Table 4.

The pre-processing work flow and settings for each investigated parameter for the glioma data.

Investigated parameter	Step 1: Resolution of DWI	Step 2: Noise in DWI	Step 3: Calculate ADC	Step 4: Select GLCM size	Step 5: Quantize image
Resolution	1.2 mm2 (1.0×)	σ = 17	200–1000 s/mm2	32	AutoROI
	1.8 mm2 (1.5×)
	3.6 mm2 (3.0×)
Noise	1.2 mm2	σ = 17 (1.0×)	200–1000 s/mm2	32	AutoROI
		σ = 34 (2.0×)
		σ = 68 (4.0×)
Diffusion b-values	1.2 mm2	σ = 17	200–1000 s/mm2	32	AutoROI
			0–1000 s/mm2
			200,1000 s/mm2
			200,1000 s/mm2
Gray levels	1.2 mm2	σ = 17	200–1000 s/mm2	8	AutoROI
				16
				32
				64
				128
Quantization method	1.2 mm2	σ = 17	200–1000 s/mm2	32	AutoROI
					AutoSlice
					Manual

Each row represents the work flow of one investigated parameter. The prostate cancer data set used a similar work flow, where the native resolution was 1.625 mm², the native noise standard deviation was σ = 2.5, and the ADC was calculated using 0, 800 s/mm² only.