HL7-Based-Nutrition-Clinic-System

Table of Contents

• Introduction
• Overview
• Features
• Dataset
• Technologies
• Demo
• Contributors

Introduction

A Nutrition Clinic Management System that is developed to help nutritionists managing their daily operations. The system is also integrated with a clinical decision support tool for lung cancer classification. This system is developed as a part of the Healthcare Information System course at the Department of Systems & Biomedical Engineering, Cairo University.

Overview

The system mainly consists of:

• Doctor Portal.
• Clincal desicion support module (CDSS).

Features

• Doctor Portal
  • Doctor Login
  • Doctor Registration
  • Doctor operations
    • View all appointments (pending, cancelled and in-progress) appointments.
    • View patient data.
      • Personal data.
      • Clinical data including (allergies & drugs).
      • View & edit some clinical data as lab tests, prescriptions & diet plans.
      • Edit some in-body test parameters as weight, weight control, fats & fat control.
      • Upload new in body test.
      • Referal of the patient to another clinic this was built using HL7 communication.
• Clinical desicion support module
  • Uploading DICOM studies.
  • Doctor Registration
  • Medical viewing.
    • Allow multiple features as Panning, Zooming & Windowing.
    • Classify each slice indepentently into malignant,benign or normal.
    • Provide information about the uploaded study as number of instances Uploaded, study Uid, series Uid and patient name.
    • Show the overall result for the uploaded study.
    • Specifically define the slices including malignant or benign tumors.

Dataset

• Source: Iraq-Oncology Teaching Hospital/National Center for Cancer Diseases, if you want to access the dataset Click here
• Size:
  • 1097 CT scan slices including 561 malignant, 120 bengin & 416 normal slices.
  • 110 cases including 40 malignant, 15 bengin & 55 normal.
• Format: Originally collected in DICOM, but the dataset was available only in JPG format, so we added the metadata and converted it into DICOM.

Technologies

• System

  • Frontend: React with Typescript
  • Backend: NodeJs with Typescript
  • Database: MongoDB
  • CDSS module: Python FastAPI

• Deep learning model

  • Used MobileNet pre-trained model with a Softmax output layer for classification.
  • Model input: DICOM image pixel array
  • Model output: class with the highest probability from Softmax layer

Demo

• CDSS module

Contributors

Submitted to:

Dr. Eman Ayman & Eng. Yara Wael All rights reserved © 2024 to MDIMA team (Systems & Biomedical Engineering, Cairo University Class 2024)