A complete AI imaging suite — from deep-learning core engine to clinical integration — designed to bring functional imaging to every diagnostic setting worldwide.
Insilico PET® is built on a conditional generative deep learning architecture, trained end-to-end on a large, multi-institutional dataset of paired CT and PET acquisitions. The model learns the complex, non-linear relationship between anatomical CT features and the underlying functional metabolic state of tissue.
At inference time, a DICOM CT series is passed through the model, which outputs a synthetic PET volume at full spatial resolution. The output is quantified in standardised uptake value (SUV) units, directly comparable to a true PET acquisition, and delivered as a standard DICOM series.
Our model architecture incorporates uncertainty quantification, providing clinicians with confidence maps alongside each functional output — enabling informed, safety-conscious clinical decision-making.
Insilico PET® is available across six complementary clinical and research use cases, each built on the same validated core engine.
Generate functional metabolic maps from CT to identify primary tumours, metabolically active lymph nodes, and distant metastases. Supports initial staging, mid-treatment response assessment, and end-of-treatment evaluation across lung, colorectal, head and neck, and other solid tumours.
Request clinical accessMap cerebral glucose metabolism to support early diagnosis of Alzheimer's disease, Parkinson's disease, Lewy body dementia, and other neurodegenerative conditions. Our model captures hypometabolic patterns characteristic of neurodegeneration, enabling earlier intervention.
Request clinical accessAssess myocardial viability, perfusion defects, and inflammatory cardiac disease from standard cardiac CT. Provides functional data critical for decisions around coronary revascularisation, device implantation, and heart failure management.
Request clinical accessIncorporate functional imaging endpoints into clinical trials at a fraction of the cost and logistical complexity of conventional PET. Insilico PET® enables metabolic response biomarkers in Phase II and III trials without requiring PET-capable sites or radiotracer supply chains.
Discuss partnershipA DICOM-native, HL7-compatible API that slots into existing PACS, RIS, and EHR environments. Insilico PET® outputs appear directly in the radiologist's reading workstation alongside standard CT series — zero workflow disruption.
Schedule integration callCollaborate with RADiCAIT on funded research programmes. We provide access to our models, datasets, and technical team for university and NHS research partners investigating AI in medical imaging, with co-publication opportunities.
Explore research accessInsilico PET® has been developed and validated against ground-truth PET imaging across thousands of paired cases. Our validation programme spans multiple institutions, patient demographics, and clinical indications to ensure robust real-world performance.
We measure performance using standardised metrics including SUV concordance, Dice coefficient for lesion segmentation, and overall diagnostic accuracy benchmarked against radiologist reads of true PET images.
Request Validation DataRADiCAIT handles the full implementation journey — from initial integration to ongoing model updates and clinical support.
We work with your IT and radiology teams to map existing infrastructure and define integration requirements, timelines, and success metrics.
Our DICOM-native API is configured to your PACS environment. Typical integration takes 2–4 weeks. No new hardware required.
Dedicated onboarding, radiologist training, and ongoing technical support. Continuous model improvements are deployed automatically.
Request a personalised demo and see Insilico PET® applied to your imaging use case.