“…We extract 18 features from each local patch including autocorrelation, contrast, entropy, variance, dissimilarity, homogeneity, cluster shade, energy, maximum probability, sum of squares of variance, sum of averages, sum of variance, sum of entropy, difference of entropy, difference of variance, normalized inverse difference moment, cluster prominence, and mutual information. Readers are encouraged to refer to [35] for further details on these well-established features in machine learning, and [12]–[15] for particular CAD systems in identification of lung abnormalities from CT scans in general.…”