Advanced Intelligent Clear-IQ Engine (AiCE)

AI in Medical Imaging

AiCE is the next generation of CT reconstruction technology. The world’s first Deep Learning Reconstruction method, AiCE quickly produces stunning CT images that are exceptionally detailed and with the low-noise properties you might expect of a future advanced MBIR (Model-based Iterative Reconstruction) algorithm.

Better IQ. Lower dose. Easy workflow.

“On our Aquilion ONE GENESIS a new Deep Learning Reconstruction, named AiCE, was recently installed. Compared to our previous reconstructions we noticed significantly improved IQ with AiCE.

The clinical images have less noise and are sharper, we noticed an increase of ~25% SNR and CNR.

Moreover the dose decreased by ~20% for body imaging and ~40% for cardiac examinations.

The AiCE implementation was easy and just turned ON by the Canon application specialists. We scan approximately 80 patients per day and therefore high throughput is essential for our site. AiCE is implemented into our protocols helping to streamline our daily workflow.”

Prof. Frédéric Ricolfi
Centre Hospitalier Universitaire
Dijon Bourgogne, France


“Tomorrow’s radiology available today”

Listen to Ewoud Smit, MD, PhD, from Radboud University Medical Center, Nijmegen, the Netherlands, explain how AiCE has won over even the most conservative radiologists in his department by offering them more diagnostic confidence.

Watch now

Integrated Intelligence

Sharp, clear and distinct images. At low dose.

Harnessing the enormous computational power of a Deep Convolutional Neural Network (DCNN), AiCE is trained to restore low-quality CT data to match the properties of an advanced MBIR. Without the need for multiple forward-projected iteration cycles, reconstructions are ultra-fast and highly accurate.

AiCE features:

  • Exceptional low-noise properties
  • Enhanced anatomical resolution
  • Superb image homogeneity
  • Fast zero-impact reconstruction

Deep Learning Reconstruction (DLR)

The DCNN learns which methods are best applied to maintain the spatial resolution and low-noise properties contained in the advanced MBIR algorithm. The more variations of data supplied during training, the better the final algorithm will perform in terms of image quality and processing speeds.