These examples provide a starting point for issuing http connections and requests to the dosimetry app on the thinknode™ framework. They are provided as is, and are written in python. Any further dependencies are listed along with the provided scripts.

The provided python scripts and libraries are meant to be a foundation and starting point for using the astroid apps on the thinknode™ framework. The provided scripts outline the basics of using ISS to store objects, as well as constructing and making calculation requests to the calculation provider. The below sections detail the basic usage for each script.

Download: The python astroid_script_library can be downloaded from the .decimal GitHub repository.

There is a simple configuration file (thinknode.cfg) that is used to store user data for connecting to the astroid app on the thinknode™ framework. This file is required by all scripts in the python astroid_script_library to authenticate and use the app. A sample file with no user data is available in the repository and the details of the information to include in the file are provided below.

• basic_user being a base64 encoded username and password. Refer to the thinknode documentation for more information.
• api_url being the connection string to the thinknode™ framework.
• apps
  • app_name being the current app name (e.g. dosimetry or dicom).
    • app_version being the current version of the app existing on the thinknode™ framework being used. If left blank the thinknode_worker will select the first app's version returned by the Realm Versions GET request.
    • branch_name not currently implemented
• realm_name thinknode realm
• account_name thinknode account name

thinknode.cfg

{
    "basic_user": "<Base64 encoded thinknode username:password>",
    "api_url": "https://<thinknode_account>.thinknode.io/api/v1.0",
    "apps":
    {   
        "dosimetry": 
        {    
            "app_version": "1.0.0-beta1",
            "branch_name": "master"
        },
        "dicom":
        {
            "app_version": "",
            "branch_name": "master"       
        },
        "rt_types":
        {
            "app_version": "",
            "branch_name": "master"         
        }
    }, 
    "realm_name": "<thinknode realm>",
    "account_name": "<thinknode account>"
}

The post_iss_object_generic.py is a basic python script that provides an example to post any dosimetry type as an immutable object to the dosimetry app on the thinknode™ framework. This example can be used for any immutable storage post using any datatype by replacing the json iss file. The current example posts an rt_study DICOM App datatype object that is read in from the study.json data file.

Dependencies:

• thinknode.cfg
• .decimal Python Libraries
• study.json (or any other prebuilt json file of a dosimetry object as described in the Apps Manifest Guide)

post_iss_object_generic.py

# Copyright (c) 2015 .decimal, Inc. All rights reserved.
# Desc:     Post an immutable json object to the thinknode framework
 
from lib import thinknode_worker as thinknode
import requests
import json
 
iss_dir = "iss_files"
json_iss_file = "study.json"
obj_name = "rt_study"
 
# Get IAM ids
iam = thinknode.authenticate(thinknode.read_config('thinknode.cfg'))
 
# App object to post to iss
with open(iss_dir + '/' + json_iss_file) as data_file:
    json_data = json.load(data_file)
 
# Post immutable object to ISS
res = thinknode.post_immutable_named(iam, "dicom", json_data, obj_name)

Returns:

1. The ID (in json) of the object stored in Immutable Storage.

The post_calc_request_generic.py is a basic example to post a calculation request to dosimetry. This example can be used for any calculation request using any datatype by replacing the calculation request json file. This request will post a calculation request, check the status using long polling with a specified timeout, and return the calculation result.

Dependencies:

• thinknode.cfg
• .decimal Python Libraries
• compute_aperture.json (or any other prebuilt json file of a dosimetry object as described in the Dosimetry Manifest Guide)

post_calc_request_generic.py

# Copyright (c) 2015 .decimal, Inc. All rights reserved.
# Desc:     Post a json calculation request to the thinknode framework
 
request_dir = "request_files"
json_calc_file = "compute_aperture.json"
 
# Get IAM ids
iam = thinknode.authenticate(thinknode.read_config('thinknode.cfg'))
 
# App calculation request
with open(request_dir + '/' + json_calc_file) as data_file:
    json_data = json.load(data_file)
 
# Send calc request and wait for answer
res = thinknode.do_calculation(iam, json_data)
dl.data("Calculation Result: ", str(res))

Returns:

1. The calculation result (in json) of the API function called.

The post_calc_request_sobp_dose.py and post_calc_request_sobp_dose_with_shifter.py are more complete examples that create input data and perform an sobp dose calculation function request to the dosimetry app on the thinknode™ framework.

