Department of Radiology, Medical University of Vienna, Waehringer Guertel 18-20, A-1090 Vienna, Austria.
Radiology. 2010 May;255(2):553-62. doi: 10.1148/radiol.10091096. Epub 2010 Mar 23.
To retrospectively assess the rate of detection of skull fractures at cranial computed tomography (CT) achieved with the use of curved maximum intensity projections (MIPs) compared with that achieved by reading transverse sections only.
The institutional review board approved this research and waived informed consent. A curved thin (3-mm) MIP of the skull cap and a curved thick (50-mm) MIP of the skull base were obtained from the cranial CT data in 200 consecutive patients with head trauma. Four radiologists (two residents without experience in cranial CT and two consultants) independently evaluated all cases. Each radiologist reported findings in 100 patients by using transverse sections only and findings in the other 100 patients by using the unfolded view. The radiologists were blinded to patient names, and patient and group orders were randomized. The results were compared with a standard of reference established by two experts from all prior reading results, all reconstructions, and high-spatial-resolution multiplanar reformats. Logistic regression with repeated measurements was used for statistical analysis.
The experts found 63 fractures in 30 patients. When transverse sections only were used, the mean patient-based fracture detection rate was 43% (13 of 30) for inexperienced and 70% (21 of 30) for experienced readers; with curved MIPs, the rates were 80% (24 of 30) and 87% (26 of 30), respectively. Overall sensitivity was higher with curved MIPs (P < .001); specificity was higher with transverse sections (P < .001).
Curved MIPs enable a significantly higher fracture detection rate than transverse sections. They also considerably close the experience gap in fracture detection rate between residents and experts.
回顾性评估使用曲面最大密度投影（MIP）检测颅骨 CT 颅骨骨折的检出率与仅阅读横断面相比的差异。
本研究经机构审查委员会批准，并豁免了知情同意。从 200 例颅脑外伤患者的颅脑 CT 数据中获得颅骨帽状曲面薄（3mm）MIP 和颅底曲面厚（50mm）MIP。4 名放射科医生（2 名无颅脑 CT 经验的住院医生和 2 名顾问）独立评估所有病例。每位放射科医生仅通过横断面报告 100 例患者的检查结果，通过展开视图报告另外 100 例患者的检查结果。放射科医生对患者姓名、患者和组别的顺序均不知情，且患者和组别的顺序是随机的。将结果与由两位专家根据所有先前的阅读结果、所有重建和高空间分辨率多平面重建确定的标准进行比较。采用重复测量的逻辑回归进行统计分析。
专家在 30 名患者中发现 63 处骨折。仅使用横断面时，无经验的读者基于患者的骨折检出率平均为 43%（30 例中的 13 例），有经验的读者为 70%（30 例中的 21 例）；使用曲面 MIP 时，检出率分别为 80%（30 例中的 24 例）和 87%（30 例中的 26 例）。曲面 MIP 的总体敏感性更高（P<.001）；特异性以横断面更高（P<.001）。
曲面 MIP 能显著提高骨折检出率，并且在骨折检出率方面，明显缩小了住院医生和专家之间的经验差距。
