Abstract:

Objective　The differences in the detection of pulmonary nodules between conventional thin-slice images and high resolution combined with maximum intensity projection (HR-MIP) reconstruction images were compared, and the detection efficiency of HR-MIP for pulmonary nodules was evaluated. Methods　A total of 90 patients who underwent chest CT examination in Chu Hsien-I Memorial Hospital from December 2021 to June 2022 were randomly selected, including 52 males and 38 females, aged 18-82 (43±12.5) years. Conventional thin-slice images with a thickness of 2 mm and HR-MIP reconstruction images with a thickness of 4 mm were used to observe the situation of pulmonary nodules, and the nodules were divided into <5 mm and ≥5 mm groups according to their size. The number, size, and nature of nodules (solid, part-solid, and ground glass) were recorded by two radiologists, and the time spent reading the images of each patient was recorded. The differences of nodule detection rate and time consumption were compared between the conventional and HR-MIP groups. Paired t test and χ² test were used. Results　There were 313 nodules in the 90 patients, with 255 nodules <5 mm and 58 nodules ≥5 mm. The total numbers of nodules detected in the conventional and HR-MIP groups were 253 (80.1%) and 285 (91.0%), respectively. The detection rate of HR-MIP was higher than that of conventional thin-slice images, with a statistically significant difference (χ²=13.523, P<0.001). The numbers of <5 mm and ≥5 mm nodules detected in the conventional and HR-MIP groups were 208 vs.235 and 45 vs.50, respectively. There was a statistically significant difference in the <5 mm group (χ²=12.517, P<0.001); there was no statistically significant difference in the ≥5 mm group (χ²=1.408,P=0.237). The numbers of solid, part-solid, and ground glass density nodules were 261, 22, and 30, respectively. The numbers of solid, part-solid, and ground glass density nodules in the conventional and HR-MIP groups were 210 vs.245, 15 vs.17, and 28 vs.23, respectively. There was a statistically significant difference in the detection rate of solid nodules between the two groups (χ²=34.812, P<0.001), but there were no statistically significant differences in the part-solid and ground glass nodules (χ²=0.631, P=0.407; χ²=1.231, P=0.224). The time consumption of the conventional group was (215.3±23.8) s, and that of the HR-MIP group was (133.2±19.1) s, with a statistically significant difference (t=18.013,P<0.001). Conclusions　HR-MIP reconstruction can improve the detection rate of small solid pulmonary nodules and shorten the reading time to improve diagnostic efficiency. However, part-solid and ground glass nodules still need to be carefully observed.

Key words:

CT, High resolution, Maximum intensity projection, Pulmonary nodules, Lung cancer

摘要：

目的　对比常规薄层图像和高分辨率联合最大密度投影（high resolution combined with maximum intensity projection，HR-MIP）重建图像在检出肺内结节方面的差异，评价HR-MIP对肺内结节的检出效能。方法　随机选取天津医科大学朱宪彝纪念医院2021年12月至2022年6月期间行胸部CT检查的患者共90例，其中男性52例，女性38例，年龄18～82（43±12.5）岁。分别利用2 mm层厚的常规薄层图像和HR-MIP（厚度4 mm）重建图像观察肺内结节的情况，根据结节大小分为<5 mm和≥5 mm两组。两位诊断医师分别对两组图像进行阅片并记录结节的数量、大小及性质（实性、部分实性、磨玻璃），同时记录阅读每例患者阅片时所耗费的时间。比较常规薄层图像和HR-MIP在结节检出率及耗费时间方面的差异。采用配对t检验、χ²检验进行统计学分析。结果　90例患者共有结节313个，<5 mm 255个、≥5 mm 58个，其中常规薄层图像和HR-MIP分别检出结节的总数为253个（80.1%）、285个（91.0%），HR-MIP检出率高于常规薄层图像，差异有统计学意义（χ²=13.523，P<0.001）；其中常规薄层图像和HR-MIP检出<5 mm及≥5 mm结节的数量分别为（208比235）和（45比50），<5 mm组差异有统计学意义（χ²=12.517，P<0.001），≥5 mm组差异无统计学意义（χ²=1.408，P=0.237）。结节性质方面：实性、部分实性及磨玻璃密度结节分别有261、22、30个；常规薄层图像和HR-MIP检出实性、部分实性和磨玻璃密度结节分别为（210比245）、（15比17）和（28比23），其中两种方法实性结节检出率差异有统计学意义（χ²=34.812，P<0.001），部分实性和磨玻璃密度结节检出率差异均无统计学意义（χ²=0.631，P=0.407；χ²=1.231，P=0.224）。常规薄层图像耗时为（215.3±23.8）s，HR-MIP耗时为（133.2±19.1）s，差异有统计学意义（t=18.013，P<0.001）。结论　利用HR-MIP重建后阅片能够提高肺内实性小结节的检出率并且可以缩短阅片时间、提高诊断效率，但是对于部分实性及磨玻璃密度小结节仍需仔细观察。

关键词:

计算机断层成像, 高分辨率, 最大密度投影, 肺结节, 肺癌