| ObjectiveThe objective of our study is firstly to evaluate the intra- and inter-obsever variability of solitary pulmonary nodule (SPN) enhancement assessment on computed tomography perfusion imaging combined with dynamic enhanced scanning using two different methods:whole voxel threshold segmentation versus region of interest(ROI) respectively. Secondly, we compare the repeatability between these two methods. Furthermore, a preliminary evaluation of the voxel index(VI) measured by threshold segmentation is made to distinguish malignant SPN from those of benign and inflammatory ones.Materials and Methods40 patients with SPN confirmed by pathology and clinical information were included in our study. Of them 17 were malignant nodules,15 were benign nodules and 8 were inflammatory nodules. All 40 patients underwent pre-enhanced and perfusion MSCT combined with dynamic enhanced scan. Scan protocol:(1) Pre-enhanced:2.5mm thickness,2.5mm interval; 120 kV,100mA; helical scanning mode. (2) Enhanced:2.5mm thickness,2.5mm interval; 120kV,100mA; axial scanning mode. Iodinated, nonionic contrast material (Omnipaque,350/370 mgI/ml) was administered intravenously via antecubital vein at the rate of 4.3-4.5ml/sec for a total volume of 50/80ml by using power auto-injector. Perfusion axial scan throughout whole SPN, which triggered at 7s after injection, was performed by separated two series (minimum interval time of 7.5s), with Z-axis cover 40mm, single exposure time 0.4s, interval time 1.0s, and a total scan of 29 times. Thereafter, additional dynamic series of images throughout whole SPN were obtained at 60s,90s, 120s,150s,180s,240s,300s,7min after the injection.Image data of SPN were transferred to AW 4.4 workstation, The CT image of whole SPN at different time point were reconstructed using maximum intensity projection (MIP) algorithms. The SPNs were firstly segmented from adjacent vessels and pleura structure with the 3D cutting tool of scalpel in the toolbar. SPNs of whole voxel were Segmented according to the settlement of different CT value threshold range with functional tool of 3D color ROI. Threshold ranges were set as follow:-120～0 Hu,1～15 Hu,16～30 Hu,31-45 Hu,46～60 Hu,61-75 Hu and 76-120 Hu. The pre- and post-enhanced volume constituent ratio of each threshold interval were recorded and named as voxel index (Ⅵ). The CT values in different phases were measured as well using the traditional method of ROI. Two different readers measured these parameters separately and one of them measured twice with one week interval. The intra- and inter-observer variability of the two methods were estimated and compared. The characteristic of voxel index (Ⅵ) change between malignant and benign/inflammatory nodules were explored. Preliminary evaluation is made to distinguish malignant SPN from those of benign/inflammatory ones with the method.Result1. A good intra-and inter-observer agreement was found onⅥmeasurement. The mean difference within observers was -0.0038%, and 95% limits of agreement was-0.79%～0.79%; the mean difference between observers is 0.0035%, and 95% limits of agreement:was -1.24%～1.24%.2. The mean difference within observers using traditional ROI CT value measurement was -0.1724Hu and 95% limits of agreement was -6.22Hu～5.87Hu; The mean difference between observers using traditional ROI CT value measurement was 0.0831 Hu and 95% limits of agreement was -8.55Hu～8.72Hu.3.Ⅵmeasurement showed better intra- and inter-observer agreement than traditional ROI CT value measurement. (relative difference:0.02±0.04 vs 0.13±0.22; P<0.001)4. The absoluteⅥaverage difference was significantly different at different delay time points, and in different threshold segmentations, between malignant (P<0.001), benign (P<0.001) and inflammatory (P<0.001) groups. The mean absoluteⅥchanges were plotted with time points and different threshold intervals, and evident differences in shapes of histograms were observed between different SPN groups.5. Groups of rank sum test results and histograms analysis showed that the most significant different absoluteⅥaverage difference of different groups of SPN appeared in the threshold segmentation of 76-120Hu. 6. when choosing the threshold segmentation of 76-120Hu, the time-absolute VI change curves showed evident different characteristics in malignant, benign and inflammatory SPN groups.Conclusion1. SPN adnacent to vessels and pleura can be accurately segmented by using the 3D cutting tool of scalpel.2. Both volume threshold segmentation VI measurement and traditional ROI CT value measurement showed relatively satisfactory intra-observer and inter-observer agreement.3. Volume threshold segmentation VI measurement showed better intra-observer and inter-observer agreement than the method of traditional ROI CT value measurement.4. Analysis of the characteristics of histograms of absolute VI difference in different threshold segmentation and Time-Absolute VI difference curve in the threshold of 76-120HU contributes to the diagnosis and differential diagnosis of SPN. |
PDF Full Text Request