Why Central Imaging Has Become a Strategic Asset in Clinical Trial

Central imaging in clinical trials is the standardized acquisition, management and independent review of medical images through a centralized Blinded Independent Central Review. It reduces inter-site variability, strengthens endpoint reliability and supports regulatory expectations around blinded independent central review, which turns imaging from a logistics task into a strategic trial asset.

Imaging used to be the part of a protocol that nobody argued about. Sites would scan patients, images would travel somewhere, a radiologist would read them, and a number would appear in the database. The assumption underneath that workflow was that an image is an image.

It is not, and the FDA said so plainly in its Clinical Trial Imaging Endpoint Process Standards guidance, finalized in April 2018. That document treats imaging as a measurement process with its own sources of bias and variability, and it asks sponsors to control them deliberately. Once imaging carries a primary or key secondary endpoint, the way images are acquired, transferred, read and archived stops being an operational detail. It becomes part of the evidence.

What Central Imaging Means in Clinical Trials

Central imaging clinical trials rely on centralized management of the entire imaging chain: standardizing how images are acquired at each site, collecting and quality-controlling them in one place, having them interpreted by independent readers under a documented methodology, and delivering endpoint data in a form that survives regulatory review.

The Blinded Independent Central Review is the entity that owns that chain. It sits outside the sites and outside the sponsor, which is precisely what gives its output evidentiary value.

Central reads vs site/local reads

A local read is performed by the radiologist at the investigational site, usually the same person involved in the patient's care. Fast, clinically informed, unblinded. A central read is performed by an independent reader who sees the images and nothing else: no treatment assignment, often no clinical narrative, only what the charter allows.

Local reads are not inferior medicine. They answer a different question. A treating radiologist asks what is happening to this patient. A central reader asks whether this image, measured the same way as several thousand others, meets a predefined response definition. Trials need the second answer, reproducibly.

Why Central Imaging Became Strategic: Four Drivers

Imaging as a primary or key secondary endpoint

In oncology, progression is defined by imaging. In neurology, lesion load and atrophy carry the story. In respiratory research, structural change on CT is increasingly used alongside spirometry. When the endpoint itself is an image-derived measurement, the quality of the imaging process is not supporting evidence. It is the evidence.

Multicenter variability

Forty sites means forty scanner configurations and forty radiologists with their own habits. Slice thickness, contrast timing, field of view and reconstruction algorithm all move the measurement. None of that variation relates to the drug, but all of it lands in the same dataset.

Centralization does not remove the differences between scanners. It constrains them in advance through the imaging manual, then detects the remaining deviations at quality control instead of at database lock.

Regulatory scrutiny and blinded independent central review

The 2018 FDA guidance is more nuanced than the industry shorthand suggests. It states that centralized interpretation is not always critical, even for a Phase III trial, when the measurement is quantitative, widely performed and low in variability. It becomes important when a trial is subject to bias and cannot be blinded, or when it relies on specialized imaging measures.

That nuance is worth reading carefully, because it defines where blinded independent central review actually earns its cost. An open-label study with a subjective imaging endpoint is the textbook case. A double-blind study with a simple quantitative measurement may not be.

Imaging biomarkers and new modalities

Response criteria have multiplied well beyond RECIST 1.1: iRECIST for immunotherapy, Lugano and LYRIC in lymphoma, PERCIST for PET-based response, RANO in neuro-oncology, PCWG3 in prostate cancer. Quantitative imaging biomarkers add another layer, from tumor volumetrics to regional lung measurements and dosimetry in radioligand therapy.

Each requires readers trained on that specific framework. Generic radiology experience does not transfer automatically, and a mismatch between the protocol criteria and the reader panel is one of the quieter ways a program loses time.

The Blinded Independent Central Review Difference

What separates a Blind Independent Central Review from an imaging vendor is that it takes responsibility for the measurement, not the file transfer. Acquisition is standardized before the first patient is scanned, reading is independent and documented, and quality control runs continuously rather than at the end.

A medical imaging provider typically delivers:

• An imaging manual covering site qualification, equipment settings, patient preparation and acquisition timing

• An imaging review charter defining the read design, reader qualification, reading methodology and adjudication rules

• Site training and qualification, including phantom imaging where the guidance calls for it

• De-identification, DICOM management and a validated transfer and archiving path

• Blinded independent reads with documented reader training and periodic blinded re-testing

• Adjudication of discordant reads under predefined rules

• Query management, reconciliation with the clinical database and submission-ready endpoint datasets

Where Central Imaging Creates the Most Value

Not every trial needs a central read, and saying so is more useful than pretending otherwise. Value concentrates where the measurement is subjective, where blinding is imperfect, or where the endpoint drives a go or no-go decision.

Oncology. The densest use case. Tumor response drives approval, criteria are complex, and lesion selection is a judgment call that varies between readers.

Neurology. Lesion counting, volumetry and atrophy measurement are highly sensitive to acquisition parameters and to reader consistency over long follow-up periods.

Respiratory. Structural and functional imaging is being used to detect regional change that global lung metrics flatten out, particularly in interstitial lung disease and emphysema.

Cardiology. Ejection fraction and structural measurements sit at the center of cardiotoxicity monitoring, where a small measurement drift changes a safety conclusion.

Metabolic disease and orthopaedics. DXA-based composition and structural endpoints depend almost entirely on acquisition consistency across sites and over time.

Building Imaging Into Trial Design: A Checklist

Imaging problems are almost always design problems discovered during conduct. This list belongs at protocol stage.

• Decide what the imaging endpoint is actually measuring. Response, progression, structural change, safety signal. The answer drives everything downstream.

• Choose the criteria before the modality. RECIST 1.1, iRECIST, PERCIST, RANO and the rest imply different acquisition requirements and different reader profiles.

• Test whether central reading is justified. Apply the FDA logic: is the trial blinded, is the measurement quantitative and standard, is it subject to bias? If not, say why in the protocol.

• Write the imaging charter early. Read design, blinding, reader qualification, adjudication rules, timepoint windows. It is the document a reviewer will ask for.

• Qualify the sites, do not just contract them. Equipment specification, technical settings, acceptable deviations, and phantom imaging or on-site training where appropriate.

• Define the quality control loop. When images are checked, how queries are raised, and how long a site has to re-acquire before the window closes.

• Plan archiving and retention from the start. The guidance expects images to be retained as part of subject care records, and for no less than two years following approval.

• Match the reader panel to the criteria. Specialty, criteria-specific training, documented qualification and periodic blinded re-testing during the trial.

Banook operates one of the larger independent reader networks in the sector: more than 500 board-certified specialists worldwide, each with at least ten years of experience, covering radiology, oncology, neuroradiology, nuclear medicine, dermatology and cardiology. 

Reader consistency is managed actively rather than assumed, through statistical monitoring of reader performance, consensus sessions, retraining and phantom-based bias checks. The group has supported more than 350 imaging studies, including pivotal trials, across CT, MRI, PET, SPECT, X-ray, ultrasound, angiography and DXA, and reads against RECIST 1.1, iRECIST, Lugano, LYRIC, PERCIST, mRECIST, Cheson, IMWG, CHOI, PCWG3, Macdonald, RANO, RANO-BM, INRC, WHO and tumor volumetrics. 

Each study is documented through an imaging manual, an imaging review charter and, for radioligand programs, dedicated dosimetry documentation.

Review the full central imaging offering or discuss a specific imaging endpoint with the team.