5  HL7 & FHIR

5.1 HL7 Overview

HL7 (Health Level Seven) refers to a set of international standards for the exchange of clinical and administrative data between healthcare systems. The “Level Seven” refers to the application layer in the OSI model.

5.1.1 Key HL7 Standards:

HL7 v2.x - The traditional messaging standard using pipe-delimited format HL7 v3 - XML-based standard (less widely adopted) HL7 FHIR - The modern RESTful API-based standard

5.2 FHIR (Fast Healthcare Interoperability Resources)

FHIR is the latest standard from HL7, designed to be:

  • Web-based: Uses modern web technologies (REST, JSON, XML)
  • Resource-oriented: Data organized as modular resources
  • Developer-friendly: Easy to implement compared to older standards

5.2.1 Core FHIR Concepts:

  1. Resources - Basic building blocks (Patient, Observation, DiagnosticReport, etc.)
  2. References - Links between resources
  3. Extensions - Custom data elements
  4. Profiles - Constraints on resources for specific use cases

5.2.2 Radiology-Specific FHIR Resources

For radiology data science, these resources are particularly important:

  • ImagingStudy - Represents a DICOM study
  • DiagnosticReport - Radiology reports
  • Observation - Findings and measurements
  • ServiceRequest - Imaging orders
  • Patient - Patient demographics
  • Practitioner - Radiologists and referring physicians

5.2.3 Practical Example

Let me show you a simple radiology workflow in FHIR:## Key Points for Radiology Data Science

Code 
import json
from datetime import datetime

# Example 1: Patient Resource
patient = {
    "resourceType": "Patient",
    "id": "patient-001",
    "identifier": [{
        "system": "http://hospital.example.org/patients",
        "value": "12345"
    }],
    "name": [{
        "use": "official",
        "family": "Smith",
        "given": ["John", "A."]
    }],
    "gender": "male",
    "birthDate": "1985-03-15"
}

# Example 2: ServiceRequest (Imaging Order)
imaging_order = {
    "resourceType": "ServiceRequest",
    "id": "order-001",
    "status": "active",
    "intent": "order",
    "code": {
        "coding": [{
            "system": "http://loinc.org",
            "code": "36643-5",
            "display": "Chest X-ray PA and lateral"
        }]
    },
    "subject": {
        "reference": "Patient/patient-001"
    },
    "authoredOn": "2024-01-15T09:00:00Z",
    "requester": {
        "reference": "Practitioner/dr-jones"
    },
    "reasonCode": [{
        "text": "Persistent cough for 2 weeks"
    }]
}

# Example 3: ImagingStudy Resource
imaging_study = {
    "resourceType": "ImagingStudy",
    "id": "study-001",
    "status": "available",
    "subject": {
        "reference": "Patient/patient-001"
    },
    "started": "2024-01-15T10:30:00Z",
    "basedOn": [{
        "reference": "ServiceRequest/order-001"
    }],
    "modality": [{
        "system": "http://dicom.nema.org/resources/ontology/DCM",
        "code": "CR",
        "display": "Computed Radiography"
    }],
    "numberOfSeries": 1,
    "numberOfInstances": 2,
    "series": [{
        "uid": "1.2.840.113619.2.55.3.123456.1",
        "modality": {
            "system": "http://dicom.nema.org/resources/ontology/DCM",
            "code": "CR"
        },
        "numberOfInstances": 2,
        "instance": [
            {
                "uid": "1.2.840.113619.2.55.3.123456.1.1",
                "sopClass": {
                    "system": "urn:ietf:rfc:3986",
                    "code": "urn:oid:1.2.840.10008.5.1.4.1.1.1.1"
                },
                "title": "PA view"
            },
            {
                "uid": "1.2.840.113619.2.55.3.123456.1.2",
                "sopClass": {
                    "system": "urn:ietf:rfc:3986",
                    "code": "urn:oid:1.2.840.10008.5.1.4.1.1.1.1"
                },
                "title": "Lateral view"
            }
        ]
    }]
}

# Example 4: DiagnosticReport with Observations
diagnostic_report = {
    "resourceType": "DiagnosticReport",
    "id": "report-001",
    "status": "final",
    "code": {
        "coding": [{
            "system": "http://loinc.org",
            "code": "36643-5",
            "display": "Chest X-ray PA and lateral"
        }]
    },
    "subject": {
        "reference": "Patient/patient-001"
    },
    "issued": "2024-01-15T14:00:00Z",
    "performer": [{
        "reference": "Practitioner/radiologist-001"
    }],
    "imagingStudy": [{
        "reference": "ImagingStudy/study-001"
    }],
    "conclusion": "No acute cardiopulmonary abnormality. Small left pleural effusion.",
    "result": [
        {
            "reference": "Observation/obs-001"
        },
        {
            "reference": "Observation/obs-002"
        }
    ]
}

