In cancer care, imaging has become a marvel of modern medicine. Advances in scanner technology, reconstruction algorithms, contrast agents, and diagnostic protocols have radically improved our ability to detect, track, and understand disease. Clinically, imaging is fast, accurate, and central to decision-making.
But in clinical trials? Imaging remains a bottleneck. Despite being one of the most important sources of data in oncology drug development, imaging in clinical trials is stubbornly opaque, clunky, and often wrong. Data is captured manually, transferred with delays, analyzed without traceability, and ultimately used to make go/no-go decisions that affect patient outcomes.
Sponsors, trial patients, site investigators, and study staff are frustrated, ill-served, and unable to get the accurate and timely results they need despite billions in drug development spending.
This disconnect is not benign. It means missed signals of efficacy. Patients inappropriately enrolled. Backlogs of unread scans. Errors that go undetected until it’s too late. In a time where trials are more complex, more global, and more urgent than ever, and budgets are squeezed, the status quo of imaging is not just outdated—it’s hazardous and wasteful.
The opportunity? A connected imaging platform that brings together sites, sponsors, CROs, and reader groups to deliver real-time, high-fidelity imaging endpoints for every patient and every trial, every time.
Imaging is critically important in oncology trials
Tumor progression is the primary endpoint in over 90% of oncology clinical trials. Imaging determines whether a patient qualifies for enrollment, whether they remain on study, and whether the drug is deemed effective enough to advance.
Imaging isn’t just part of the trial—it often is the trial.
That’s why tumor response criteria like RECIST 1.1, iRECIST, Lugano, RANO, and over 30 others have been developed. These frameworks standardize how tumors are measured, tracked, and interpreted. They define when a lesion is considered measurable, how to calculate progression, what constitutes a partial or complete response, and how pseudo-progression should be handled. In nearly all cases, these criteria are modified on a per-trial basis in each unique protocol.
Imaging assessments, which determine whether a patient is treated on a trial or not, are critical for these decisions made in real-time by investigators, radiologists, or both.
Yet, in most centers, the systems used to apply these criteria cannot meet these complex requirements. PACS systems show the images but can’t guide a trial-specific read. Sites rely on printouts, manual annotations, and spreadsheets. Measurement accuracy and eligibility decisions hinge on handwritten notes and siloed communication.
If imaging is wrong, the patient may be enrolled when they shouldn’t be—or excluded when they had no better option. Errors may delay therapy or compromise the trial’s credibility, resulting in significant expenses for sponsors when incorrect patients are enrolled and need to be replaced.
In an age where oncology trials are often the best hope for advanced cancer patients, the stakes could not be higher.
The new expectations of today’s sponsors
Today’s sponsors face more pressure than ever to accelerate trials and detect early signs of efficacy. With more phase I trials proceeding directly to phase III or registrational status, the first few patients scanned are no longer just data points—they’re the entire go/no-go signal.
To support this shift, sponsors need:
- Accurate imaging results at the point of care, captured during the patient visit and immediately pushed to the systems of record, including the sponsor’s EDC, eSource platform, or clinical data lake.
- Real-time EDC integration of imaging endpoints, with no manual re-entry or delayed uploads. But unlike labs or vitals, clinical trial imaging data isn’t in the EHR. Without a dedicated clinical trial imaging platform of record that can integrate with EHR-to-EDC systems or connect directly to EDC, it lives in separate silos: PACS, spreadsheets, USB drives, and image archives.
- Simultaneous visibility into site and central reads so that inconsistencies are caught early and not weeks later during a reconciliation effort. Sponsors require real-time read dashboards that show what has been read, by whom, and what the results were.
- Unified access to multi-modal data, where imaging results are contextually linked to clinical outcomes, labs, genomic markers, and other endpoints—to extract meaning earlier and improve trial design.
- Decoupling of reader workflows from the reading vendors, so sponsors can retain their data and analysis methods even if the reader group changes. Lock-in must give way to flexibility.
- Site and central reads managed on a shared platform, giving sponsors a harmonized view of how the trial is being executed across geographies, reducing variability, and improving oversight.
- Instant access to high-quality source data, including lesion annotations, DICOM metadata, and protocol-specific forms that can be exported for any type of sponsor-side analysis.
These demands are not futuristic. They’re happening now. Sponsors who can’t meet them risk slower trials, higher costs, and missed opportunities.