The post_calc_request_sobp_dose.py example creates the entire calculation request inline using thinknode structure, array, and function requests. The entire dose calculation request is performed using one thinknode calculation provider call. While this structure of a request is a little more complicated to setup and perform, it removes the need to post to ISS or perform ancillary calculations separately.

The post_calc_request_sobp_dose_with_shifter.py adds in the complication of adding a degrader to the sobp calculation. This example performs three separate calculation requests. The first two requests are used to construct the proton degrader_geometry and the third performs the actual dose calculation request using the previously constructed degrader. The entire example could be condensed into a single more complicated thinknode calculation structure, eliminating the need to perform the separate requests, but in some instances it can be more straight-forward to perform some of the calculations separately as shown. As seen in the example, the first two calculation results for the proton degrader are what is placed into the sobp calculation request, instead of the actual function calls as was done in the case of the aperture in the previous example.

Dependencies:

• thinknode.cfg
• .decimal Python Libraries

Below is an abbreviated version of the post_calc_request_sobp_dose_with_shifter.py file. The abbreviated sections are denoted as “…”. In the below sample, the dose_calc variable is a thinknode function request that is made of individually constructed arguments. Notice how some of the elements, like degrader, can be built upon seperate calculation requests.

• Modules used and explanation:
  • The thinknode_worker (thinknode) module is a library that provides worker functions for performing and building the authentication, iss, and calculation requests to the thinknode framework.
  • The dosimetry_worker (dosimetry) module is a library that provides simplified common dosimetry tasks.
  • The decimal_logger (dl) module is a library that provides nicely formatted log output. This includes optional file logging, timestamps, and message coloring (when run through command windows).

Refer to the .decimal Libraries section for more information on the provided decimal libraries.

import json
from lib import thinknode_worker as thinknode
from lib import dosimetry_worker as dosimetry
from lib import decimal_logging as dl
 
# Get IAM ids
iam = thinknode.authenticate(thinknode.read_config('thinknode.cfg'))
 
def make_grid(corner, size, spacing):
...
 
def make_water_phantom(corner, size, spacing):
    return \
        thinknode.function("dosimetry", "create_uniform_image_on_grid_3d",
            [
                make_grid(corner, size, spacing),
                thinknode.value(1),
                thinknode.value("relative_stopping_power")
            ])
 
def make_dose_points(pointCount):
...
 
def make_layers(sad, range, mod):
    return \
        thinknode.function(iam["account_name"], "dosimetry", "compute_double_scattering_layers",
            [
                thinknode.reference("55f70f5000c0a247563a909b6087ada0"), # SOBP Machine from ISS
                thinknode.value(sad),
                thinknode.value(range),
                thinknode.value(mod)
            ])
 
def make_target():
    return \
        thinknode.function("dosimetry", "make_cube",
            [      
                thinknode.value([-32, -20, -30]),
                thinknode.value([16, -10, 30])
            ])
 
def compute_aperture():
    return dosimetry.compute_aperture(iam, make_target(), beam_geometry, 20.0, 0.0, 250.5)
 
beam_geometry = \
...
 
# Get degrader geometry as calculation result
degrade_geom = \
    thinknode.function(iam["account_name"], "dosimetry", "make_shifter", 
        [
            thinknode.value(18), # thickness
            thinknode.value("mm"), # units
            thinknode.value(200) # downstream edge
        ])
res_geom = thinknode.do_calculation(iam, degrade_geom, True)
degrader = \
    thinknode.function(iam["account_name"], "dosimetry", "make_degrader", 
        [
            thinknode.value(res_geom),
            thinknode.reference("56030a9500c036a0c6393f984b25e303") # Material spec from ISS
        ])
proton_degr = thinknode.do_calculation(iam, degrader)
 
# Call compute_sobp_pb_dose2
dose_calc = \
    thinknode.function("dosimetry", "compute_sobp_pb_dose2",
        [
            make_water_phantom([-100, -100, -100], [200, 200, 200], [2, 2, 2]), #stopping_power_image
            thinknode.value(make_dose_points(181)), # dose_points
            beam_geometry, #beam_geometry
            make_grid([-75, -75], [150, 150], [2, 2]), # bixel_grid
            make_layers(2270.0, 152.0, 38.0),
            compute_aperture(), # aperture based on targets
            thinknode.value([proton_degr]) # degraders
        ])
 
# Perform calculation
res = thinknode.do_calculation(iam, dose_calc)
dl.data("Calculation Result: ", res)

The rt_types module is a reconstruction of astroid manifest types in python class format. This includes interdependencies between types (e.g. the class “aperture_creation_params.view” requires the class “multiple_source_view”).

