Governance infrastructure · Australian healthcare

Patient identity
never leaves.

The centralised data privacy & governance platform for Australian healthcare. Anonymise, govern and audit every modality before it reaches an AI service.

DICOM PS3.15HL7 / FHIROAIC AlignedAU Data ResidencyTGA SaMD READY
Secuva · live de-identification pipeline
PixelIQ · DICOM CT study
(0010,0010)████████████stripped
(0010,0020)MRN-████████stripped
(0008,0060)CTretained
(0008,103E)CHEST W CONTRASTretained
ClinicalIQ · HL7 ORU message
PID|1||CLINIQ-E7A2||PATIENT^A||196200|F
PV1|1|I|WARD^^^^||||PHYSICIAN REDACTED
OBX|1|NM|WBC||7.2|10*3/uL|4.5-11.0|N
GenomeIQ · VCF file · de-identified2.1s
##SAMPLE=<ID=PSEUDO-2E9F4A>
##INDIVIDUAL=<Name=REDACTED>
#CHROM POS REF ALT PSEUDO-2E9F4A
1 925952 G A 0/1:18,9:27:...
3 modalities processed · 0 raw PHI in transit · audit log written
100%
Australian data residency
Sydney + Melbourne
113+
DICOM PS3.15 attributes
Anonymised by default
Zero
Raw PHI in transit
Removed on-prem
<50ms
Per-study overhead
Inline pipeline
The problem

Healthcare AI is moving fast.
PHI is moving with it.

Every hospital, lab and clinic is being asked to share data with AI vendors, researchers and overseas platforms. DICOM headers, burnt-in pixel text, clinical notes, VCF metadata - all of it carries patient identity. Most de-identification pipelines miss most of it.

PHI leaks at the edges - header tags, pixel overlays, note free-text
Manual de-id doesn't scale across multi-modal AI workflows
Boards and regulators want proof. Most teams can't produce it.
Typical PHI coverage gap by modality
DICOM / Radiology
Header tags
Pixel overlays
Structured reports
Digital Pathology
TIFF metadata
Label image
LIMS barcode
Genomic (VCF / BAM)
Header fields
Sample IDs
Variant re-id risk
Cardiology (ECG / Echo)
Report header
SCP-ECG fields
HL7 wrapper
Clinical (HL7 / FHIR)
Struct. segments
Free-text NLP
FHIR resources
Physiological signals
Device metadata
Session header
Alarm logs
Covered by typical tools Missed by typical tools
Your hospital network · on-prem
PACS / DICOM
EMR / FHIR
Research DB
↓ raw clinical data ↓
SECUVA Layer
anonymise · validate · route · govern
PixelIQSlideIQGenomeIQCardioIQClinicalIQSignalIQ
governed output only · raw PHI never leaves your firewall
SECUVA control plane · AU sovereign cloud
Policy engine
Audit trail
Routing rules
↓ approved recipients only ↓
AI vendor
de-id data only
Researcher
HREC approved
Internal model
within perimeter
The platform

One privacy layer between
your data and every AI service.

SECUVA runs entirely within your environment, under your control, paired with an Australian cloud-hosted control plane. PHI is removed where it lives. Only governed, auditable outputs ever cross your firewall.

Raw PHI never crosses your firewall
Australian data residency, sovereign control plane
Connects to PACS, EMR, FHIR, research systems and AI vendors
Every action cryptographically signed and exportable for audit
How it works

Four steps. Zero raw PHI in transit.

01

Connect

SECUVA agent registers as a DICOM node, HL7 listener, and FHIR proxy inside your network - minimal network configuration required.

02

Anonymise

PHI removed from headers, pixel overlays, free-text fields and structured reports using modality-specific PS3.15-aligned profiles.

03

Govern

Policy engine determines what leaves your perimeter, where it goes, and which AI service is permitted to receive it.

04

Audit

Every transformation cryptographically signed and logged. Exportable for OAIC, TGA, HREC, and internal governance review.

The difference

The day you switch SECUVA on.

Scenario
Without SECUVA
With SECUVA
AI vendor requests DICOM data
Manual export, zip, email - with full patient headers attached
Automated pipeline: de-identified within your environment, then securely routed through approved pathways.
Researcher needs a study cohort
IT project, 3–6 months, manual spreadsheet de-id, no audit trail
Automated cohort build, HREC-exportable audit trail, days not months
OAIC asks what left your network
"We believe it was de-identified" - no records to show
Cryptographic log: what, when, who, which profile - exportable in minutes
New AI vendor onboarding
Legal review cycle, custom de-id pipeline per vendor, inconsistent output
Vendor added to allowlist, same pipeline, same audit standard
Multi-site clinical trial
Each site uses different scripts, inconsistent PHI removal, no central log
Same profile deployed everywhere, consistent pseudonymous IDs, central audit

Patient identity
never leaves.

Show us your data flows - PACS, EMR, research, AI vendors. We will show you exactly where SECUVA fits and what changes the day you switch it on.