Objective:

To investigate the effects of variety in region of interest (ROI) and slice on the measurement of the first-order statistical features of apparent diffusion coefficient (ADC) map of prostate peripheral zone.

Methods:

Retrospective analysis of 18 subjects with prostate cancer in peripheral zone was performed. The first-order statistical features of cancer and n

ormal tissues were measured with different ROIs referring to the pathological data. ROI 1: the cancer lesions and normal tissues wereidentified by T2WI image characteristic on the largest slice; ROI 2: identified by ADC map; the areas of ROI 3 and ROI 4 were 3/4 and 1/2 of ROI 1

respectively; ROI 5 was the whole volume of the cancer lesions and normal tissues with multi-slice. Data which were significantly different between cancer lesions and normal tissues were selected and further assessed their variety in different ROIs.

Results:

Data were obtained from 19 cancer lesions and 21 normal tissues. There was no significant differences in skewness and kurtosis between cancer lesions and normal tissues. The mean and percentile ADC values from all ROIs were all significantly higher in normal tissues than cancer lesions. As the area of ROI reduced

the mean ADC value in the cancer lesions decreased

but in the normal tissues remained stable. Their standard deviation (SD) values were not significantly different

but the SD values for most of the percentile ADC values were significantly higher in normal tissues compared with cancer lesions. The ADC values

the percentile ADC values and their SD values measured on the largest slice were not significantly different from those by whole volume.

Conclusion:

The area of ROI has remarkable effect on the mean and percentile ADC values

and first-order statistical features could provide more significative data.