# Example 5: Observation (Finding)
observation_pleural_effusion = {
    "resourceType": "Observation",
    "id": "obs-001",
    "status": "final",
    "code": {
        "coding": [{
            "system": "http://snomed.info/sct",
            "code": "60046008",
            "display": "Pleural effusion"
        }]
    },
    "subject": {
        "reference": "Patient/patient-001"
    },
    "bodySite": {
        "coding": [{
            "system": "http://snomed.info/sct",
            "code": "264231000",
            "display": "Left lung"
        }]
    },
    "valueCodeableConcept": {
        "coding": [{
            "system": "http://snomed.info/sct",
            "code": "255507004",
            "display": "Small"
        }]
    }
}

# Example 6: Observation (Measurement)
observation_cardiothoracic_ratio = {
    "resourceType": "Observation",
    "id": "obs-002",
    "status": "final",
    "code": {
        "coding": [{
            "system": "http://loinc.org",
            "code": "81269-2",
            "display": "Cardiothoracic ratio"
        }]
    },
    "subject": {
        "reference": "Patient/patient-001"
    },
    "valueQuantity": {
        "value": 0.45,
        "unit": "ratio"
    },
    "referenceRange": [{
        "high": {
            "value": 0.5
        },
        "text": "Normal < 0.5"
    }]
}

# Print examples in readable format
print("=== FHIR Radiology Workflow Example ===\n")

resources = [
    ("1. Patient", patient),
    ("2. Service Request (Order)", imaging_order),
    ("3. Imaging Study", imaging_study),
    ("4. Diagnostic Report", diagnostic_report),
    ("5. Observation - Finding", observation_pleural_effusion),
    ("6. Observation - Measurement", observation_cardiothoracic_ratio)
]

for title, resource in resources:
    print(f"\n{title}:")
    print(json.dumps(resource, indent=2))
    print("-" * 50)
=== FHIR Radiology Workflow Example ===


1. Patient:
{
  "resourceType": "Patient",
  "id": "patient-001",
  "identifier": [
    {
      "system": "http://hospital.example.org/patients",
      "value": "12345"
    }
  ],
  "name": [
    {
      "use": "official",
      "family": "Smith",
      "given": [
        "John",
        "A."
      ]
    }
  ],
  "gender": "male",
  "birthDate": "1985-03-15"
}
--------------------------------------------------

2. Service Request (Order):
{
  "resourceType": "ServiceRequest",
  "id": "order-001",
  "status": "active",
  "intent": "order",
  "code": {
    "coding": [
      {
        "system": "http://loinc.org",
        "code": "36643-5",
        "display": "Chest X-ray PA and lateral"
      }
    ]
  },
  "subject": {
    "reference": "Patient/patient-001"
  },
  "authoredOn": "2024-01-15T09:00:00Z",
  "requester": {
    "reference": "Practitioner/dr-jones"
  },
  "reasonCode": [
    {
      "text": "Persistent cough for 2 weeks"
    }
  ]
}
--------------------------------------------------

3. Imaging Study:
{
  "resourceType": "ImagingStudy",
  "id": "study-001",
  "status": "available",
  "subject": {
    "reference": "Patient/patient-001"
  },
  "started": "2024-01-15T10:30:00Z",
  "basedOn": [
    {
      "reference": "ServiceRequest/order-001"
    }
  ],
  "modality": [
    {
      "system": "http://dicom.nema.org/resources/ontology/DCM",
      "code": "CR",
      "display": "Computed Radiography"
    }
  ],
  "numberOfSeries": 1,
  "numberOfInstances": 2,
  "series": [
    {
      "uid": "1.2.840.113619.2.55.3.123456.1",
      "modality": {
        "system": "http://dicom.nema.org/resources/ontology/DCM",
        "code": "CR"
      },
      "numberOfInstances": 2,
      "instance": [
        {
          "uid": "1.2.840.113619.2.55.3.123456.1.1",
          "sopClass": {
            "system": "urn:ietf:rfc:3986",
            "code": "urn:oid:1.2.840.10008.5.1.4.1.1.1.1"
          },
          "title": "PA view"
        },
        {
          "uid": "1.2.840.113619.2.55.3.123456.1.2",
          "sopClass": {
            "system": "urn:ietf:rfc:3986",
            "code": "urn:oid:1.2.840.10008.5.1.4.1.1.1.1"
          },
          "title": "Lateral view"
        }
      ]
    }
  ]
}
--------------------------------------------------

