Summary of the open source clinical studies up to May 2017 that perform texture analyses on Dynamic Contrast Enhanced (DCE)-MRI

Table_Review_DCEandTexture.docx and Table_Review_DCEandTexture.pdf are text files with the information

This is a Table with the Results of the systematic search in PubMed and Web of Science of open source clinical studies published up to May 2017, where texture analysis has been used in the analysis of Dynamic Contract Enhanced Magnetic Resonance Images (DCE-MRI). Search terms were combinations of words derived from “texture analysis” (i.e. texture analysis, textural, texture descriptors, texture characteristic) AND “DCE-MRI”. The time points where the texture analysis was used on each paper appears highlighted in the correspondent column. Texture analysis has been used in human studies for tissue classification (i.e. discriminating malignant vs. benign types of tumours) (Agner S.C. et al., 2011; Alic et al., 2011; Chaudhury et al., 2015; Eliat et al., 2012; Jianhua et al., 2009; Kale M.C. et al., 2013; Karahaliou et al., 2010; Nie et al., 2008; Rose et al., 2009; Soares et al., 2014; Wang et al., 2014; Woods et al., 2007), for predicting response to cancer treatment (Ahmed et al., 2013; Golden et al., 2013; Teruel et al., 2014; Torheim et al., 2014), and for characterising tumours (Alic et al., 2011; Chen et al., 2007; Viswanath et al., 2008). The studies summarised in this Table use texture analysis as an alternative or complement to the use of the more established pharmacokinetic models or area under the signal enhancement curve in response to the low specificity and high variability in the outcome of these methods mainly due to: 1) the heterogeneity of lesion enhancement and the effects of MR artefacts such as bias field inhomogeneity and intensity non-linearity on these conventional methods, 2) reported minor variations in the B1 magnetic field resulting in up to 50% error on the estimated contrast agent concentration vs. time curve, 3) these existent methods not taking into account the spatial distribution of the different contrast uptake parameters, 4) intra- and inter-observer variability while interpreting the outcome, and 5) these methods being a trade-off between spatial and temporal resolution providing only an imperfect gold standard that does not necessarily reflect the biological truth. Some studies did texture analysis based on encouraging results previously reported or unsuitability of the pharmacokinetic parameters obtained from the conventional models for their purpose.

Valdés Hernández, Maria; González Castro, Victor. (2017). Summary of the open source clinical studies up to May 2017 that perform texture analyses on Dynamic Contrast Enhanced (DCE)-MRI, 2007-2017 [text]. University of Edinburgh Centre for Clinical Brain Sciences. Department of Neuroimaging Sciences. http://dx.doi.org/10.7488/ds/2052.