Search

Aim: To investigate the impact of different kVp and mAs values on effective dose and image quality using a chest phantom that simulates a normal sized and an obese patient. Methods and materials: A chest phantom with simulated pathological nodules was imaged at various kVp and mAs values. To determine the image quality, CNR and SNR were calculated. An observer study was carried out using relative visual grading with a 3-point Likert scale to assess image quality and nodule visibility. The VGA-study reference image was of the phantom at standard size without the chest plates using 125kVp, 2.4mAs by AEC and 24μSv. Visual grading scores were compared against SNR and CNR values in order to determine the optimal acquisition parameters. Effective dose was calculated using Monte Carlo simulation software, and a Figure of Merit was calculated. Results: The image obtained with 125 kVp and 4.0 mAs had the highest SNR, and the one with 125 kVp and 2.0 mAs had the highest CNR. The observers found that 125 kVp/4.0 mAs was the most optimal image and 125 kVp/6.88 mAs had the least image quality, when compared to the reference image. On calculating the Figure of Merit, 125 kVp/2.0 mAs has the highest score. The effective dose varied from 5.34 μSv to 73.5 μSv for the range of parameters used. Conclusion: It is possible to get higher SNR, CNR and VGA-scores in large sized patient chest radiography at lower mAs than that given by using standard AEC, due to post-processing. Manual mAs better control the image quality than using AEC. Anatomical features are better detected using a higher mAs and a standard kVp. Better image contrast is achieved when a lower kVp and standard mAs is utilised. A protocol for larger patients needs to be tailored accordingly.

Aim: To investigate the effect of source-to-image distance (SID) on effective dose and image quality at a fixed mAs. Furthermore, to determine whether effective dose varies as SID is varied after compensating mAs for the inverse square law. Method: A chest phantom with a simulated pathology was imaged at varying SIDs at a fixed mAs and kV. Observers visually compared the experimental images with a reference image using criteria divided into 3 categories: anatomy, noise and nodule image quality. A 2AFC program was used to display the images. A 5-point Likert scale was used ranging from ‘much worse’ to ‘much better’. PCXMC 2.0 84 was used to estimate effective dose. At each SID, a new mAs was calculated using inverse square law correction. Effective dose was estimated again. Results: Modal response for 2AFC visual evaluation of anatomy image quality was ‘equal’ for every SID, implying no observable change. Modal response for 2AFC visual evaluation of noise changed from ‘better’ to ‘worse’ as SID increased. Modal response for visual evaluation of nodule image quality was ‘worse’ at the smallest SIDs and ‘better’ at the largest SIDs. Decreasing SID from 190-100 cm: increased effective dose by 247%; compensating mAs reduced the effective dose by 36%. Conclusion: For variable SIDs used in this study, there is little perceived difference in anatomical image quality. Nodule image quality may decrease at shorter SIDs while effective dose increases. Therefore, when shorter SIDs are necessary due to lack of space, practitioners must be aware of patient dose.

Tese (doutorado) - Universidade Federal de Santa Catarina, Centro de Ciências Físicas e Matemáticas. Programa de Pós-Graduação em Física. O trabalho descreve quantitativa e qualitativamente a formação do feixe de raios X produzido por um tomógrafo computadorizado que opera no Hospital Regional de São José/SC. Para isso, dividiu-se a abordagem em 3 etapas: i) determinação criteriosa da mancha focal por meio de um código computacional que resolve a equação de Laplace e do movimento de elétrons em colisão com o ânodo; ii) cálculo analítico do espectro de radiação X emitido através de bremsstrahlung, partindo de primeiros princípios, a partir do processo de colisão do feixe de elétrons com o ânodo (sólido), incluindo a auto-atenuação dos fótons gerados; iii) uso do código PENELOPE (algoritmo de Monte Carlo) para a determinação do espectro total de radiação X emitido pelo ânodo. Em todas as etapas, os valores calculados estão muito próximos dos encontrados na literatura. A área da mancha focal simulada possui uma diferença menor que 5% em relação aos valores do fabricante, apesar das limitações do simulador. A equação proposta para o cálculo do espectro de bremsstrahlung emitido, incluindo-se a auto-atenuação, mostrou-se muito simples de calcular, pela ausência de variáveis empíricas ou funções multiparamétricas, em relação a outras apresentadas na literatura. Os resultados, comparados a outros gerados por equações semi-empíricas ou simulação por método Monte Carlo, também são excelentes, com diferenças abaixo de 5%.