Volterrani L, Mazzei M A, Scialpi M, Carcano M, Carbone S F, Ricci V, Guazzi G, Lupattelli L
Dipartimento di Patologia Umana ed Oncologia, Sezione di Radiologia Oncologica, Medicina Nucleare e Radioterapia, Università degli Studi di Siena, Policlinico Santa Maria alle Scotte, Viale Bracci 2, I-53100, Siena (SI), Italy.
Radiol Med. 2006 Apr;111(3):343-54. doi: 10.1007/s11547-006-0033-4. Epub 2006 Apr 11.
The purpose of this study was to test the reproducibility of the three-dimensional (3D) Advanced Lung Analysis software (3D-ALA, GE Healthcare) in the estimation of pulmonary nodule volume.
We retrospectively reviewed the unenhanced multislice CT scans (Lightspeed Pro 16 GE) of 77 patients with a solitary pulmonary nodule (n=71) or metastatic pulmonary disease (n=6). A total of 103 pulmonary nodules (19 well-circumscribed, 45 juxtavascular and 39 juxtapleural) were analysed grouped into five classes based on diameter: <5 mm, 10 nodules (9.7%); >or=5 to <10 mm, 25 nodules (24.2%); >or=10 mm to <15 mm, 41 nodules (39.8%); >or=5 to <18 mm, 14 nodules (13.6% ); >or=8 to <30 mm, 13 nodules (12.62%). The following acquisition parameters were used: slice thickness 0.625 mm, reconstruction interval 0.4 mm, pitch 0.562:1, 140 kV, 300 mAs, field of view 13 cm, bone kernel. For each of the 103 nodules three, 3D volume measurements were obtained by the 3D-ALA software. The reproducibility of nodule segmentation was evaluated according to a visual score (1=optimal, >or=95%; 2=fair, 90-95%; 3=poor, <or=90%) by three observers working in consensus. The reproducibility of volume estimation was evaluated by comparing all 3D volume measurements and all segmentations obtained for each pulmonary nodule using the ANOVA test.
ALA-1 software allowed segmentation in all nodules (type 1 segmentation n=43, type 2 n=35, type 3 segmentation n=25). ALA-1 provided an identical 3D volume measurement in 62 nodules: [16 out of 19 well circumscribed (84.2%), 31 out of 45 juxtavascular (68.8%), 15 out of 39 juxtapleural (38.4%)]. Repeatability of 3D volume measurement was not possible in 41 out of 103 nodules [3 out of 19 (15.7%) well-circumscribed, 14 out of 45 (31.1%) juxtavascular, 24 out of 39 (61.5%) juxtapleural]. Among the 41 nodules with nonrepeatable 3D volume measurement, segmentation was scored as 1 in 2 out of 41 (4.8%), as 2 in 15 out of 41 (36.5%) and as 3 in 24 out of 41 (58.5%). The difference between the mean volume on three measurements and each type of nodule was not statistically significant (p>0.05).
Three-dimensional volume measurement with ALARiassunto 1 software is reproducible for all nodules as regards dimension and site. ALA-1 software provided a good and reproducible volume measurement in well-circumscribed and most juxtavascular nodules. Volumetric evaluation and reproducibility of volume estimation in juxtapleural pulmonary nodules, particularly those adjacent to diaphragmatic pleura, is inadequate, and software improvement is needed.
本研究旨在测试三维（3D）高级肺部分析软件（3D-ALA，通用电气医疗集团）在估计肺结节体积方面的可重复性。
我们回顾性分析了77例患有孤立性肺结节（n = 71）或转移性肺部疾病（n = 6）患者的非增强多层CT扫描（Lightspeed Pro 16，通用电气）。总共分析了103个肺结节（19个边界清晰、45个血管旁和39个胸膜旁），根据直径分为五类：<5mm，10个结节（9.7%）；≥5至<10mm，25个结节（24.2%）；≥10至<15mm，41个结节（39.8%）；≥15至<18mm，14个结节（13.6%）；≥18至<30mm，13个结节（12.62%）。使用以下采集参数：层厚0.625mm，重建间隔0.4mm，螺距0.562:1，140kV，300mAs，视野13cm，骨算法。对于103个结节中的每一个，通过3D-ALA软件获得三次3D体积测量值。由三位观察者共同评估结节分割的可重复性，根据视觉评分（1 = 最佳，≥95%；2 = 尚可，90 - 95%；3 = 差，≤90%）进行。通过使用方差分析比较每个肺结节获得的所有3D体积测量值和所有分割结果，评估体积估计的可重复性。
ALA-1软件能够对所有结节进行分割（1型分割n = 43，2型分割n = 35，3型分割n = 25）。ALA-1在62个结节中提供了相同的3D体积测量值：[19个边界清晰的结节中有16个（84.2%），45个血管旁结节中有31个（68.8%），39个胸膜旁结节中有15个（38.4%）]。103个结节中有41个无法进行3D体积测量的重复性分析[19个边界清晰的结节中有3个（15.7%），45个血管旁结节中有14个（31.1%），39个胸膜旁结节中有24个（61.5%）]。在41个无法进行3D体积测量重复性分析的结节中，分割评分为1的有41个中的2个（4.8%），评分为2的有41个中的15个（36.5%），评分为3的有41个中的24个（58.5%）。三次测量的平均体积与每种类型结节之间的差异无统计学意义（p>0.05）。
就尺寸和位置而言，使用ALA-1软件进行三维体积测量对所有结节具有可重复性。ALA-1软件在边界清晰和大多数血管旁结节中提供了良好且可重复的体积测量。胸膜旁肺结节，尤其是那些靠近膈胸膜的结节，体积评估和体积估计的可重复性不足，需要对软件进行改进。
