SubtlePET
AI solution for PET scan optimization
SubtlePET is a deep learning solution allowing nuclear medicine practices to reduce acquisition time on existing machines and/or reduce injected dose during acquisition. From a low quality image, SubtlePET corrects and returns an image of a quality equivalent to an image taken under standard acquisition conditions, directly into the radiologist reading tools.

Reduced acquisition time
SubtlePET improves quality of images taken up to 4X faster. The image returned by SubtlePET is of standard clinical quality.
Reduction of injected dose of FDG, FCH, FDOPA, Ga-PSMA…
SubtlePET improves quality of images taken with a reduced injected dose. The image returned by SubtlePET is of standard clinical quality.
Compatible with all PET scan models and supplier
SubtlePET delivers the exam directly into the radiologist’s reading tool and therefore fits seamlessly into the existing infrastructure.
- PET scan
- nuclear medecine
- scanner
- low dose
- full-length scan acquisition
- denoise
- scarce nuclear tracers
Save time, reassure your diagnosis and streamline your workflow with Incepto
Testimonial
The quality of the images improved by the Subtle tool makes them indistinguishable from images obtained by conventional means. In addition, the results are free of artifacts and lesions can be easily identified. I can already see how the tool will improve both our workflow and the patient experience.
Peter Giuliano,
Director of the Department of Nuclear Medicine
HOAG Hospital (California – USA)
Publications
- Quantitative Standardized Uptake Value Evaluation of 4x Faster PET Scans Enhanced using Deep Learning – A Chaudhari, PhD; P Gulaka, PhD; T Zhang; S Srinivas, MD, PhD; G Zaharchuk, MD, PhD; E Gong, PhD, Stanford.
- 200x Low-dose PET Reconstruction using Deep Learning – Junshen Xu*, Enhao Gong*, John Pauly, Greg Zaharchuk.
- Accelerating Whole-Body PET Acquisitions Using Deep Learning: External validation on foreign country data – Jose Leite, Gustavo Tukamoto1, Akshay Chaudhari, Praveen Gulaka, Enhao Gong, GregZaharchuk, Igor Rafael Martins dos Santos, Flávia Paiva Proença Lobo Lopes, Felipe Campos, Kitamura.