Raccoon - Web-based DICOMWeb & FHIR ImagingStudy Server

English | 繁體中文

---

Raccoon 是使用 no-SQL 資料庫實作的醫學影像儲存系統(DICOMweb PACS)，目前主要由北護影像資訊學實驗室維護。 Raccoon 使用 MongoDB 管理 DICOM 影像並提供 DICOMweb 以及 FHIR ImagingStudy RESTful API 功能進行儲存、查詢、調閱。 另外 Raccoon 使用了 Burni FHIR Server為底延伸出 FHIR 與 DICOM 結合的功能。

---

安裝

• 請注意！在使用 Raccoon 前，務必安裝 MongoDB、ImageMagick
• Raccoon 在 git submodule 額外使用了 BlueLight 作為 Raccoon 的 DICOM Viewer。如果您想擴充 DICOM Viewer的功能請運行 git submodule init 指令 git submodule update 安裝 BlueLight
• Node.js >= 14
• Python >= 3.7

Windows

:yellow_heart::yellow_heart::yellow_heart:

您可以跟隨完整安裝教學的步驟安裝 Raccoon 🎉

此篇教學列出了每項必要軟體的安裝步驟、 Raccoon 的詳細設定、 Raccoon 的部屬以及 Raccoon 的測試方法

😄 對新手來講是一篇非常有好的教學！ 😄

Test OS

• Windows 10 64bits

必要軟體

• MongoDB >= 4.0
• Anaconda (非必要))
• GDCM environment in Anaconda (當您使用Anaconda時，請務必確認環境支援 GDCM)
• imagemagick

安裝依賴

• 進到專案根目錄並使用 cmd 執行以下指令:

npm install
npm run build #此指令會下載 dcmtk 執行檔到 ./models/dcmtk 並產生 dotenv 檔案範例.
pip install -r requirements.txt #安裝 Python 相依性套件

💙💙💙

如果您是 Windows 的使用者，並已經結束安裝的步驟，您可以往下閱讀Configuration & Deploy的說明

Linux

Test OS

• Ubuntu v21.10

必要軟體

• MongoDB >= 4.0
• GDCM

  bash sudo apt-get install python3-gdcm libgdcm3.0

• imagemagick

  bash sudo apt-get install imagemagick #ubuntu

Install dependencies

• 進到專案根目錄並使用 cmd 執行以下指令:

sudo apt-get install dcmtk
npm install
npm run build #This will download dcmtk using apt-get install and generate example dotenv file.
pip install -r requirements.txt #安裝 Python 相依性套件

Troubleshooting

• Unknown VR: Tag not found in data dictionary when using STOW-RS

  • 您必須設定 DCMDICTPATH 環境變數

  • dicom.dic 檔案可以在/usr/share/libdcmtk{version}找到

    {version} 對應到dcmtk的版本, e.g. 3.6.5 => libdcmtk15

  • 使用指令設定 DCMDICTPATH 或者您可以將指令加入到profile檔案中(~/.bashrc,~/.profile etc.), example with dcmtk 3.6.5: sh export DCMDICTPATH=/usr/share/libdcmtk15/dicom.dic

  • 檢查環境變數 sh echo $DCMDICTPATH

---

設定

Server dotenv

• .env 檔案位於專案根目錄
• 您可以參考 .env.template 進行修改

ENV='windows'  #input the os type. enum: windows , linux

MONGODB_NAME="Raccoon" 
MONGODB_HOSTS=["mongodb"]
MONGODB_PORTS=[27017]
MONGODB_USER="user"
MONGODB_PASSWORD="password"
MONGODB_SLAVEMODE=false


SERVER_HOST="localhost"
SERVER_PORT=8081

DICOM_STORE_ROOTPATH="C:/"  #The root path that DICOMweb STOW to store 
DICOMWEB_HOST="localhost" 
DICOMWEB_PORT=8081
DICOMWEB_API="dicom-web"

FHIRSERVER_APIPATH="api/fhir"
FHIRSERVER_HOST="localhost"
FHIRSERVER_PORT=8081
FHIR_NEED_PARSE_PATIENT=true #STOW will generate Patient data using DICOMTag. If you want custom FHIR patient , please change to false.

USE_CONDA=false
CONDA_PATH="C:\\Users\\chinHPlaptop\\anaconda3\\Scripts\\conda.exe"
CONDA_GDCM_ENV_NAME="gdcm"

USE_DCM2JPEG_PYTHONAPI=true
DCM2JPEG_PYTHONAPI_HOST="127.0.0.1"
DCM2JPEG_PYTHONAPI_PORT=5000

前端 web 設定

• 設定檔路經 : public/scripts/config.js
• Change all hostname and port with your server config.

var envConfig = {
    QIDO : {
        hostName :'127.0.0.1' , 
        port : '9090' , 
        api : 'dicom-web' , 
        http : "http" //http or https
    } , 
    WADO : {
        hostName : '127.0.0.1' ,
        port : '9090' , 
        api : 'dicom-web' ,
        http : "http"
    } , 
    FHIR : {
        hostName : '127.0.0.1' , 
        port : '9090' , 
        api : 'api/fhir' , 
        http : "http"
    },
    login: {
        enable: false, // 是否啟動 login 功能
        jwt: false // 是否儲存 token、檢查 token、使用 token 登入
    },
    backend: {
        baseUrl: "http://localhost:8081" // 登入功能的後端前綴網址 (Raccoon)
    }
}

• 使用者介面提供簡單的管理功能管理 Raccoon 內的影像
• For DICOMWeb client, Raccoon can integrate with BlueLight, a lightweight Web-based DICOM Viewer.

