Purpose
Extracting image features can predict the prognosis and treatment effect of non-small cell lung cancer, which has been increasingly confirmed. However, the specific operation using 3D-Slicer still lacks standardization. For example, image segmentation is manually performed based on the lung window or automatically performed through the mediastinal window. The images used for feature extraction are either enhanced or plain scanned. It is questionable whether these influencing factors will affect the extraction results and which results will be affected. This article intends to preliminarily explore the above issues.
Methods
This article downloaded images of 22 patients with lung cancer from The Cancer Imaging Archive (TCIA), including 11 cases of adenocarcinoma and 11 cases of squamous cell carcinoma. It is required that the lung lesions have clear boundaries and have plain and enhanced images. Draw the region of interest on the lung window and mediastinal window of the plain scan image, and then draw the region of interest on the lung window and mediastinal window of the enhanced image. Manual drawing is used on the lung window, and automatic drawing is used on the mediastinal window with the CT value between − 150-500HU). 22 patients, each with 4 sets of imaging features, extracted a total of 88 sets of imaging features, each containing 107 features entered the analysis. Firstly, analyze the image features of the original sequence and perform the Shapiro test. If it follows a normal distribution, perform an analysis of variance. If it does not follow a normal distribution, perform the Friedman test. Compare the significantly different image features pairwise. Then, according to the pathology, they are divided into two groups. First, Shapiro test is performed. If the distribution is normal, independent sample T-test is performed. If the distribution is not normal, Wilcoxon rank test is performed.
Results
Four sets of influencing features were extracted for each patient, namely the plain lung window group, plain mediastinal window group, enhanced lung window group, and enhanced mediastinal window group. A total of 88 sets of imaging features were extracted, with 107 features in each group. Among them, 33 features showed significant differences. Continuing with pairwise repeated testing, it was found that there were 2 significant differences between enhanced and plain lung windows, namely original-glcm-Imc1,original-ngtdm-Complexity. There were 12 significant differences between enhanced lung windows and plain mediastinal windows, namely original-firstorder-10Percentile, original-firstorder-Energy, original-firstorder-Skewness, original-glcm-Imc1, original-glszm-SmallAreaHighGrayLevelEmphasis, original-glszm-SmallAreaLowGrayLevelEmphasis, original-glszm-ZoneEntropy, original-glszm-ZonePercentage, original-ngtdm-Busyness, original-ngtdm-Complexity, original-ngtdm-Contrast, original-ngtdm-Strength. There is one significant difference between plain scanning and enhancement mediastinal window, which is the original-firstorder-90Percentile. There are 14 significant differences between the plain lung window and the enhanced mediastinal window groups, including original-firstorder-10Percentile, original-firstorder-Energy, original-glcm-Idmn, original-gldm-DependenceVariance, original-glrlm-LowGrayLevelRunEmphasis, original-glszm-SmallAreaHighGrayLevelEmphasis, original-glszm-SmallAreaLowGrayLevelEmphasis, original-glszm-ZoneEntropy, original-glszm-ZonePercentage, original-glszm-ZoneVariance, original-ngtdm-Busyness, original-ngtdm-Complexity, original-ngtdm-Contrast, original-ngtdm-Strength. There are 14 significant differences between the lung window and the mediastinal window in the plain scan, which are the original firststorder-10Percentile, original-firstorder-Energy, original-firstorder-Skewness, original-glcm-Idmn, original-gldm-DependenceVariance, original-glszm-SmallAreaHighGrayLevelEmphasis, original-glszm-SmallAreaLowGrayLevelEmphasis, original-glszm-ZoneEntropy, original-glszm-ZonePercentage, original-glszm-ZoneVariance, original-ngtdm-Busyness, original-ngtdm-Complexity, original-ngtdm-Contrast, original-ngtdm-Strength. There are 13 significant differences between the enhanced lung window and the mediastinal window, which are original-firstorder-10Percentile, original-firstorder-Energy, original-glcm-Imc1, original-glrlm-LongRunHighGrayLevelEmphasis, original-glrlm-LowGrayLevelRunEmphasis, original-glszm-SmallAreaHighGrayLevelEmphasis, original-glszm-SmallAreaLowGrayLevelEmphasis, original-glszm-ZoneEntropy, original-glszm-ZonePercentage, original-ngtdm-Busyness, original-ngtdm-Complexity, original-ngtdm-Contrast, original-ngtdm-Strength. According to pathological grouping testing, it was found that there 54 significant differences between squamous cell carcinoma and adenocarcinoma in terms of original-shape-Elongation, original-shape-Flatness, original-shape-LeastAxisLength, original-shape-MajorAxisLength, original-shape-Maximum2DDiameterColumn, original-shape-Maximum2DDiameterRow, original-shape-Maximum2DDiameterSlice, original-shape-Maximum3DDiameter, original-shape-MeshVolume, original-shape-MinorAxisLength, original-shape-SurfaceArea, original-shape-SurfaceVolumeRatio, original-shape-VoxelVolume, original-firstorder-Entropy, original-firstorder-InterquartileRange, original-firstorder-Kurtosis, original-firstorder-Range, original-firstorder-RobustMeanAbsoluteDeviation, original-firstorder-Skewness, original-firstorder-Uniformity, original-glcm-Idn, original-glcm-Imc1, original-glcm-Imc2, original-glcm-SumEntropy, original-gldm-DependenceEntropy, original-gldm-DependenceNonUniformity, original-gldm-GrayLevelNonUniformity, original-gldm-HighGrayLevelEmphasis, original-gldm-LowGrayLevelEmphasis, original-gldm-SmallDependenceLowGrayLevelEmphasis, original-glrlm-GrayLevelNonUniformity, original-glrlm-GrayLevelNonUniformityNormalized, original-glrlm-HighGrayLevelRunEmphasis, original-glrlm-LongRunHighGrayLevelEmphasis, original-glrlm-LowGrayLevelRunEmphasis, original-glrlm-RunLengthNonUniformity, original-glrlm-ShortRunHighGrayLevelEmphasis, original-glrlm-ShortRunLowGrayLevelEmphasis, original-glszm-GrayLevelNonUniformity, original-glszm-HighGrayLevelZoneEmphasis, original-glszm-LargeAreaEmphasis, original-glszm-LargeAreaHighGrayLevelEmphasis, original-glszm-LargeAreaLowGrayLevelEmphasis, original-glszm-LowGrayLevelZoneEmphasis, original-glszm-SizeZoneNonUniformity, original-glszm-SizeZoneNonUniformityNormalized, original-glszm-SmallAreaEmphasis, original-glszm-SmallAreaHighGrayLevelEmphasis, original-glszm-SmallAreaLowGrayLevelEmphasis, original-glszm-ZoneVariance, original-ngtdm-Busyness, original-ngtdm-Coarseness, original-ngtdm-Complexity, original-ngtdm-Strength.
Conclusion
The enhancement of lung CT has a relatively small impact on extracting image features, while selecting lung or mediastinal windows for image segmentation has a significant impact on extracting image features. Therefore, choosing lung or mediastinal windows for feature extraction should be carefully considered, as the size of the image segmentation range has a significant impact on image features. The impact of lung squamous cell carcinoma and adenocarcinoma on imaging features is also significant, indicating a high possibility of distinguishing between squamous cell carcinoma and adenocarcinoma based on radiomics.