Most platforms used in imaging clinical trials grew through acquisition of disparate tools, or started as specialized reader software later stretched to handle image collection. DYNAMIKA™ is not a merger of tools, and it's not a single-modality or single-purpose review system. It is the enterprise platform, designed and built from the ground up for imaging clinical trials – delivering regulatory-grade data and giving every trial stakeholder the workflow they need to succeed.
Six trials. Six imaging platforms. Six disconnected databases.
When a biotech's Chief Medical Officer asks which sites consistently underperform, or which imaging endpoints are generating the cleanest signal across the program, there is often no fast answer. The imaging trial data lives in separate silos: sites upload scans into a 'Dropbox-like' website, and the Quality Control team only reaches for these images once a month, by which point multiple baseline scans have already been acquired incorrectly. Radiologists, asked to receive data on their workstations and turn reads around within 24 hours, are frustrated, and the pressure leads to mistakes. Last year's Phase 2 reader discrepancy metrics are gone, buried in an Excel sheet no one can find. QC benchmarks, site feasibility scores, and performance data that took months to establish have to be rebuilt or awkwardly adapted for every new study using the same outdated tools. And choosing readers is always a headache, since no one can easily tell their calibration rates or availability.
This is not a hypothetical scenario. It is the default operating model for most biotech sponsors running concurrent imaging programs today. And it is why, when something goes wrong, whether in site performance, reader inconsistency, or image quality, the company tends to find out at the interim analysis, when the window to fix it has already closed.
By the time these problems surface, the cost is no longer measured in hours or Excel sheets. It shows up as protocol amendments, extended timelines, and data too inconsistent to support a confident go/no-go call. The irony is that none of this is a data problem. It is an infrastructure problem, one created by stitching together tools that were never designed to work as a single system in the first place.
The Imaging Clinical Trial Infrastructure Decision: Critical to Success of the Trial
Most sponsors treat the imaging platform decision the way they treat vendor selection for any other clinical service: availability, cost, whether someone else has used it before. In the modern world of quantitative endpoints, use of AI-powered workflows, reliance on exploratory endpoints to build competitive drug profile while saving costs and time, the platform for the trial is the single most important decision, one that should get priority review and C-level sign off.
The consequence of the decision is visible in the data. In oncology trials, we often see reader agreement rates collapse under independent adjudication. In obesity trials, DXA calibration gaps surface at the interim analysis. In long trials which are often needed for drugs targeting rheumatic or musculoskeletal diseases, scanner upgrades mid-study silently rewrite months of longitudinal data. None of these are science problems. They are clinical trial infrastructure problems, and they are entirely predictable, which means they are entirely preventable if the infrastructure was designed to prevent them.
Image Analysis Group (IAG) has spent almost 20 years and 700 global trials studying these failure modes. DYNAMIKA™ is the platform built to address them. It is not a radiology PACS system adapted for research, not a general-purpose EDC with an imaging module bolted on, and definitely not a Dropbox-style solution for image collection. It is a purpose-built operating system for the imaging layer of a drug development program – engineered on cloud-native infrastructure to run the full lifecycle, from site qualification and the first test image upload through final database lock. Along the way, it provides real-time quality control, orchestrated workflows for every trial stakeholder, a centralized performance dashboard, and a compliance architecture that never needs to be rebuilt from scratch for each new study. All read methods, all AI-workflows and API enablement for specific reads stay on, thus saving time and costs for study start up and reading tools.
The enterprise model is what truly sets DYNAMIKA™ apart. Many imaging platforms treat each trial as an isolated instance, so every new trial means standing up the data management system again. Sponsors can end up running dozens of near-identical instances, each configured slightly differently.
DYNAMIKA™ takes the enterprise-level approach: it consolidates all of a sponsor's studies, sites, and readers under one shared infrastructure. Site performance history from a previous program, reader calibration records, and QC rules that caught specific failure modes in a prior indication all carry forward automatically. By the third study, the platform already understands the sponsor's entire network.
What DYNAMIKA™ Covers
DYNAMIKA™'s underlying infrastructure stays constant across every therapeutic area IAG supports: the same cloud-native architecture, the same real-time QC at upload, the same centralized adjudication workflow, and the same 21 CFR Part 11 audit trail. What changes is the indication-specific endpoint knowledge applied when images are read, whether by a radiologist or an AI model. This consistency matters more than ever, because modern trials for advanced therapies and complex designs increasingly rely on a mix of imaging modalities and read criteria within a single protocol. A cachexia trial, for example, needs DXA for bone mass alongside CT-based RECIST 1.1 for tumor assessment, while an obesity trial needs MRI-based muscle mass assessment paired with DXA-based bone health measurements.