Each data type detailed in the astroid Manifest Guide has a corresponding class in this python module.

Below you will see as snippet from the rt_types module that shows the class for the aperture_creation_params rt_type along with its default initializations and .out function.

• Interdependence: When rt_types are constructed of other or multiple named types, they will be constructed as such in each class as displayed by the view parameter of the aperture_creation_params in this example. The sobp dose calculation sample python script provides an example of this usage in actual practice.
• out function: Each class's .out function provides an ordered dictionary of each of the values in the class. This is explicitly an ordered dictionary since when calling a function in a calculation request, the order of the values provided matters if constructing the request by thinknode value type.

class aperture_creation_params(object):
 
	#Initialize
	def __init__(self):
		self.targets = [] 
		self.target_margin = 0.0 
		self.view = multiple_source_view()
		self.mill_radius = 0.0 
		self.organs = [] 
		self.half_planes = [] 
		self.corner_planes = [] 
		self.centerlines = [] 
		self.overrides = [] 
		self.downstream_edge = 0.0 
 
	def expand_data(self):
		data = {}
		target = []
		for x in self.targets:
			s = triangle_mesh()
			s.from_json(x)
			target.append(s.expand_data())
		data['targets'] = target
		data['target_margin'] = self.target_margin
		data['view'] = self.view.expand_data()
		data['mill_radius'] = self.mill_radius
		organ = []
		for x in self.organs:
			s = aperture_organ()
			s.from_json(x)
			organ.append(s.expand_data())
		data['organs'] = organ
		half_plane = []
		for x in self.half_planes:
			s = aperture_half_plane()
			s.from_json(x)
			half_plane.append(s.expand_data())
		data['half_planes'] = half_plane
		corner_plane = []
		for x in self.corner_planes:
			s = aperture_corner_plane()
			s.from_json(x)
			corner_plane.append(s.expand_data())
		data['corner_planes'] = corner_plane
		centerline = []
		for x in self.centerlines:
			s = aperture_centerline()
			s.from_json(x)
			centerline.append(s.expand_data())
		data['centerlines'] = centerline
		override = []
		for x in self.overrides:
			s = aperture_manual_override()
			s.from_json(x)
			override.append(s.expand_data())
		data['overrides'] = override
		data['downstream_edge'] = self.downstream_edge
		return data
 
	def from_json(self, jdict):
		for k, v in jdict.items():
			if hasattr(self,k):
				if k == 'view':
					self.view.from_json(v)
				else:
					setattr(self, k, v)

The thinknode_worker module is the main work horse for communication with the astroid app and thinknode. The module will handle authentication, posting objects to ISS, creating most of the common calculation request structures, and posting the calculation request.

Refer to the .decimal GitHub repository for the complete module. Below are a few of the more common thinknode http worker and their intended usages:

# Authenticate with thinknode and store necessary ids
# Gets the realm_id, bucket_id, and context_id for the current iam configuration
#   param config: connection settings (url and unique basic user authentication)
def authenticate(config):
 
# Send calculation request to thinknode api
#   param config: connection settings (url, user token, and ids for context and realm)
#   param json_data: calculation request in json format
#   param return_data: True = returns calculation result; False = returns calculation id
def do_calculation(config, json_data, return_data=True):
 
# Post immutable object to ISS
#   param config: connection settings (url, user token, and ids for context and realm)
#   param json_data: immutable object in json format
#   param obj_name: object name of app to post to
def post_immutable(config, json_data, obj_name):
 
# Post immutable object to ISS
#   param config: connection settings (url, user token, and ids for context and realm)
#   param obj_id: thinknode iss reference id for object to get
def get_immutable(config, obj_id):

The decimal_logging module provides formatted and detailed output window and file logging.

The following settings are available in the decimal_logging.py file: display_timestamps: display timestamps in the output window/logfile display_types: display message types (e.g. debug, data, alert) in the output window/logfile log_file: sets the logfile name and location

The following image shows the logging settings for each message type as:

1. Timestamps = True; Types = True
2. Timestamps = False; Types = True
3. Timestamps = False; Types = False