A RECIST read takes a dozen careful measurements and reports one of four words. That compression has consequences far beyond the imaging report, and none of them is about speed or cost.
A 28% response is not a 30% response, and that gap can decide whether the next round gets funded.
A few millimeters of measurement noise at the 20% progression line can end treatment with a drug that was working.
A misread baseline enrolls a patient who never had measurable disease, and the error surfaces only after dosing.
Tumors in the same patient rarely move together. A summed diameter across five target lesions can hide which lesion is growing, and that is the one that matters.
FDA investigators published a retrospective analysis of 22 randomized trials in metastatic non-small cell lung cancer, finding faster tumor growth associated with shorter overall and progression-free survival. The method is public. Delivering it inside the read is not.
A multi-stakeholder working group’s 2026 white paper names continuous response measures and pseudoprogression as open questions.
Smaller biotechs often contract their imaging directly, and they get the least attention from core labs built for registrational scale.
AI can now find, measure and track lesions, with a radiologist confirming every measurement.
FDA investigators published a retrospective analysis of 22 randomized trials in metastatic non-small cell lung cancer, finding tumor growth rate inversely associated with overall and progression-free survival. The authors concluded it may provide a supplementary approach to assessing clinical activity.
What the FDA Tumor Growth Rate Analysis Does and Does Not Establish →
A multi-stakeholder working group’s 2026 white paper on AI-enabled tumor assessment names continuous response measures and pseudoprogression as open questions.
Published analyses report reader-to-reader disagreement on tumor measurement in a substantial share of cases, which is why trial reads are blinded, central and performed by protocol-trained readers.
What to Require From an Imaging Core Lab Before the First Patient Is Scanned →
ONIX AI is the environment the read happens in, so response analytics, now in development, is being built into the signed read rather than delivered beside it weeks later.
ONIX AI runs today in hospitals, diagnostic centers and teleradiology groups in India: real operating experience, and a cost base built for early-phase work.
Our roadmap came from what oncology sponsors told us costs them, not from technology looking for a use.
A credentialed radiologist signs every read inside a 21 CFR Part 11 audit trail. That is how trust is earned in this market.
DICOM, HL7 v2.x, FHIR and REST. Third-party algorithms plug into the same surface as our own AI models.
We run real clinical workflows on our own platform. Validation documentation available for sponsor review.
Digital pathology capability is available on request and confirmed against a specific protocol.
Central check of measurable disease before enrollment.
A Phase I or II study with a non-registrational endpoint.
Independent review of scans from a completed study.
The core lab for when the number has to mean something.
We will map the imaging plan to them, from eligibility confirmation to the data transfer plan, and come back with a scoped plan and pricing within five business days.
1990 N California Blvd., 8th Floor
Walnut Creek, CA 94596
(844) 922-7468
hello@genphase.ai
What happens next. Send us the protocol and the enrollment timeline. Our clinical team responds within one business day, and you will have a scoped plan and pricing within five business days.
By submitting this form, you agree to let GenPhase use the information provided to send relevant content about our services. You can unsubscribe at any time.