Merge pull request #58 from benjamin-russoniello-pa/57-enh-handle-dic…

…om-files-that-dont-have-pixel-data

Add handling for dicom files without pixel data

dbx/pixels/dicom/dicom_udfs.py

@@ -3,7 +3,7 @@
 import os
 
 from pyspark.sql.functions import udf
-
+from pydicom import Dataset
 
 def cloud_open(path: str, anon: bool = False):
     try:
@@ -22,6 +22,18 @@ def cloud_open(path: str, anon: bool = False):
     except Exception as e:
         raise Exception(f"path: {path} is_anon: {anon} exception: {e} exception.args: {e.args}")
 
+def check_pixel_data(ds: Dataset) -> Dataset | None:
+    """Check if pixel data exists before attempting to access it.
+        pydicom.Dataset.pixel_array will throw an exception if the
+        image contains no pixel data.
+    params:
+        df -- An object of type pydicom.Dataset
+    """
+    try:
+        a = ds.pixel_array
+    except:
+        return None
+    return a
 
 @udf
 def dicom_meta_udf(path: str, deep: bool = True, anon: bool = False) -> dict:
@@ -38,22 +50,27 @@ def dicom_meta_udf(path: str, deep: bool = True, anon: bool = False) -> dict:
 
         fp, fsize = cloud_open(path, anon)
         with dcmread(fp, defer_size=1000, stop_before_pixels=(not deep)) as ds:
+            a = None
             js = ds.to_json_dict()
             # remove binary images
             if "60003000" in js:
                 del js["60003000"]
             if "7FE00010" in js:
                 del js["7FE00010"]
-
             if deep:
+                a = check_pixel_data(ds)
+            if deep and a is not None:
                 a = ds.pixel_array
                 a.flags.writeable = False
+                js["has_pixel"] = True
                 js["hash"] = hashlib.sha1(a).hexdigest()
                 js["img_min"] = np.min(a).item()
                 js["img_max"] = np.max(a).item()
                 js["img_avg"] = np.average(a).item()
                 js["img_shape_x"] = a.shape[0]
                 js["img_shape_y"] = a.shape[1]
+            elif deep:
+                js["has_pixel"] = False
 
             js["file_size"] = fsize
 
@@ -62,4 +79,4 @@ def dicom_meta_udf(path: str, deep: bool = True, anon: bool = False) -> dict:
         except_str = str(
             {"udf": "dicom_meta_udf", "error": str(err), "args": str(err.args), "path": path}
         )
-        return except_str
+        return except_str