Table 2.
Predictor values of included radiomic features in the machine learning model
| Radiomics feature (raw output) | Feature description [12] | Normalized model coefficient |
|---|---|---|
| Entire cohort (n = 236): UTUC vs. RCC | ||
| Firstorder.histogram.Entropy | First order feature: Entropy as parameter of randomness in image values | -0.146725622 |
| Firstorder.histogram.Mean | First order feature: Mean gray level intensity in ROI | -0.093077333 |
| Firstorder.histogram.Median.abs.deviation | First order feature: Mean absolute deviation of intensity values from the median value of the image array | -0.108286287 |
| Firstorder.intensity.Robust.mean.abs.deviation | First order feature: Mean absolute deviation of intensity values from the mean value of 10–90% of the image array | -0.384201940 |
| Glcm.Difference.average | Gray level co-occurrence matrix feature: Relationship between pairs of similar and pairs of dissimilar gray level values and | -0.136673047 |
| Glcm.Difference.variance | Gray level co-occurrence matrix feature: Weights means of dissimilar intensity level pairs that differ more from the mean higher | -0.313556241 |
| Glcm.Dissimilarity | Gray level co-occurrence matrix feature: Relationship between pairs of similar and pairs of dissimilar gray level values and | -0.003591719 |
| Glcm.Joint.entropy | Gray level co-occurrence matrix feature: Describes the randomness between neighboring intensity values | 0.114638300 |
| Glcm.Std | Gray level co-occurrence matrix feature: Standard deviation of GLCM features | -0.053630813 |
| Tumor.short.axis | Second-largest axis length of the ellipsoid enclosed by the ROI upon segmentation in the CT series | -1.888042670 |
| Subcohort (n = 172): high grade UTUC vs. ccRCC | ||
| Firstorder.histogram.Median.abs.deviation | First order feature: Mean absolute deviation of intensity values from the median value of the image array | -0.2014530418 |
| Firstorder.histogram.Std | First order feature: Standard deviation of the distribution of intensity values of the image array (Histogram) | -0.0125926496 |
| Firstorder.Robust.mean.abs.deviation | First order feature: Mean absolute deviation of intensity values from the mean value of 10–90% of the image array | -0.1378609126 |
| Firstorder.intensity.Std | First order feature: Standard deviation of intensity values | -0.0000503306 |
| Glcm.Difference.variance | Gray level co-occurrence matrix feature: Weights means of dissimilar intensity level pairs that differ more from the mean higher | -0.1467854801 |
| Glcm.Std | Gray level co-occurrence matrix feature: Standard deviation of GLCM features | -0.3490017115 |
| Tumor.short.axis | Second-largest axis length of the ellipsoid enclosed by the ROI upon segmentation in the CT series | -1.2137183749 |
| Subcohort (n = 117): high-grade vs. low-grade UTUC | ||
| Glcm.Difference.variance | Gray level co-occurrence matrix feature: Weights means of dissimilar intensity level pairs that differ more from the mean higher | -0.2453007 |
| Glcm.Joint.average | Gray level co-occurrence matrix feature: Mean gray level intensity of the analyzed matrix | -0.1000878 |
| Glcm.Std | Gray level co-occurrence matrix feature: Standard deviation of GLCM features | 0.0481093 |
| Glcm.Sum.of.averages | Gray level co-occurrence matrix feature: Relationship between pairs of lower intensity and higher intensity values | 0.4578403 |
| Subcohort (n = 144): UTUC vs. nccRCC | ||
| Firstorder.histogram.Entropy | First order feature: Entropy as parameter of randomness in image values | -0.005620842 |
| Firstorder.intensity.Max | First order feature: Maximum gray level intensity in ROI | -0.209872612 |
| Firstorder.intensity.Root.mean.square | First order feature: Represents the square root of the mean of all squared intensity values and serves as a measure of the overall magnitude of image intensities | 0.216805958 |
| Glcm.Joint.entropy | Gray level co-occurrence matrix feature: Describes the randomness between neighboring intensity values | 0.063284407 |
| Tumor.short.axis | Second-largest axis length of the ellipsoid enclosed by the ROI upon segmentation in the CT series | -1.676368373 |