Table 1.

Previous studies about potentials of integrating DWI in RT. Purpose – 1: pre-treatment outcome prediction, 2: response assessment, inter-treatment prediction, 3: tumor delineation; Mexp - mono-exponential model, IVIM – intra-voxel incoherent motion; NGK – non-gaussian kurtosis, ADC – apparent diffusion coefficient, $\mathrm{\Delta }$ADC – difference of inter- or post-treatment ADC to baseline; DTI - diffusion tensor imaging, DCE – dynamic-contrast enhanced imaging.

Purpose	Site	Author/year citation	# patients	b-Values in s/mm2	Imaging time point	Model	Main findings
1	HNSCC	Lambrecht 2014 [44]	161	0, 50, 100, 500, 750, 1000	pre-RT	Mexp (high-, low- and full b-value range)	higher pre-treatment ADC in tumor, when derived from the high b-value range, is related to disease recurrence
1	“	Noij 2015 [30]	78	0, 750 and 0, 1000	pre-(C) RT	Mexp (ADC750, ADC1000)	higher pre-treatment ADC1000 in lymph nodes is related to lower disease-free survival
1	“	Hauser 2013 [53]	22	0, 50, 100, 150, 200, 250, 700, 800	pre-RT	IVIM	high perfusion fraction f in tumor may be related to poor prognosis
1,2	Rectal cancer	Jung 2012 [45]	35	0, 500, 1000	pre- and post-CRT (neoadjuvant)	Mexp	significant correlation between pre-treatment ADC and tumor volume reduction, as well as between $\mathrm{\Delta }$ADC and tumor volume reduction
1,2	“	Lambrecht 2012 [46]	20	0, 50, 100, 500, 750, 1000	pre-, inter-, and post-CRT (neoadjuvant)	Mexp	pre-treatment ADC as well as inter- and post-treatment $\mathrm{\Delta }$ADC may be useful for prediction and early assessment of treatment response; pretreatment ADC is significantly lower in patients with pathologic complete response
1	“	Joye 2017 [47]	85	0, 50, 100, 300, 600, 1000	pre-, inter-, and post-CRT	Mexp (high-, low- and full b-value range)	DWI is predictive for treatment response; the predictive power can be improved by combining DWI with FDG-PET and T2-weighted volumetry
1	Glioblastoma	Pramanik 2015 [48]	21	0, 1000, 3000	pre-CRT	no model applied	hypercellularity volume as defined on the b = 3000 acquisition is a significant prognostic factor for progression-free survival
1	Cervical cancer	Heo 2013 [49]	42	3 0, 500, 1000	pre-CRT	Mexp	higher mean ADC related to tumor recurrence; 75th percentile ADC predictor for tumor recurrence
1	“	Onal 2016 [50]	44	0, 800	pre-CRT, post-CRT	Mexp	lower ADC values pre-RT and post-RT associated to disease recurrence
1	“	Marconi 2016 [51]	66	0, 600 and 0, 800	pre-CRT	Mexp	Pre-treatment minimum ADC may be a prognostic factor for disease-free survival
1	“	Gladwish 2016 [52]	85	0, 50, 400, 1000, and 0, 100, 800 and 0, 50, 400, 800	pre-CRT	Mexp	95th percentile ADC might be a metric to predict treatment failure
2	HNSCC	Dirix 2009 [56]	15	0, 50, 100, 500, 750, 1000	Pre-, inter-, and post-CRT	Mexp	lesions showing loco-regional recurrence had a significantly lower inter-treatment ADC
2	“	King 2013 [57]	30	0, 100, 200, 300, 400, 500	Pre- and inter-CRT	Mexp	local failure is associated with lower relative increase of ADC compared to local control, as well as with a decrease of skewness and kurtosis in GTV-based ADC histograms
2	“	Marzi 2015 [65]	34	0, 25, 50, 75, 100, 150, 300, 500, 800	Pre-, inter-, and post-CRT	IVIM	pre-treatment f and D are independent predictors for shrinkage of major salivary glands