The table below shows examples of FDA-required and exploratory imaging endpoints in these cross-modality settings, which DYNAMIKA™ supports across IAG's core therapeutic areas.
Key indication examples.
IAG also supports cardiac safety, cardiovascular imaging, radiopharmaceutical drug development, rare diseases, and additional therapeutic areas where imaging endpoints are protocol-specific and require configuration from the ground up. The breadth is a function of the platform architecture: because DYNAMIKA™ was built to be configurable at the protocol level, each new indication does not require rebuilding the underlying infrastructure.
One Platform. Every Trial. Every Site. Every Reader.
Quality of imaging data will impact the quality of final efficacy. DYNAMIKA™’s instantaneous automated QC at image upload is a foundational capability, not a downstream check. When a site submits an image that deviates from protocol parameters (wrong echo time, mismatched voxel size, scanner model not on the approved list), DYNAMIKA™ flags it before it enters the dataset. The sponsor does not find out at the interim. They find out that morning.
The IROC National Clinical Trial Network demonstrated what active monitoring achieves in practice: overall protocol conformance across a multi-site oncology network increased from 73% to 85% over four years. PET uptake time conformance improved from 47% to 84%. Those gains were not produced by better protocols or more experienced sites. They were produced by the feedback loop. Sites that receive immediate QC results correct their behavior. Sites that receive results weeks later cannot.
The enterprise model compounds over time. Site activation, QC calibration, and reader performance history from a previous study carry forward to the next. A sponsor running their third program with IAG knows before enrollment opens which sites historically deliver conformant data and which readers demonstrate the tightest agreement rates on the specific endpoints the new protocol requires.
Why Infrastructure Decisions Cannot Be Undone
The data integrity problems that DYNAMIKA™ is designed to prevent are documented in the peer-reviewed literature and in FDA records. Understanding them is useful, not as the lead of this story but as the evidence base for why the platform choices described above matter.
On DXA: when the same protocol runs across a multi-site network, the inter-center CV for bone mineral density reaches 8.18%, against intra-center precision of 1.40%. For lean mass the ratio is worse: a tenfold amplification of technical noise. The ISCD cross-calibration mandate requires formal cross-calibration within 60 days of any scanner replacement, or a new baseline must be established, nullifying every patient's longitudinal contribution to the primary endpoint.
On scanner upgrades: a Siemens hardware upgrade produced a 30% increase in neocortical contrast-to-noise ratio, with WMH volumes dropping 68%, indistinguishable from a drug effect in a neurodegeneration trial. Cortical volume differences of 2-4% are comparable to two years of atrophy. The FDA's guidance on mid-study changes classifies major scanner upgrades as potential significant changes requiring IDE supplements.
On audit trails: during the review of NDA 208772 for Brigatinib, the FDA's Office of Scientific Investigations stated it could not recommend data reliability from the contracted imaging vendor due to inadequate 21 CFR Part 11-compliant audit trails. The submission was jeopardized by the imaging infrastructure, not by the drug.
Human Expertise, Scaled by Technology
The conversation about AI in clinical trial imaging is generating more confusion than clarity. IAG's position is precise: machine intelligence supports the workflows that human experts run. QC flags that a human investigates. Documentation that a human reviews. Reader calibration data that a specialist interprets before it becomes a protocol intervention.
The value of 20 years of experience is not the volume of data. It is the institutional knowledge about what goes wrong and when. The failure modes that DYNAMIKA™’s rules are designed to prevent were not discovered by an algorithm. They were observed by specialists who reviewed the data, spoke with multiple expert radiologists and impacted approval of many drugs. This is the team behind DYNAMIKA™.
Sponsors working with IAG have helped shape the current regulatory standards for imaging endpoints. The ones defining the next standard, in GLP-1 programs, radiopharma trials, and the first wave of AI-augmented oncology studies, are doing it now.
The Operating System Decision
The operating system decision gets made once. It determines whether imaging data holds up at interim, whether the endpoint survives regulatory scrutiny, and whether the program can scale from Phase 2 to Phase 3 without rebuilding its infrastructure from scratch.
It is made at protocol design, before the first site is activated, before the first patient is scanned. The sponsors who make it well tend to be the ones presenting at the podium. The ones who don't tend to be the ones requesting an extension.