部屬

使用 Node.JS

node server

---

使用 docker-compose

• 您可以在專案的根目錄找到 docker-compose 的範例檔

Example:

version: '3.4'
services:
  mongodb:
    image: mongo:4.2
    container_name : mongodb
    restart: always
    ports:
      - 27017:27017
    volumes:
      - ./mongodb/db:/data/db
    environment:
      # provide your credentials here
      - MONGO_INITDB_DATABASE=admin
      - MONGO_INITDB_ROOT_USERNAME=root
      - MONGO_INITDB_ROOT_PASSWORD=Raccoon#Admin2Mongo
      - MONGO_PORT=27017
  raccoon:
    build:
      context : ./
      dockerfile : Dockerfile
    container_name: raccoon
    command: >
      /bin/sh -c '
      while ! nc -z mongodb 27017;
      do
        echo "waiting for database ...";
        sleep 3;
      done;
      echo "db is ready!";
      npm install;
      pm2-runtime start ecosystem.config.js --node-args="--max-old-space-size=4096";
      '
    volumes :
      - ./:/nodejs/raccoon
      - ./raccoon-storage:/dicomFiles
      - ./raccoon-null/node_modules:/nodejs/raccoon/node_modules
      - ./raccoon-null/build:/nodejs/raccoon/build
      - ./raccoon-null/models:/nodejs/raccoon/models/dcmtk/linux-lib
    ports:
      - 8081:8081
    depends_on:
      - mongodb
    tty : true
    restart: on-failure:3
    stdin_open : true

Set-up

docker-compose up -d

---

Usage

DICOMweb

QIDO-RS、WADO-RS、STOW-RS : /dicom-web/studies

WADO-URI : /api/dicom/wado

FHIR

base : /api/fhir/{resource} metadata : /api/fhir/metadata

---

About

• Raccoon支援DICOMWeb標準傳輸協定，包含QIDO-RS, WADO-RS, WADO-URI, STOW等。
• 支援各種Transfer Syntax 以及SOP Class影像
• 通過台灣醫學資訊聯測 MI-TW 2020 - 項目: Track #4 - 醫學影像影像 DICOMWeb Query/Retrieve Source
• 通過台灣醫學資訊聯測 MI-TW 2021 - 項目: Track #6 - 數位病理影像 DICOMWeb Query/Retrieve Source

Key Features

Cybersecurity

• JSON Web Token(JWT) authentication
• ID/Password login

DICOMWeb capabilities

• Raccoon supports the following DICOMWeb API
• metadata (JSON)
• QIDO-RS: (JSON)
• WAOD-RS
• WADO-URI: supported both: DICOM and JPEG
• STOW-RS

FHIR Resources

• Raccoon can act a FHIR server supporting the following FHIR resources and FHIR API which can be found in FHIR metadata services
• patient
• organization
• ImagingStudy
• endpoint
• Note: Raccoon is focused on medical imaging-related resources, not all FHIR resources, built on the top of the Simple-Express-FHIR-Server. If you are interesting in FHIR soultion, please visit Simple-Express-FHIR-Server.

Supported SOP Classes (particular)

Image

• General image storages, e.g., CT, MR, X-ray, etc.
• Multiframe Image Storage - partical support
• Specifal SOP Class: VL Microscopic Image Storage: DICOM WSI

Non-Image

• GSPS, Segementation, SR, etc.
• Supplement 222: Whole Slide Microscopy Bulk Annotations Storage SOP Class

Supported library

• Raccoon DICOM Server uses several open source libraries as following:
• dcmtk use dcm2json to generating DICOM json and use dcmj2pnm to create jpeg image.
• gdcm and pydicom convert DICOM JPEG2000 to jpeg for the retrieve option of the WADO-URI service

Related toolkits

• fhir-mongoose-schema-generator. It can generate the collection's schema in MongoDB from mapping to FHIR resources used in Raccoon automatically.
• Raccoon provides a tool to convert DICOM objects included in a study to a FHIR ImagingStudy resources stored as a MononDB document.
• We hava another FHIR Server, is designed to creating a tiny FHIR server supported specified FHIR resources.

Roadmap

• Dockerize
• DICOM Protocol: C-ECHO SCP, C-STORE SCP, C-MOVE SCP.
• IHE Invoke Image Display (IID) Profile [RAD-106]
• DICOM Supplement 219 - JSON Representation of DICOM Structured Reports. Referenced standard: DICOM Sup 219
• Write a new proposal of FHIR Implementation Guide of Raccoon modified from FHIR ImagingStudy Core Resource which lacks many tags commonly used in PACS.

Acknowledgement

• This project is supported by a grant from the Ministry of Science and Technology Taiwan.
• Thanks 琦雯Queenie, Queenie's github for contributing the logo design.