2	“	Vandecaveye 2012 [66]	29	0, 50, 100, 500, 750, 1000	Pre- and post-CRT	Mexp	$\mathrm{\Delta }$ADC three weeks after RT allows for early treatment response assessment
2	Cervical cancer	Haack 2015 [59]	11	0, 150, 600, 1000	Pre- and inter-RT	Mexp	volume with reduced diffusion as derived from DWI changes significantly during treatment, along with a significant mean ADC increase
2	“	Das 2015 [60]	24	0, 400, 800	Pre- and inter-CRT	Mexp	inter-treatment $\mathrm{\Delta }$ADC can be used for early response prediction
2	“	Zhu 2017 [21]	30	0, 10, 20, 30, 40, 50, 100, 150, 200, 350, 500, 650, 800, 1000	Pre- and inter-CRT	IVIM	D at 2 weeks as well as D and f 4 weeks after start of RT prognostic for therapy outcome
2	“	Daniel 2017 [61]	10	0, 850	Pre-, inter-, and post- CRT	Mexp	Patient averaged ADCs increased from baseline to follow up, low-ADC regions spatially varied over time
2	“	Schreuder 2015 [7]	231 (review)	mixed	Pre-, inter- and post-RT	Mexp	DWI can be used for early post-RT assessment, but not for early response monitoring
2	Glioma	Kassubek 2017 [63]	18	0, 800	Pre- and post-RT	DTI	DTI can potentially be used to asses irradiation-induced microstructural white matter damage
2	Glioblastoma	Nagesh 2008 [64]	25	0, 1000	Pre-, inter- and post-RT	DTI	DTI has potential for the assessment of radiation-induced white matter injury
2	“	Chu 2013 [67]	30	0, 1000 and 0, 3000	post-RT	Mexp (ADC1000, ADC3000)	Fifth percentiles of cumulative histograms of ADC1000 as well as of ADC3000 promising for the differentiation between true progression and pseudo-progression
2	Esophageal cancer	van Rossum 2015 [58]	20	0, 200, 800	Pre-, inter-, post-CRT (neoadjuvant)	Mexp	inter-treatment $\mathrm{\Delta }$ADC is a predictive factor for histopathologic response
3	“	Hou 2013 [68]	42	400, 600, 800	pre-treatment	no model applied	DWI is superior to CT or anatomical MR in GTV delineation
3	Pancreas cancer	Kartalis 2016 [26]	15	0, 50, 150, 200, 300, 600, 1000	pre-treatment	IVIM, Mexp, NGK	ADC and $D_{\mathrm{K}}$ might be valuable for differentiating between tumorous and non-tumorous parenchyma
3	Glioblastoma	Jensen 2017 [69]	11	0, 1000	pre-RT	DTI	DTI in combination with a model for the microscopic spread of tumor cells along white matter fiber tracts might be of value for defining the clinical target volume (CTV) of glioblastomas
3	Cervical cancer	Schernberg 2017 [70]	44	0, 1000	after CRT, before image guided adaptive brachytherapy	no model applied	DWI images (without applying quantitative models) might lead to modifications in high-risk clinical target volumes
3	Prostate cancer	Langer 2009 [71]	25	0, 600	pre-treatment	Mexp	ADC is superior to DCE and T2-mapping for differentiating between tumorous and non-tumorous tissue; classification accuracy can be increased by using a multi-parametric model
3	“	Groenendaal 2012 [72]	87	300, 500, 1000	pre-treatment	Mexp	Logistic regression-derived model including DWI, DCE can define different risk levels for tumor presence on a voxel level
3	“	Yu 2017 [73]	140	50, 600, 1000	pre-treatment	Mexp	Multiparametric model of DWI, T1, and T2 may discriminate between tumorous tissue and normal peripheral zone