Value of whole-tumor histogram-based texture analysis of baseline ADC map in prediction of tumor response to neoadjuvant chemoradiotherapy in locally advanced rectal cancer.

There has been increasing interest in quantitative methods to excavate more information than traditional descriptive features, facilitated by the availability of texture analysis software platforms. By doing whole-tumor histogram-based texture analysis on pre-treatment ADC map, histogram parameters reflecting the pixel distribution of ROIs were derived. Our study found that LARC that achieved pCR after NCRT appeared less heterogeneous on ADC map and had lower high percentile pre-treatment ADC values. Whole-tumor histogram parameters of pre-treatment ADC map were feasible to predict pCR in LARC, including the absolute value of relative deviation, frequency size, quantile 75%, 90%, and 95% of ADC value histogram.