Automated assessment of low contrast sensitivity for CT systems using a model observer
A constructed 38 mm CT slice of the CTP515 module. Each contrast group and the low contrast object diameters have been tagged in the figure for the supraslice region.
Mask consisting of templates with respect to each low contrast object in the Catphan module.
Detectability index d′ and proportion correct PC (left and right column, respectively) as a function of low contrast circle diameters of the Catphan phantom for the 1% contrast group for different kV (A, D), mAs (B, E), and reconstruction filters (C, F). The parameters kept constant in each case appear as a legend on the figures to identify them as image series in Table I. The black line on PC graphs (D, E, F) represents the visibility threshold criterion applied (PC = 75%).
Smallest object diameters visible (with a PC = 75%), λ, obtained with the psychometric fitting for all contrast groups for different kV values. Lines are a mere data connector in the graph.
Smallest object diameters visible (with a PC = 75%), λ, obtained with the psychometric fitting for all contrast groups for different mAs values. Lines are a mere data connector in the graph.
Smallest object diameters visible (with a PC = 75%), λ, obtained with the psychometric fitting for all contrast groups for different reconstruction filters. Lines are a mere data connector in the graph.
Psychometric fitting functions of PC as a function of object size for different mAs and 1% contrast group. Fitting parameters, λ and f, have been included in the legend for each image series.
PC curves normalized to the reference dose (16.9 mGy) for different tube charge per rotation values. The black line represents the global psychometric fit for all data.
PC curves as a function of object size for the average expert (lines, Exp) and nonexpert (dots, NE) observers for varying kV (A), mAs (B), and reconstruction filter (FC12 and FC50, soft body and lung, respectively, (C). Observers were not able to score images related to the sharp filters (FC81, FC53). Note that PC values run from 0 to 1.
PC curves as a function of kV (left column) and mAs (right column) for different object sizes for the average expert observer (first row, (A) and (D), 1% contrast series) and the LCD software (second and third row, 1% and 0.5% contrast series, respectively). Note that PC values run from 0 to 1 for human observer (first row) and from 0.5 to 1 for the software (second and third row).
Overview of the acquisition and reconstruction parameters.
Values of D1 (largest diameter object) and its corresponding PC value (PC1) which fail the PC ≥ 75% criteria to be considered visible, for the three contrast groups and all image series.
Summary of the results of the individual psychometric fittings for all contrast groups and acquisition conditions considered. Parameter λ represents the smallest object diameter which reached PC ≥ 75% and so, considered visible. The error associated with this parameter is also shown as a (%).
Lambda values (λNorm) obtained performing the psychometric fits taking as f the value obtained with the psychometric fit of the normalized data to the reference dose (CTDIvol ref = 16.9 mGy) for all contrast groups and different mAs values. The relative differences (ɛrelative) between lambda values obtained for the individual fits (λIndiv.Fit) for each series (with lambda and f unbound) and this method, are also shown.
Measured and nominal contrast values (%) and contrast-to-noise ratio (CNR) for the low contrast module of the Catphan phantom.
Article metrics loading...
Full text loading...