Optimization of the imaging quality of 64-slice CT acquisition protocol using Taguchi analysis: A phantom study.

Abstract

In this study, the phantom imaging quality of 64-slice CT acquisition protocol was quantitatively evaluated using Taguchi. The phantom acrylic line group was designed and assembled with multiple layers of solid water plate in order to imitate the adult abdomen, and scanned with Philips brilliance CT in order to simulate a clinical examination. According to the Taguchi L8(2(7)) orthogonal array, four major factors of the acquisition protocol were optimized, including (A) CT slice thickness, (B) the image reconstruction filter type, (C) the spiral CT pitch, and (D) the matrix size. The reconstructed line group phantom image was counted by four radiologists for three discrete rounds in order to obtain the averages and standard deviations of the line counts and the corresponding signal to noise ratios (S/N). The quantified S/N values were analyzed and the optimal combination of the four factor settings was determined to be comprised of (A) a 1-mm thickness, (B) a sharp filter type, (C) a 1.172 spiral CT pitch, and (D) a 1024×1024 matrix size. The dominant factors included the (A) filter type and the cross interaction between the filter type and CT slice thickness (A×B). The minor factors were determined to be (C) the spiral CT pitch and (D) the matrix size since neither was capable of yielding a 95% confidence level in the ANOVA test.