Analysis of Fourier‐domain task‐based detectability index in tomosynthesis and cone‐beam CT in relation to human observer performance

Abstract

Purpose: Design and optimization of medical imaging systems benefit from accurate theoretical modeling that identifies the physical factors governing image quality, particularly in the early stages of system development. This work extends Fourier metrics of imaging performance and detectability index ͑dЈ͒ to tomosynthesis and cone-beam CT ͑CBCT͒ and investigates the extent to which dЈ is a valid descriptor of task-based imaging performance as assessed by human observers. Methods:The detectability index for tasks presented in 2D slices ͑d slice Ј ͒ was derived from 3D cascaded systems analysis of tomosynthesis and CBCT. Anatomical background noise measured in a physical phantom presenting power-law spectral density was incorporated in the "generalized" noise-equivalent quanta. Theoretical calculations of d slice Ј were performed as a function of total angular extent ͑ tot ͒ of source-detector orbit ranging 10°-360°under two acquisition schemes: ͑i͒ Constant angular separation between projections ͑constant-⌬͒, giving variable number of projections ͑N proj ͒ and dose vs tot and ͑ii͒ constant number of projections ͑constant-N proj ͒, giving constant dose ͑but variable angular sampling͒ with tot . Five simple observer models were investigated: Prewhitening ͑PW͒, prewhitening with eye filter and internal noise ͑PWEi͒, nonprewhitening ͑NPW͒, nonprewhitening with eye filter ͑NPWE͒, and nonprewhitening with eye filter and internal noise ͑NPWEi͒. Human observer performance was measured in 9AFC tests for five simple imaging tasks presented within uniform and power-law clutter backgrounds. Measurements ͑from 9AFC tests͒ and theoretical calculations ͑from cascaded systems analysis of d slice Ј ͒ were compared in terms of area under the ROC curve ͑A z ͒ Results: Reasonable correspondence between theoretical calculations and human observer performance was achieved for all imaging tasks over the broad range of experimental conditions and acquisition schemes. The PW and PWEi observer models tended to overestimate detectability, while the various NPW models predicted observer performance fairly well, with NPWEi giving the best overall agreement. Detectability was shown to increase with tot due to the reduction of out-of-plane clutter, reaching a plateau after a particular tot that depended on the imaging task. Depending on the acquisition scheme, however ͑i.e., constant-N proj or ⌬͒, detectability was seen in some cases to decline at higher tot due to tradeoffs among quantum noise, background clutter, and view sampling. Conclusions: Generalized detectability index derived from a 3D cascaded systems model shows reasonable correspondence with human observer performance over a fairly broad range of imaging tasks and conditions, although discrepancies were observed in cases relating to orbits intermediate to 180°and 360°. The basic correspondence of theoretical and measured performance supports the 1754 1754 Med. Phys. 38 "4…,

Three-dimensional cascaded system analysis of a 50µm pixel pitch wafer-scale CMOS active pixel sensor x-ray detector for digital breast tomosynthesis

“…It was demonstrated that a high spatial resolution of 10 mm -1 can be achieved using the DynAMITe detector with 50 μm pixel pitch (Zhao et al 2015b). In this study, the validated 2D cascaded system analysis model is extended to 3D in combination with the FBP reconstruction method (Zhao and Zhao 2008b, Gang et al 2011, Tward and Siewerdsen 2008). This 3D model previously validated, using the Siemens MAMMOMAT Inspiration direct conversion a-Se detector with a-Si:H TFT PPS readout under DBT conditions (Zhao et al 2009), is expected to accurately predict the 3D imaging performance of the CMOS APS detector.…”

“…In this study, the magnification factor was ignored. The aliased 3D NPS is given by (Zhao and Zhao 2008b, Tward and Siewerdsen 2008, Gang et al 2011 14 13 14 ,, ( , , ) ( , , )***III ( , , ) ( , , ) , , ,…”

“…Previously, 3D cascaded system analysis models have been developed and validated for both cone-beam computed tomography (CT) and DBT (Tward and Siewerdsen 2008, Zhao and Zhao 2008b, Zhao et al 2009, Hu and Zhao 2014, Siewerdsen and Jaffray 2003, Gang et al 2011. We have already reported a 2D cascaded system analysis model for a 50 μm pixel pitch CMOS APS detector named DynAMITe (Zhao et al 2015b).…”

CT protocol management: simplifying the process by using a master protocol concept

“…A scanner specific ${\mathrm{F}}_{\text{SS}}$ term can be calculated by keeping all controllable acquisition parameters constant and measuring the ratio of mA values required to obtain equal detectability. As mentioned previously, unless the resolution and noise texture remain unchanged when changing from scanner to scanner, a detectability metric that incorporates the influence of the MTF and NPS on object detection 19 , 20 should be used to match two different scanners. In place of such an approach, a qualitative assessment of high and low contrast resolution could be made using a Catphan (Models 500, 600, or 700, The Phantom Laboratory, Salem, NY) (or similar phantom).…”

“…If the noise structure changes, however, the noise (pixel standard deviation) may vary with minimal change in low contrast detectability. In these cases it is better to use a more advanced model observer‐based detectability metric that incorporates the influence of the MTF and NPS on object detection 19 , 20 …”