Performance of multi-slice channelized Hotelling observer for low-contrast signal detection in simulated CT data(PDF)
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- Issue:
- 2018年第12期
- Page:
- 1462-1467
- Research Field:
- 医学信号处理与医学仪器
- Publishing date:
Info
- Title:
- Performance of multi-slice channelized Hotelling observer for low-contrast signal detection in simulated CT data
- Author(s):
- WU Wangjiang; LI Yan; YANG Zhi
- College of Biomedical Engineering, Capital Medical University, Beijing 100069, China
- Keywords:
- ; Keywords: channelized Hotelling observer; image quality assessment; low contrast detection; computed tomography
- PACS:
- R312
- DOI:
- DOI:10.3969/j.issn.1005-202X.2018.12.018
- Abstract:
- Abstract: Objective To compare the performances of conventional single-slice channelized Hotelling observer (ssCHO) and multi-slice channelized Hotelling observer (msCHO) in the detection of low-contrast signals based on simulated CT data. Methods Three-dimensional (3D) signal-present and signal-absent images were generated from computer simulation. The amplitude and size of signals were changed to investigate the performances of ssCHO and msCHO in the detection of different signals, and the percentage of correct detection (PC) was used to quantitatively evaluate the performances of ssCHO and msCHO in the detection of low-contrast signals. Results With the signal size of 2.5 mm and signal amplitudes of 2.5, 5.0, 10.0 HU respectively, the PC was saturated when msCHO used 10 slices of 3D image (contrast object was fully covered by these slices), and the maximum PC were 0.79, 0.96, and 1.00 for msCHO, and 0.67, 0.78, and 0.93 for ssCHO, respectively. Compared to ssCHO, msCHO achieved the performance improvement by 17.91%, 23.08%, and 7.53%, respectively. When the signal amplitude remained at 2.5HU, and the signal size were 2.0, 2.5, 3.5, 5.0, 10.0 mm, respectively, the PC of msCHO using 8, 10, 8, 6, 3 slices of 3D image reached the maximum (0.69, 0.79, 0.95, 0.99, and 1.00 mm, respectively), improved by 9.5%, 17.9%, 28.4%, 15.1%, and 4.2% compared to 0.63, 0.67, 0.74, 0.86, and 0.96 of ssCHO. Conclusion msCHO using more information for decision-making, therefore, is more reliable in low-contrast signal detection than ssCHO. However, the performance of msCHO depends on the selection of the slices. The optimal number of slices should be determined based on the factors of signal amplitude, signal size, and background noise.
Last Update: 2018-12-26