Performance Evaluation of a Commercial CMOS Detector for Dynamic High-Resolution Imaging
A Jain*, C Ionita, D Bednarek, S Rudin,Toshiba Stroke and Vascular Research Center, Univ. at Buffalo, Buffalo, NYSU-E-I-22 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: Present state-of-the art flat-panel detectors have limitations in terms of noise and resolution for dynamic imaging. With recent developments, CMOS-based detectors show promise to fulfill the ever increasing need for better image quality. Here, we evaluated a CMOS-based detector by measuring its sensitivity, MTF, DQE and INEE.
Methods: The detector (Hamamatsu Model # C10901D-40) featured 12-bit image acquisition, CsI phosphor, 100 μm detector pixel pitch and an active area of 9.7cm X 6.7cm and had a maximum frame rate of 30 fps. The sensitivity and instrumentation noise equivalent exposure (INEE) were calculated using flat-field images taken at various exposures (4.2 μR to 334 μR). The detector resolution was evaluated by measuring the modulation transfer function (MTF) using a tilted edge. The noise power spectra (NPS) and detective quantum efficiency (DQE) were measured using the RQA5 spectrum for three different exposures (4.2 μR, 16.5 μR and 174 μR).
Results: The signal vs. exposure curve showed excellent linearity of the detector and the sensitivity of the detector was found to be 5.8 DN/μR. The slope and the intercept of the variance vs. exposure curve were used to calculate the INEE of the detector, which was found to be 7.46 μR. The measured MTF values were 0.61, 0.21 and 0.07 at 10%, 50% and 90% of the Nyquist frequency of 5 cycles/mm. The corresponding values of the DQE at these indicated frequencies were 0.47(0.56), 0.19(0.32) and 0.03(0.09) at 4.2 μR (16.5 μR).
Conclusion: The imaging performance of Hamamatsu detector was evaluated in terms of sensitivity,
linear system metrics and INEE. The INEE of this detector appears higher than desirable for
fluoroscopy, but the detector shows good DQE at high exposure when the effect of the high
instrumentation noise is not significant.
(Support: NIH grants R01EB002873 and R01EB008425)
Funding Support, Disclosures, and Conflict of Interest: NIH grants R01EB002873 and R01EB008425