A Conceptual Abstraction for Human-Computer Interaction: Data Notes for the Development of New Techniques for Magnetic Resonance Imaging

Abstract

In applied artificial intelligence, abstraction is a conceptual process by which complex systems are simplified, for the sake of improving human-computer interaction. This manuscript presents MaZda 5 RC, a software framework for magnetic resonance imaging (MRI). Though its intended application is MRI, this abstraction system can be used in any area of research and professional practice ? especially wherever high-dynamic range (HDR) still and motion media need to be analyzed with computer vision. In this data descriptor, a set of techniques was introduced to investigate the visually-lossless aspects of different image acquisition formats. 1.5/3T DICOM files were decoded to RAW images (sRGB-48) for investigation of noise and artifacts. Additional techniques were used to quantify and visualize the invisible effects of noise and artifacts in magnetic resonance (MR) images of two human fetuses. Algorithmic sorting was used to reduce more than 300 computerized values to 13 epitomic parameters (i.e. a numerical representation of the difference between the original and the denoised images). The 13 epitomic parameters were subsequently used to investigate bit-depth and artifacts in uncompressed HDR images of prenatal brain originally captured with 1.5/3T MRI scanners. As hypothesized, no qualitative difference was observed when LCD monitors were set to

Ähnliche Arbeiten