Home/Services/GCP-compliant studies
Research phase 03
GCP-compliant studies
When image analysis data is going into a regulatory submission, the number is only half the deliverable. The other half is the evidence that the number was produced under control — by trained people, following approved procedures, in a validated environment, with every step recorded.
Analysis inside a controlled quality framework
We operate a quality management framework that provides a structured environment and the associated processes to ensure image analysis and reporting activities are performed to written SOPs and against a predefined statement of work. Nothing about a run is improvised.
- Written procedures
Standard operating procedures covering image receipt, QC, annotation, algorithm development, verification, analysis, reporting, archiving and deviation handling.
- Trained and recorded personnel
Role-based training records, documented competency, and a defined approval chain for each activity.
- Validated environment
AI-enabled image analysis runs within a scalable, access-controlled cloud environment that has been built and validated inside the existing quality framework.
- Audit trail
Every analysis run records who ran it, when, on which images, with which algorithm version and which parameters. Records are retained and retrievable.
- Deviation management
Departures from the SOW or the SOPs are logged, assessed for impact, and reported in the study documentation rather than absorbed quietly.
Does your study need a GCP environment?
Yes, generally
Regulated use
Image analysis endpoints that form part of a clinical trial dataset, feed into a submission, support a companion diagnostic claim, or will be inspected.
Usually not
Exploratory use
Internal decision-making, target validation, publication figures and early translational work. Our standard clinical service already carries QC and algorithm locking.
Not sure which side you fall on? Send the protocol section describing the endpoint. It is usually resolvable in one conversation, and moving up a tier later costs more than starting there.
What the study file contains
Doc 01
Statement of work
Approved scope, endpoints, acceptance criteria and delivery schedule, signed by both parties before work starts.
Doc 02
Image analysis plan
The pre-specified method: annotation rules, algorithm approach, verification criteria and the definition of a pass.
Doc 03
Verification record
Ground-truth comparison results, agreement statistics, and the sign-off that locked the algorithm.
Doc 04
QC records
Per-image and per-batch quality control outcomes, including every exclusion and the reason for it.
Doc 05
Deviation log
Any departure from plan, its impact assessment, and the corrective action taken.
Doc 06
Final report
Results, methods, personnel, software versions and environment details, in a form suitable for inclusion in a submission.
Planning a regulated study?
Bring us in while the endpoint is still being written. Pre-specifying the image analysis plan is far cheaper than retrofitting one.