Most platforms were built for something else entirely. DYNAMIKA™ is here to ensure the success of imaging clinical trials.
About Image Analysis Group
Image Analysis Group (IAG) is a specialist imaging CRO which delivers end-to-end solutions to sponsors who use medical imaging to assess safety or efficacy of their new drug candidates. IAG build DYNAMIKA™, our cloud-native, GxP-validated, SOC 2 Type II and 21 CFR Part 11-compliant platform for clinical trial imaging. Single platform for end-to-end upload, QC and central review of imaging data in phase I-III trials, with embedded AI-powered workflows, FDA required endpoints, centralized image review and real-time trial oversight of trial progress and stakeholders’ performance. DYNAMIKA™ is helping Sponsors move from first scan to database lock faster, thus accelerating clinical trials and increases chances of drug development success.
Email: [email protected]
Website: www.ia-grp.com LinkedIn: https://www.linkedin.com/company/image-analysis-ltd/
References
- Sutter et al.: DXA intra-center CV 1.40% BMD / 0.76% lean mass; inter-center CV 8.18% BMD / 7.89% lean mass. Bone 2021. https://pubmed.ncbi.nlm.nih.gov/33045389/
- ISCD 2023 Official Adult Positions: cross-calibration mandate, precision thresholds, LSC methodology. https://iscd.org/wp-content/uploads/2024/03/2023-ISCD-Adult-Positions.pdf
- PINTAD consortium (Borradaile et al.): 77% of reader disagreements from image perception; 40 trials, 12,299 participants. Ther Innov Regul Sci 2021. https://pmc.ncbi.nlm.nih.gov/articles/PMC8259547/
- Basty et al., UK Biobank n=32,961+: DXA overestimates lean mass; MRI detects 4-5% longitudinal muscle loss DXA misses. Commun Med 2026. https://pmc.ncbi.nlm.nih.gov/articles/PMC13086945/
- Beaumont et al., RECIST 1.1 BICR vs local: 85% different target lesions, 59% different baseline scans, 36.7% adjudication rate. Cancer Imaging 2018. https://pmc.ncbi.nlm.nih.gov/articles/PMC6288919/
- Backhausen et al., MRI scanner upgrade (Siemens Verio vs Skyra): cortical volume differences 2-4% across regions. PLoS One 2021. https://pmc.ncbi.nlm.nih.gov/articles/PMC8491918/
- Potvin et al., MRI system upgrade (Trio to Prismafit): 30% CNR increase, neocortical volumes 1.9-6.4% larger, WMH -68%. Front Neurol 2019. https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2019.00726/full
- Clarkson et al., ADNI phantom voxel scaling: 0.20% mean absolute volume change from scanner drift. Neuroimage 2009. https://pmc.ncbi.nlm.nih.gov/articles/PMC2800076/
- FDA Clinical Trial Imaging Endpoint Process Standards Guidance for Industry (2018). https://www.fda.gov/files/drugs/published/Clinical-Trial-Imaging-Endpoint-Process-Standards-Guidance-for-Industry.pdf
- FDA Guidance: Changes or Modifications During the Conduct of a Clinical Investigation. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/changes-or-modifications-during-conduct-clinical-investigation-final-guidance-industry-and-cdrh
- Poon et al. (IROC), QC in multicenter clinical trials: conformance 73% to 85% over 4 years; PET uptake 47% to 84%. RSNA 2016. https://www.rsna.org/uploadedFiles/RSNA/Content/Science/Quality/Storyboards/2016/Poon-QS018.pdf
- FDA NDA 208772, Brigatinib (Alunbrig): OSI could not recommend data reliability from imaging vendor; inadequate 21 CFR Part 11 audit trails. https://accessdata.fda.gov/drugsatfda_docs/nda/2017/208772Orig1s000OtherR.pdf
- Trevisan et al., DXA: 1 cm hip placement shift = 4% BMD change; 3 cm scan window enlargement = 3% BMD increase. J Bone Miner Res 1992. https://pubmed.ncbi.nlm.nih.gov/1481723/
Moe Alsumidaie, MBA, MSF, is founder and Chief Editor of Vanguard Publications, which publishes Clinical Trial Vanguard, Pharma Vanguard and BullScope, and Head of Research at CliniBiz. He has two decades in clinical trial operations and data science, with earlier roles at Genentech, Abbott Vascular and Stanford University Medical Center, and is a guest lecturer in clinical trial sciences at Rutgers University.
