adding dicom example (#16)

* adding dicom example
Signed-off-by: vsoch <vsoch@users.noreply.github.com>

.github/workflows/main.yaml

@@ -44,6 +44,8 @@ jobs:
           python examples/python/tensorflow-function-example.py
           python examples/spack/example-python.py
           python examples/spack/example-singularity.py
+          pip install -r examples/dicom/requirements.txt
+          python examples/dicom/run.py
 
           cd examples/dwarf 
           pip install -r requirements.txt

examples/dicom/README.md

@@ -0,0 +1,52 @@
+# Dicom Changes
+
+This is a quick example to show how to do a diff between dicom headers.
+
+## Usage
+
+This assumes you have a Python environment with compspec installed.
+Install deid:
+
+```bash
+pip install -r requirements.txt
+```
+
+And then run the example
+
+```bash
+$ python run.py
+```
+```
+Result for image1.dcm vs. image4.dcm group 'header'
+{
+    "changed_node_value": [
+        [
+            "A",
+            "B",
+            "id12",
+            "id12",
+            "field-value",
+            "F",
+            "M",
+            "header:image->field:(0010, 0040)->field-name:PatientSex->field-value:F",
+            "header:image->field:(0010, 0040)->field-name:PatientSex->field-value:M"
+        ]
+    ]
+}
+Result for image4.dcm vs. image1.dcm group 'header'
+{
+    "changed_node_value": [
+        [
+            "A",
+            "B",
+            "id12",
+            "id12",
+            "field-value",
+            "M",
+            "F",
+            "header:image->field:(0010, 0040)->field-name:PatientSex->field-value:M",
+            "header:image->field:(0010, 0040)->field-name:PatientSex->field-value:F"
+        ]
+    ]
+}
+```

examples/dicom/model.py

@@ -0,0 +1,38 @@
+__author__ = "Vanessa Sochat"
+__copyright__ = "Copyright 2022, Vanessa Sochat"
+__license__ = "MPL 2.0"
+
+# This is the base model for deriving facts from ast in json
+
+from deid.dicom.parser import DicomParser
+import compspec.graph
+import compspec.solver
+
+
+class DicomGraphs(compspec.graph.GraphGroup):
+    """
+    A namespace to hold more than one graph about dicom files.
+    """
+
+    def __init__(self, dcm, ns):
+        self.dcm = DicomParser(dcm)
+        self.dcm.parse()
+        self.ns = ns
+        super().__init__()
+
+    def extract(self):
+        """
+        Extract named groups into different graphs
+        """
+        # First try just the header
+        g = compspec.graph.Graph()
+        root = g.new_node("header", self.ns)
+
+        for name, field in self.dcm.fields.items():
+            f, _ = g.gen("field", name, parent=root.nodeid)
+            n, _ = g.gen("field-name", field.name, parent=f.nodeid)
+            g.gen("field-value", field.element.value, parent=n.nodeid)
+
+        # Save the named graph - the header is tested first so we can stop
+        # if there are differences.
+        self.graphs["header"] = g

examples/dicom/requirements.txt

@@ -0,0 +1 @@
+deid

examples/dicom/run.py

@@ -0,0 +1,42 @@
+__author__ = "Vanessa Sochat"
+__copyright__ = "Copyright 2022, Vanessa Sochat"
+__license__ = "MPL 2.0"
+
+from compspec.runner import Difference
+from model import DicomGraphs
+
+import os
+import sys
+import json
+
+here = os.path.abspath(os.path.dirname(__file__))
+sys.path.insert(0, here)
+
+
+def main():
+    dicom_dir = os.path.join(here, "lib")
+    dicoms = os.listdir(dicom_dir)
+    graphs = {}
+    for dcm in dicoms:
+        dcm_file = os.path.join(dicom_dir, dcm)
+        if not os.path.exists(dcm_file):
+            sys.exit(f"{dcm_file} does not exist.")
+        graphs[dcm] = DicomGraphs(dcm_file, "image")
+
+    # Run the diff! This is an iterative diff, meaning we are comparing subgraphs
+    for nameA, A in graphs.items():
+        for nameB, B in graphs.items():
+            if nameA == nameB:
+                continue
+            for group, graph in A:
+                if group in B:
+                    gA = A[group]
+                    gB = B[group]
+                    runner = Difference(gA, gB, "A", "B", quiet=True)
+                    result = runner.run()
+                    print(f"Result for {nameA} vs. {nameB} group '{group}'")
+                    print(json.dumps(result, indent=4))
+
+
+if __name__ == "__main__":
+    main()

examples/dwarf/model.py

@@ -491,12 +491,16 @@ def parse_member(self, die):
 
         is_connector = self.is_connector(die)
 
-        node = self.new_node("member", get_name(die), self.ids[die], is_connector=is_connector)
+        node = self.new_node(
+            "member", get_name(die), self.ids[die], is_connector=is_connector
+        )
 
         # Just generate relation for path logic!
-        relation = self.new_relation(fromid=self.ids[die.get_parent()], toid=node.nodeid, relation="has")
+        relation = self.new_relation(
+            fromid=self.ids[die.get_parent()], toid=node.nodeid, relation="has"
+        )
         self.add_relation(relation)
- 
+
         self.gen(
             "type", underlying_type, parent=self.ids[die], is_connector=is_connector
         )

examples/dwarf/tests.py

@@ -30,7 +30,7 @@
 
 # Add remainder
 for name in os.listdir(os.path.join(here, "lib")):
-    if name not in seen and not name.startswith('_'):
+    if name not in seen and not name.startswith("_"):
         tests.append((name, "lib.v1.so", "lib.v2.so"))
     seen.add(name)
 

examples/nifti/model.py

@@ -42,12 +42,12 @@ def extract(self):
         data = self.nii.get_data()
 
         # Since we cannot represent floats, we need to apply scale
-        # Yes this isn't precise, it's an example        
+        # Yes this isn't precise, it's an example
         for x in range(data.shape[0]):
             for y in range(data.shape[1]):
                 for z in range(data.shape[2]):
                     voxel, _ = g.gen("voxel", f"{x}.{y}.{z}", parent=root.nodeid)
-                    g.gen("value", int(data[x,y,z] * 100), parent=voxel.nodeid)
+                    g.gen("value", int(data[x, y, z] * 100), parent=voxel.nodeid)
 
         self.graphs["image"] = g
 

examples/nifti/run.py

@@ -16,7 +16,7 @@
 def main():
     """
     The truth (changed values) is:
-   
+
     x  y  z  zeros modified
     16 16 5  0.0   1.0
     32 32 10 0.0   1e-09
@@ -44,7 +44,9 @@ def main():
             # Omit default logic program for image data
             else:
                 runner = Difference(gA, gB, "zeros", "modified", quiet=True)
-                result = runner.run(logic_programs=['compare-voxels.lp'], omit_default=True)
+                result = runner.run(
+                    logic_programs=["compare-voxels.lp"], omit_default=True
+                )
             print(f"Result for group '{group}'")
             print(json.dumps(result, indent=4))