diff --git a/server/__pycache__/helper.cpython-36.pyc b/server/__pycache__/helper.cpython-36.pyc
new file mode 100644
index 0000000..2ea804e
Binary files /dev/null and b/server/__pycache__/helper.cpython-36.pyc differ
diff --git a/server/__pycache__/protocol.cpython-36.pyc b/server/__pycache__/protocol.cpython-36.pyc
new file mode 100644
index 0000000..bd07ad7
Binary files /dev/null and b/server/__pycache__/protocol.cpython-36.pyc differ
diff --git a/server/helper.py b/server/helper.py
index 6601606..16a45e6 100644
--- a/server/helper.py
+++ b/server/helper.py
@@ -13,6 +13,11 @@ def print_matrix(itkmat, size=(3, 3)):
             sys.stdout.write('{} '.format(itkmat(i, j)))
         sys.stdout.write('\n')
 
+def itk_image_dimension(itkimage):
+    return int("".join([char for char in repr(itkimage).split('itkImagePython.')[1].split(';')[0][8:] if char.isnumeric()]))
+def itk_pixel_type(itkimage):
+    return getattr(itk, "".join([char for char in repr(itkimage).split('itkImagePython.')[1].split(';')[0][8:] if char.isupper()]))
+
 # modified from: https://github.com/InsightSoftwareConsortium/itk-jupyter-widgets/blob/master/itkwidgets/trait_types.py#L49
 def _itk_image_to_type(itkimage):
     component_str = repr(itkimage).split('itkImagePython.')[1].split(';')[0][8:]
diff --git a/server/protocol.py b/server/protocol.py
index 895a137..a099a30 100644
--- a/server/protocol.py
+++ b/server/protocol.py
@@ -7,6 +7,7 @@ class Protocol(helper.ObjectProtocol):
     @rpc('median_filter')
     @helper.objdir_wrap
     def median_filter(self, image, radius):
+        print (itk.__path__)
         itk_image = helper.vtkjs_to_itk_image(image)
         
         median_filter = itk.MedianImageFilter[type(itk_image), type(itk_image)].New()
@@ -19,6 +20,23 @@ def median_filter(self, image, radius):
         # TODO auto-serialize in objdir_wrap?
         return helper.itk_to_vtkjs_image(result)
 
+    @rpc('otsu_segment_filter')
+    @helper.objdir_wrap
+    def otsu_segment_filter(self, image):
+        print (itk.__path__)
+        itk_image = helper.vtkjs_to_itk_image(image)
+        
+        otsu_filter = itk.TubeTK.SegmentUsingOtsuThreshold[helper.itk_pixel_type(itk_image), 
+                                              helper.itk_image_dimension(itk_image), 
+                                              helper.itk_pixel_type(itk_image)].New()
+        otsu_filter.SetInput(itk_image)
+        otsu_filter.Update()
+
+        result = otsu_filter.GetOutput()
+
+        # TODO auto-serialize in objdir_wrap?
+        return helper.itk_to_vtkjs_image(result)
+
     @rpc('segment')
     @helper.objdir_wrap
     def segment(self, image, point):