind of check, but with a value? For example, let's say that we have
+a regular expression to describe a number, and we want to remove any field
+that matches. We could do:
+
+```
+REMOVE ALL contains:(\d{7,0})
+```
+
+Would parse through ALL fields, and remove those that contain a match to the regular
+expression. All supported expanders include:
+
+ - contains
+ - notcontains
+ - equals
+ - notequals 
+
 Now that you know how configuration works, you have a few options.
 You can learn how to define groups of tags based on fields or values in [groups]({{ site.baseurl }}/user-docs/recipe-groups/),
 or if you want to write a text file and get going with cleaning your files, you should