The imaging workflow is broken
At most cancer centers, research imaging is a broken patchwork of good intentions and disconnected, manual processes. It isn’t because people aren’t working hard. It’s because the tools they have were never designed for this.
- PACS systems used by radiologists have no awareness of trial protocols. They don’t support lesion tracking, response criteria, or conformance to trial-specific image acquisition or analysis standards.
- Bolt-on measurement tools can do the math but don’t solve the workflow. They aren’t web-based. They aren’t collaborative. And they don’t reduce the burden on study staff, who still spend eight hours prepping each read.
- Radiologists often don’t want to give up research work but can’t dedicate the time to do it well or in time for the patient’s next visit. Meanwhile, trial coordinators are left chasing unread time points, reconciling discrepancies, and manually uploading results.
- Trial imaging backlogs are a silent epidemic. Some centers have hundreds of unread time points spanning years. Others have scans that were never reviewed or scored against the protocol, leaving patients and sponsors in the dark.
- Cancer Centers have published that 25% to 50% of imaging time points contain an error—from missing measurements to incorrect calculations to protocol violations. Most aren’t caught unless you look.
- Sponsors often rely on the eCRF values and a PDF signed by the investigator, but there is no traceability back to the annotated images. There is no audit trail, and there is no link from the measurement in the database to the actual pixel on the scan.
- FDA audits are beginning to scrutinize this more closely, and some sponsors are being asked to justify decisions made on data that cannot be traced to the source image or verified for accuracy.
The system is collapsing under its weight. The people doing the work are overburdened. The processes are delayed and error-prone, and the consequences are growing more severe.
AI won’t save imaging (yet), but technology can help
Artificial intelligence is often proposed as the solution to the imaging crisis. And in time, it may be. But not today.
- Tumor response criteria like RECIST are human-readable standards defined by regulators. An AI model that calculates response would need to itself be a new standard—which means regulatory validation, large-scale reproducibility, and global clinical acceptance. That’s 20+ years away, not two.
- While over 1,000 imaging AI solutions have been cleared by the FDA, most are used sparingly in clinical care. FDA clearance typically means “substantial equivalence” to a prior device, not breakthrough performance.
- Clinical trials require lesion tracking, selection, and measurements that are both auditable and accurate. AI models that draw bounding boxes on CTs aren’t enough. And even when models work, they must generalize across sites, scanner types, patient populations, and disease indications.
- For AI to help in trials, it needs a modern imaging workflow platform—one that can ingest, organize, display, and validate the data. AI requires a purpose-built stage. Without it, the models are irrelevant.
But technology can help today. A modern imaging platform can:
- Guide radiologists through protocol-specific reads.
- Automate workflow for lesion tracking, measurements, and calculations.
- Validate conformance to trial-specific response criteria.
- Share results instantly with sites, sponsors, and CROs.
- Create a permanent, traceable, and audit-ready link from image to result.
- Enable collaboration across geographies, time zones, and institutions.
This is not science fiction. It is already being done.
What this means for sites, sponsors, and patients
Sites can finally standardize every trial—no matter the sponsor, the criteria, or the CRO—on a single cloud-based imaging platform that reduces effort and increases accuracy. Study staff can focus on patient care, not image handling.
Sponsors gain real-time access to imaging data from both sites and central readers. They can track reads, detect errors, and get complete endpoint packages faster and at a lower cost. They can analyze lesion-level data across trials and optimize their pipelines based on early signals.
And patients? They receive better care. Imaging results are reviewed on time. Errors that might disqualify them or delay treatment are eliminated. The trial becomes a more reliable bridge to the therapy they need.
Imaging has ridden to the bottom in clinical trials. But the path up is now available and clear. It starts with connected platforms, collaborative workflows, and a refusal to accept that the status quo is good enough.
The cancer patients of today—and the medicines of tomorrow—deserve nothing less.
Yunu provides innovative imaging workflow and data management solutions designed to optimize clinical trial processes. By integrating advanced technologies, Yunu enables life sciences organizations to streamline imaging workflows, improve accuracy, and accelerate timelines. Yunu’s platform supports clinical trials across various therapeutic areas, offering scalability and flexibility for organizations of all sizes. For more information, visit yunu.io and follow us on LinkedIn or X @Yunu_Inc.
