Aquilion Prime SP (TSX-303B8) V10.2 with AiCE-i

The Aquilion Prime SP (TSX-303B/8) V10.2 with AiCE-i is a computed tomography system designed for whole-body imaging, which incorporates AiCE-i, an image reconstruction technology based on Deep Convolutional Neural Networks. This system aims to enhance diagnostic capabilities by significantly reducing image noise and radiation dose while improving image quality, spatial resolution, and low-contrast detectability compared to previous reconstruction methods.

The Aquilion Prime SP (TSX-303B/8) V10.2 with AiCE-i is a 40-row computed tomography (CT) system, optionally expandable to 80 active detector rows, designed to acquire and display cross-sectional volumes of the whole body, including the head. Its primary function is to provide volume sets that can be utilized by trained and qualified physicians for specialized studies using indicated software/hardware. This system builds upon the technology of previously marketed Canon CT systems and introduces AiCE-i (Advanced intelligent Clear-IQ Engine - integrated), a reconstruction technology intended to reduce image noise and improve image quality. The core innovation, AiCE-i, employs Deep Convolutional Neural Network (DCNN) methods. These methods are designed to explore the statistical properties of signal and noise, learning to differentiate structure from noise to produce fast, high-quality CT reconstructions. This technology is applicable across various anatomical regions, including the abdomen, pelvis, lung, cardiac, extremities, head, and inner ear. Performance testing, utilizing phantoms, demonstrated significant improvements including: improved quantitative high contrast spatial resolution over AIDR 3D with reduced noise; an 82% dose reduction relative to Filtered Backprojection (FBP); and for body imaging, 13.8% improved low-contrast detectability and a reduction in the Standard Deviation of Noise by 21.4 (units not specified) at the same dose compared to AIDR 3D. Furthermore, AiCE produces images with noise texture more similar to high-dose FBP than Model-Based Iterative Reconstruction (MBIR), and supports low-contrast detectability of 1.5 mm at 0.3% (21.8mGy) for body applications using AiCE Body. Review by an American Board Certified Radiologist confirmed that AiCE reconstructed images were of diagnostic quality for representative anatomical regions.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the performance of AiCE-i, particularly its noise reduction and low-contrast detectability, vary across different anatomical regions (e.g., cardiac vs. inner ear) and patient demographics (e.g., pediatric vs. adult)?; What are the specific computational requirements (e.g., processing time, hardware) for the AiCE-i reconstruction algorithm compared to traditional methods like FBP or AIDR 3D, and how might this impact clinical workflow?; Beyond the reported 82% dose reduction compared to Filtered Backprojection (FBP), what is the typical dose reduction achieved with AiCE-i when compared to the more recent iterative reconstruction technique AIDR 3D in routine clinical practice?; How does the stated 'noise appearance/texture more similar to high dose filtered backprojection compared to MBIR' achieved by AiCE-i affect diagnostic confidence and interpretation for radiologists accustomed to traditional image appearances?; What are the characteristics of the training data (e.g., size, diversity, annotation methods) for the Deep Convolutional Neural Network underlying AiCE-i, and how does the network generalize to unseen patient data or rare pathologies?