Optimisation of radiographic imaging by means of factorial experiments

Sammanfattning: In the optimisation process of radiographic imaging, factorial designed experiments can be applied. The parameters (factors) are varied together instead of one at a time, making it possible to discover interactions between the factors as well as main influences of them on the result variable. A 2k design implies having k number of factors each one set to two different levels (low and high).A computer program, CoCIQ, designed to automatically analyse and evaluate test images of a contrast-detail phantom, was evaluated and adjusted to clinical situations using a flat panel detector. The program gives a quantified measurement of image quality by calculating an Image Quality Figure (IQF) for the X-ray image. It was shown that the program produces IQF with small variations. It was also found that there was a strong linear statistical relation between the computerised evaluation and the evaluation performed by human observers.2k factorial experiments were evaluated by investigating the influence of tube potential, tube loading, focus size and filtration on the result variables IQF, Kerma Area Product (KAP) and effective dose using a flat panel detector. It was found that the result variables were mainly influenced by tube loading, tube potential and filtration. Interactions between tube potential and filtration as well as between tube loading and filtration were observed, too. This work demonstrates that accepted knowledge was reproduced and that the effects of interactions between parameters were revealed.Extended 2k experiments were then applied at three different optimisation procedures. Two studies were performed using a flat panel detector for lumbar spine radiography. The aim was to find optimal settings for tube potential, system sensitivity and filtration for different sized patients and, in a separate study, to investigate the effect of the image post processing parameters and the possibility for dose reduction by adjusting these. The parameters are ROI (Region Of Interest) density, gamma, detail contrast enhancement, unsharp masking, kernel size and noise compensation.After determining the optimal settings from these experiments, X-ray images of the lumbar spine of an Alderson phantom were acquired and evaluated in a visual grading analysis (VGA).The results illustrated that the image quality was maintained at a lower effective dose by operating with a reduced tube potential and increased sensitivity of the X-ray system.The experiments on image post process parameters revealed their influence on image quality and indicated that image quality could be improved by changing the settings of the process parameters.Factorial experiments were also performed, using a multislice CT scanner to investigate the possibility for dose reduction at paediatric head examinations. An anthropomorphic phantom simulating a one-year-old child was scanned using different settings of tube potential, tube loading and reconstruction filter.The study showed that a 25 % reduction of dose was possible with maintained image quality by reducing the tube loading.Factorial designed experiments provide an effective method to simultaneously predict the influence of various parameters on image quality and radiation dose in the optimisation in diagnostic radiology.