对比噪声比
材料科学
成像体模
层析合成
光学
计算机射线照相术
射线照相术
医学影像学
点间距
像素
核医学
乳腺摄影术
物理
图像质量
医学
放射科
计算机科学
癌症
人工智能
乳腺癌
内科学
图像(数学)
作者
Jeffrey C. Hammonds,R. R. Price,Edwin F. Donnelly,David R. Pickens
摘要
Purpose: A laboratory‐based phase‐contrast radiography/tomosynthesis imaging system previously (Med. Phys. Vol. 38, 2353 May 2011) for improved detection of low‐contrast soft‐tissue masses was used to evaluate the sensitivity for detecting the presence of thin layers of corrosion on aluminum aircraft structures. Methods: The evaluation utilized a test object of aluminum (2.5 inch × 2.5 inch × 1/8 inch) on which different geometric patterns of 0.0038 inch thick anodized aluminum oxide was deposited. A circular area of radius 1 inch centered on the phantom's midpoint was milled to an approximate thickness of 0.022 inches. The x‐ray source used for this investigation was a dual focal spot, tungsten anode x‐ray tube. The focal used during the investigation has a nominal size of 0.010 mm. The active area of the imager is 17.1 cm × 23.9 cm (2016 × 2816 pixels) with a pixel pitch of 0.085 mm. X‐ray tube voltages ranged from 20–40 kVp and source‐ to‐object and object‐to‐image distances were varied from 20–100 cm. Performance of the phase‐contrast mode was compared to conventional absorption‐based radiography using contrast ratio and contrast‐to‐noise ratios (C/N). Phase‐contrast performance was based on edge‐enhancement index (EEI) and the edge‐enhancement‐to‐noise (EE/N) ratio. Results: for absorption‐based radiography, the best C/N ratio was observed at the lowest kVp value (20 kVp). The optimum sampling angle for tomosynthesis was +/− 8 degrees. Conclusions: Comparing C/N to EE/N demonstrated the phase‐contrast techniques improve the conspicuity of the oxide layer edges. This work provides the optimal parameters that a radiographic imaging system would need to differentiate the two different compounds of aluminum. Subcontractee from Positron Systems Inc. (Boise, Idaho) through United States Air Force grant (AF083‐225).
科研通智能强力驱动
Strongly Powered by AbleSci AI