4. Diagnostic Report:
{
  "resourceType": "DiagnosticReport",
  "id": "report-001",
  "status": "final",
  "code": {
    "coding": [
      {
        "system": "http://loinc.org",
        "code": "36643-5",
        "display": "Chest X-ray PA and lateral"
      }
    ]
  },
  "subject": {
    "reference": "Patient/patient-001"
  },
  "issued": "2024-01-15T14:00:00Z",
  "performer": [
    {
      "reference": "Practitioner/radiologist-001"
    }
  ],
  "imagingStudy": [
    {
      "reference": "ImagingStudy/study-001"
    }
  ],
  "conclusion": "No acute cardiopulmonary abnormality. Small left pleural effusion.",
  "result": [
    {
      "reference": "Observation/obs-001"
    },
    {
      "reference": "Observation/obs-002"
    }
  ]
}
--------------------------------------------------

5. Observation - Finding:
{
  "resourceType": "Observation",
  "id": "obs-001",
  "status": "final",
  "code": {
    "coding": [
      {
        "system": "http://snomed.info/sct",
        "code": "60046008",
        "display": "Pleural effusion"
      }
    ]
  },
  "subject": {
    "reference": "Patient/patient-001"
  },
  "bodySite": {
    "coding": [
      {
        "system": "http://snomed.info/sct",
        "code": "264231000",
        "display": "Left lung"
      }
    ]
  },
  "valueCodeableConcept": {
    "coding": [
      {
        "system": "http://snomed.info/sct",
        "code": "255507004",
        "display": "Small"
      }
    ]
  }
}
--------------------------------------------------

6. Observation - Measurement:
{
  "resourceType": "Observation",
  "id": "obs-002",
  "status": "final",
  "code": {
    "coding": [
      {
        "system": "http://loinc.org",
        "code": "81269-2",
        "display": "Cardiothoracic ratio"
      }
    ]
  },
  "subject": {
    "reference": "Patient/patient-001"
  },
  "valueQuantity": {
    "value": 0.45,
    "unit": "ratio"
  },
  "referenceRange": [
    {
      "high": {
        "value": 0.5
      },
      "text": "Normal < 0.5"
    }
  ]
}
--------------------------------------------------

5.2.3.1 1. DICOM to FHIR Mapping

FHIR ImagingStudy maps to DICOM hierarchy:

  • ImagingStudy → DICOM Study
  • Series → DICOM Series
  • Instance → DICOM Instance (individual images)

5.2.3.2 2. Common Terminologies

FHIR uses standard coding systems:

  • LOINC - For procedure codes
  • SNOMED CT - For anatomical sites, findings
  • RadLex - Radiology-specific terminology
  • DICOM - For modality codes

5.2.3.3 3. RESTful Operations

Basic FHIR operations:

# Create: POST /Patient
# Read: GET /Patient/123
# Update: PUT /Patient/123
# Delete: DELETE /Patient/123
# Search: GET /Patient?name=Smith

5.3 Resource

5.3.1 What is a FHIR Resource?

A Resource is the fundamental unit of data in FHIR. Think of it as a standardized “container” or “building block” that holds specific types of healthcare information. Each resource represents a single concept in healthcare - like a patient, a medication, a lab result, or a radiology report.

Key Characteristics of Resources:

  1. Self-contained - Each resource can stand alone with its own meaning
  2. Identifiable - Has a unique ID and can be referenced by other resources
  3. Standardized structure - Consistent format across all FHIR implementations
  4. RESTful - Can be created, read, updated, deleted via web APIs

Simple Analogy

Think of FHIR resources like LEGO blocks: - Each block (resource) has a specific shape and purpose - You can connect blocks together to build complex structures - Every block follows the same connection system - You can use blocks individually or combine them

5.3.2 Basic Structure of Any Resource

Every FHIR resource has these common elements:

{
    "resourceType": "TypeName",     // What kind of resource
    "id": "unique-identifier",      // Unique ID
    "meta": {...},                  // Metadata (version, last updated)
    // Resource-specific content
}

5.3.3 Simple Example: Patient Resource

Let me show you a minimal Patient resource:

Code 
import json
from datetime import datetime

# Simplest possible Patient resource
simple_patient = {
    "resourceType": "Patient",
    "id": "12345"
}

print("1. Minimal Patient Resource:")
print(json.dumps(simple_patient, indent=2))
print("\n" + "="*50 + "\n")

# More realistic Patient resource
realistic_patient = {
    "resourceType": "Patient",
    "id": "12345",
    "active": True,
    "name": [{
        "family": "Yamamoto",
        "given": ["Hiroshi"]
    }],
    "gender": "male",
    "birthDate": "1980-05-15",
    "address": [{
        "line": ["123 Sukhumvit Road"],
        "city": "Bangkok",
        "country": "Thailand"
    }]
}

print("2. Realistic Patient Resource:")
print(json.dumps(realistic_patient, indent=2))
print("\n" + "="*50 + "\n")

