Intercomparison of methods for image quality characterization. I. Modulation transfer function

• Published in: Medical physics (Lancaster) 2006-05, Vol.33 (5), p.1454-1465
• Main Authors: Samei, Ehsan, Ranger, Nicole T., Dobbins, James T., Chen, Ying
• Format: Article
• Language: English
• Subjects: ACCURACY; Algorithms; APERTURES; BIOMEDICAL RADIOGRAPHY; COLLIMATORS; digital radiography; edge; FOURIER ANALYSIS; glare; IEC 62220; image quality; IMPACT PARAMETER; linear systems analysis; MODULATION; Modulation transfer function (MTF); Quality Assurance, Health Care - methods; Radiographic Image Enhancement - methods; Radiographic Image Interpretation, Computer-Assisted - methods; RADIOLOGY AND NUCLEAR MEDICINE; Reproducibility of Results; resolution; Sensitivity and Specificity; slit; SPATIAL RESOLUTION; SYSTEMS ANALYSIS
• ISSN: 0094-2405; 2473-4209
• DOI: 10.1118/1.2188816

The modulation transfer function (MTF) and the noise power spectrum (NPS) are widely recognized as the most relevant metrics of resolution and noise performance in radiographic imaging. These quantities have commonly been measured using various techniques, the specifics of which can have a bearing on the accuracy of the results. As a part of a study aimed at comparing the relative performance of different techniques, in this paper we report on a comparison of two established MTF measurement techniques: one using a slit test device [ Dobbins , Med. Phys. 22 , 1581-1593 ( 1995 ) ] and another using a translucent edge test device [ Samei , Med. Phys. 25 , 102-113 ( 1998 ) ], with one another and with a third technique using an opaque edge test device recommended by a new international standard (IEC 62220-1, 2003). The study further aimed to substantiate the influence of various acquisition and processing parameters on the estimated MTF. The slit test device was made of 2 mm thick Pb slabs with a 12.5 μ m opening. The translucent edge test device was made of a laminated and polished Pt 0.9 Ir 0.1 alloy foil of 0.1 mm thickness. The opaque edge test device was made of a 2 mm thick W slab. All test devices were imaged on a representative indirect flat-panel digital radiographic system using three published beam qualities: 70 kV with 0.5 mm Cu filtration, 70 kV with 19 mm Al filtration, and 74 kV with 21 mm Al filtration (IEC-RQA5). The latter technique was also evaluated in conjunction with two external beam-limiting apertures (per IEC 62220-1), and with the tube collimator limiting the beam to the same area achieved with the apertures. The presampled MTFs were deduced from the acquired images by Fourier analysis techniques, and the results analyzed for relative values and the influence of impacting parameters. The findings indicated that the measurement technique has a notable impact on the resulting MTF estimate, with estimates from the overall IEC method 4.0 % ± 0.2 % lower than that of Dobbins and 0.7 % ± 0.4 % higher than that of Samei averaged over the zero to cutoff frequency range. Over the same frequency range, keeping beam quality and limitation constant, the average MTF estimate obtained with the edge techniques differed by up to 5.2 % ± 0.2 % from that of the slit, with the opaque edge providing lower MTF estimates at lower frequencies than those obtained with the translucent edge or slit. The beam quality impacted the average estimated MTF by as muc