Review Article

Current and Potential Applications of Artificial Intelligence in Gastrointestinal Stromal Tumor Imaging

Table 1

Feature metrics extracted in the radiomic analysis of images.

Texture metricMethod (s)Descriptors

First-order (statistical)Histogram analysisMean, median, kurtosis, skewness, quartiles, minimum, maximum, energy (uniformity), entropy, standard deviation
Second-order (statistical)GLCM, GLDM, NGTDM, GLRLM, GLSZMHomogeneity, contrast, autocorrelation, prominence, maximum probability, difference variance, dissimilarity, inverse difference moment, sum entropy, sum variance, sum average, inertia, coarseness, busyness, complexity, texture strength, short run emphasis, long run emphasis, gray-level nonuniformity, run-length nonuniformity, intensity variability, run-length variability, long-zone emphasis, short-zone emphasis, intensity nonuniformity, intensity, zone percentage, variability, size zone variability
Transform (statistical)Fourier, wavelets, discrete cosine, Gabor, law, LoG, LBPMetrics assessing magnitude, phase, direction, edge, noise, and other descriptors
Structural analysisFractal analysisHurst component, mean fractal dimension, standard deviation, lacunarity

Note. GLCM = gray-level cooccurrence matrix, GLDM = gray-level difference matrix, NGTDM = Neighborhood gray-tone difference matrix, GLRLM = gray-level run-length, GLSZM = gray-level size zone matrix, LoG = Laplacian of Gaussian, LBP = local binary pattern.