The baseline Lung Cancer Screening protocol was selected from the American Association of Medical Physicists (AAPM) protocol page for the CT system (Aquilion ONE Prism Edition, Canon Medical Systems, Otawara, Japan)5. This baseline protocol uses the standard x-ray beam filter with conventional HIR. Because the AAPM protocol uses automatic exposure control, the recommended AEC setting was used to establish the fixed tube current used for each phantom evaluation.
Low Contrast Detectability (LCD) was examined using the 10HU, 6-15mm test objects in the low contrast detectability module of the 20cm CatphanTM 600 phantom with the supplemental 25-35cm body ellipse attached to mimic patient shape and attenuation. The 10HU, 6mm6 combination was selected as the lowest clinically relevant contrast/diameter combination for Lung Cancer Screening. The protocol details are given in Table 1. For LCD, only the Body reconstruction (“Body Sharp”) was used.
Low contrast detectability was assessed via a non-prewhitening model observer. The experimental set-up was based on a two-alternative forced choice scenario, utilizing a signal-known-exactly, background-known-exactly (SKE/BKE) format. For each test object, 780 signal-present and 780 signal-absent ROIs were generated. The signal-absent ROIs were located at the uniform background near the object where background bias is less than 6.25%. The output variable for the observer study was the detectability index (d’), as measured directly from the model observer’s distribution of responses to the signal-present and signal-absent ROIs.
Noise, Noise Texture, and CNR: Noise magnitude and CNR were examined using the 20cm Gammex 464 phantom with a supplemental 25-35cm body ellipse attached to mimic patient shape and attenuation. Noise texture was evaluated by calculating the Noise Power Spectrum (NPS) in a 32cm cylindrical water phantom. The protocol details are given in Table 2. For these tests, both the Body reconstruction (“Body Sharp”) and Lung reconstruction were used.
The noise magnitude was assessed via the standard deviation (SD) of pixel values in an ROI, while the NPS was assessed via standard Fourier techniques and normalized to an AUC of unity.
The CNR ratio was measured by comparing the Acrylic test object HU to the water equivalent background HU and dividing by the SD.
Spatial Resolution was examined using the sensitometry module of the 20cm Catphan 600 phantom. To minimize the noise in the measurement, no supplement attachment was used. Both a high contrast object (Teflon) and a lower contrast object (Polystyrene) were used to generate Edge Spread Functions for conversion to a Modulation Transfer Function (MTF) via standard Fourier approaches.7 For these tests, both the Body reconstruction (“Body Sharp”) and Lung reconstruction were used. The protocol details are given in Table 3.
A clinical example of the CNR properties is provided in Figure 4, which shows a small lung nodule well-demonstrated in the upper right lung using a low dose, 0.9mGy, chest CT combining a beam-shaping Silver filter with DLR.