Aquilion ONE (TSX-305A/6) V8.9 with AiCE

The Aquilion ONE (TSX-305A/6) V8.9 with AiCE is a computed tomography system designed for whole-body imaging, utilizing an advanced image reconstruction engine based on Deep Convolutional Neural Networks. This technology aims to significantly improve image quality by reducing noise and enhancing spatial and low-contrast resolution, while also enabling substantial radiation dose reduction compared to previous methods.

The Aquilion ONE (TSX-305A/6) V8.9 with AiCE is a whole-body, multi-slice helical Computed Tomography (CT) scanner designed to acquire and display cross-sectional volumetric data of the entire body, including organs like the brain and heart in a single rotation, for specialized diagnostic studies. Its core innovation is the Advanced intelligent Clear-IQ Engine (AiCE), a reconstruction technology that employs Deep Convolutional Neural Network (DCNN) methods. AiCE is engineered to differentiate signal from noise more effectively, thereby reducing image noise and enhancing overall image quality, particularly for abdomen, pelvis, lung, and cardiac applications. Performance evaluations utilizing phantoms demonstrated significant improvements with AiCE. These include an enhanced quantitative spatial resolution (7.4 lp/cm at 10% MTF versus 5.7 lp/cm with AIDR 3D), substantial dose reduction (79.6-82.4% compared to filtered back projection), a 12.2% improvement in low-contrast detectability, and a 29.2% noise reduction at equivalent dose levels compared to AIDR 3D. Additionally, AiCE achieved low-contrast detectability of 1.5 mm at 0.3% with 22.6mGy, and its noise texture was found to be more akin to standard filtered backprojection than iterative methods like FIRST, with diagnostic quality confirmed by radiologists for key applications.

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's Deep Convolutional Neural Network compare to other DCNN-based image reconstruction algorithms in terms of computational efficiency and the potential for introducing new image artifacts?; What were the specific characteristics of the training dataset used for the AiCE Deep Convolutional Neural Network, and how might variations in this dataset impact its robustness across diverse patient populations or rare pathological findings?; Beyond the abdomen, pelvis, lung, and cardiac applications, what is the potential applicability and validated performance of AiCE for other anatomical regions or specialized CT examinations, such as neurological or musculoskeletal imaging?; What is the quantitative impact of AiCE on diagnostic accuracy or inter-reader variability in clinical practice when compared to predecessor technologies like AIDR 3D or FIRST, particularly in challenging low-contrast or high-noise scenarios?; Given that AiCE's noise texture is more similar to filtered backprojection, how does this affect radiologists' learning curve and subjective image assessment compared to the distinct textures often produced by iterative reconstruction algorithms like FIRST?