| Περίληψη: | Multi Detector Computed tomography (MDCT) is one of the basic imaging modalities in clinical radiology. Reconstruction algorithms are the most important asset of computed tomography nowadays. Iterative algorithms evolution enables reconstructed images with lower noise levels. Optimization and evaluation of iterative algorithms in computed tomography is an ongoing process.
Computed Tomography Texture Analysis (CTTA) has a great impact in image analysis. Most recent studies suggest that CTTA may add informations and even help clinical diagnosis. There is, a great number of pub-lished studies mentioning the use of reconstructed images from phantoms, anthropomorphic phantoms and even clinical data to evaluate hybrid iterative algorithms reconstruction techniques in multiple detector computed tomography of all vendors.
Most frequently used image quality indices include: noise, noise power spectrum, standard deviation, signal to noise ratio, contrast to noise ratio, low contrast resolution and most recently first order statistics: standard deviation, mean, min, max, skewness, kurtosis and entropy. Open issues include appropriate Iterative Reconstruction blending, for specific clinical protocols that ensure noise reduction, acceptable CNR levels, diagnostical image texture in the reconstructed images. Quantitative image quality evaluation on phantom image data is a prerequisite for patient dose optimization.
This study focuses on the quantitative evaluation of image quality indices using standard deviation, CNR, first and second order statistics to evaluate General Electric hybrid iterative algorithm ASIR (Adaptive Statistical Iterative Reconstruction) performance using phantom data. The aim of this study is to evaluate the effect of ASIR reconstruction blending on multi-detector computed tomography image quality, assessing the effect of exposure settings and IR blending on image quality indices: Noise, CNR, First order Statistics. This study also embodies an entrepreneurial evaluation of Second order statistics .There is a presentation of second order statistics features that seem to contribute in image texture analysis such as ASM (Angular Second Moment), IDM(Inverse Difference Moment), second order Entropy, second order Contrast, Correlation.
Images were acquired using 2 types of phantoms 1)Mini CT QC Phantom (nuclear associates 76-430) and 2) standard homogenous General Electric acrylic phantom .Mini CT QC Phantom consists of 7 parts with 7 dif-ferent materials : Bone equivalent, Polyethylene, Polystyrene, Plastic Water, Nylon, Polycarbonate and a homogenous region simulating soft tissue.
Results validate the effect of milliamper (mA) via Noise Index manipulation and ASIR blending on recon-structed image noise and CNR. First and Second order texture features were also capable of capturing image quality alterations, due to exposure and ASIR blending parameters.
|
|---|