# Example showing how resources reference each other
# An Observation (lab result) that belongs to the patient above
lab_result = {
    "resourceType": "Observation",
    "id": "lab-001",
    "status": "final",
    "code": {
        "text": "Blood Glucose"
    },
    "subject": {
        "reference": "Patient/12345",  # This links to our patient!
        "display": "Hiroshi Yamamoto"
    },
    "valueQuantity": {
        "value": 5.4,
        "unit": "mmol/L"
    },
    "effectiveDateTime": "2024-01-20T08:30:00+07:00"
}

print("3. Observation Resource (linked to Patient):")
print(json.dumps(lab_result, indent=2))
print("\n" + "="*50 + "\n")

# Radiology-specific example: A simple chest X-ray order
xray_order = {
    "resourceType": "ServiceRequest",
    "id": "xray-order-001",
    "status": "active",
    "intent": "order",
    "code": {
        "text": "Chest X-ray PA view"
    },
    "subject": {
        "reference": "Patient/12345"  # Same patient
    },
    "authoredOn": "2024-01-20T09:00:00+07:00",
    "requester": {
        "display": "Dr. Smith"
    }
}

print("4. ServiceRequest Resource (X-ray order):")
print(json.dumps(xray_order, indent=2))

# Demonstrate the relationship
print("\n" + "="*50 + "\n")
print("RELATIONSHIPS:")
print("- The Patient resource (id: 12345) represents Mr. Hiroshi Yamamoto")
print("- The Observation references this patient via 'Patient/12345'")
print("- The ServiceRequest also references the same patient")
print("- This creates a connected graph of healthcare data!")
1. Minimal Patient Resource:
{
  "resourceType": "Patient",
  "id": "12345"
}

==================================================

2. Realistic Patient Resource:
{
  "resourceType": "Patient",
  "id": "12345",
  "active": true,
  "name": [
    {
      "family": "Yamamoto",
      "given": [
        "Hiroshi"
      ]
    }
  ],
  "gender": "male",
  "birthDate": "1980-05-15",
  "address": [
    {
      "line": [
        "123 Sukhumvit Road"
      ],
      "city": "Bangkok",
      "country": "Thailand"
    }
  ]
}

==================================================

3. Observation Resource (linked to Patient):
{
  "resourceType": "Observation",
  "id": "lab-001",
  "status": "final",
  "code": {
    "text": "Blood Glucose"
  },
  "subject": {
    "reference": "Patient/12345",
    "display": "Hiroshi Yamamoto"
  },
  "valueQuantity": {
    "value": 5.4,
    "unit": "mmol/L"
  },
  "effectiveDateTime": "2024-01-20T08:30:00+07:00"
}

==================================================

4. ServiceRequest Resource (X-ray order):
{
  "resourceType": "ServiceRequest",
  "id": "xray-order-001",
  "status": "active",
  "intent": "order",
  "code": {
    "text": "Chest X-ray PA view"
  },
  "subject": {
    "reference": "Patient/12345"
  },
  "authoredOn": "2024-01-20T09:00:00+07:00",
  "requester": {
    "display": "Dr. Smith"
  }
}

==================================================

RELATIONSHIPS:
- The Patient resource (id: 12345) represents Mr. Hiroshi Yamamoto
- The Observation references this patient via 'Patient/12345'
- The ServiceRequest also references the same patient
- This creates a connected graph of healthcare data!

5.4 Resource Types Most Relevant to Radiology

Here are the main resources you’ll work with in radiology:

  1. Patient - Who is being imaged
  2. ServiceRequest - The imaging order/request
  3. ImagingStudy - The actual DICOM study metadata
  4. DiagnosticReport - The radiology report
  5. Observation - Individual findings or measurements
  6. Practitioner - Radiologists, technicians, referring doctors

5.4.1 How Resources Connect

Resources don’t exist in isolation. They reference each other to form a complete picture:

Patient <-- ServiceRequest (order)
   ^              |
   |              v
   |        ImagingStudy (images)
   |              |
   |              v
   +------- DiagnosticReport (report)
                  |
                  v
            Observation(s) (findings)

5.4.2 Why This Matters for Radiology Data Science

Using standardized resources helps you:

  1. Extract consistent data from different hospital systems
  2. Build datasets where a “patient” means the same thing everywhere
  3. Link related information (connect reports to images to findings)
  4. Share AI results in a standard format that any system can understand

The beauty of FHIR resources is that whether you’re at Siriraj Hospital, Ramathibodi, or any hospital worldwide, a “Patient” resource has the same structure, making data science and interoperability much easier!