Skip to main content

Table 4 A list features to be used for lesion-based outcome prediction

From: Mining multi-site clinical data to develop machine learning MRI biomarkers: application to neonatal hypoxic ischemic encephalopathy

CategoriesDetails of features
I. Lesion anatomyI.1. Mass center in standard neonatal atlas space
I.2. Percentage of the whole-brain volume and the volume of each of the 61 auto-segmented brain structures being injured [76, 78, 122]
I.3. Ratios of volumetric injury in the same brain structures between the left and right hemisphere
I.4. Percentage and distribution of HIE lesions in 28 major fiber tracts as defined in the JHU atlas [123]
II. Lesion geometryII.1. Lesion volume
II.2. Maximum diameter along different orthogonal directions, maximum surface of lesion, lesion compactness, lesion spherecity, surface-to-volume ratio
III. Lesion heterogeneityIII.1. Histogram analysis (0, 25, 50, 75 and 100-percentile) of T1, T2, DWI, ADC, ZT1, ZT2, ZDWI, ZADC signal values within the lesion regions
III.2. Skewness (asymmetry), kurtosis (flatness), uniformity and randomness (entropy and standard deviations) of T1, T2, DWI, ADC, ZT1, ZT2, ZDWI, ZADC signal values within the lesion regions
IV. Lesion textureIV.1. gray-level co-occurrence matrix (GLCM) features and gray-level run-length matrix (GLRLM) of T1, T2, DWI, ADC, ZT1, ZT2, ZDWI, ZADC signal values within lesion regions
IV.2. fractal analysis, Minkowski functionals, wavelet transform and Laplacian transforms of Gaussian-filtered images for the lesion regions