MRI-based assessment of extraprostatic tumor extension showed good diagnostic performance, substantial agreement, and association with histopathologic tumor extension. Background There are no standardized and well-validated criteria for assessing the risk of extraprostatic extension (EPE) of prostate cancer at preoperative multiparametric MRI. Purpose To compare ...
The reliability of radiomics features (RFs) is crucial for quantifying tumour heterogeneity. We assessed the influence of imaging, segmentation, and processing conditions (quantization range, bin number, signal-to-noise ratio [SNR], and unintended outliers) on RF measurement. Low SNR and unintended outliers increased the standard deviation and mean values of histograms to calculate the first-order RFs. Variations in imaging processing conditions significantly altered the shape of the probability distribution (centre of distribution, extent of dispersion, and segmentation of probability clusters) in second-order RF matrices (i.e. grey-level co-occurrence and grey-level run length), thereby eventually causing fluctuations in RF estimation. Inconsistent imaging and processing conditions decreased the number of reliably measured RFs in terms of individual RF values (intraclass correlation coefficient ≥0.75) and inter-lesion RF ratios (coefficient of variation <15%). No RF could be reliably estimated under inconsistent SNR and inclusion of outlier conditions. By contrast, with high SNR and no outliers, all first-order RFs, 11 (42%) grey-level co-occurrence RFs and five (42%) grey-level run length RFs showed acceptable reliability. Our study suggests that optimization of SNR, exclusion of outliers, and application of relevant quantization range and bin number should be performed to ensure the robustness of radiomics studies assessing tumor heterogeneity.
8 February 2016 in Journal of Magnetic Resonance Imaging
OBJECTIVE. The objective of our study was to compare the diagnostic accuracy of virtual cystoscopy, multiplanar reformation, and source CT images for lesion detection in the contrast material-filled bladder. SUBJECTS AND METHODS. Two observers independently evaluated 47 patients (28 men and 19 women; mean age ± SD, 59 ± 16 years) with virtual cystoscopy, multiplanar reconstruction, and source CT images acquired with contrast material-filled bladder using an MDCT scanner (detector array, 4 × 1.25 mm; beam pitch, 0.75). Agreement between the two observers was evaluated for the three reconstruction methods using kappa statistics. Using the conventional cystoscopic findings as a reference, we compared the results of the three reconstruction techniques both by bladder site and by patient using the McNemar test. RESULTS. The interobserver agreement for the number of positive sites was excellent for virtual cystoscopy (κ = 0.816), fair for multiplanar reconstruction (κ = 0.461), and good for source CT images κ = 0.676). For both observers, the sensitivity for lesion detection by bladder site was significantly greater with virtual cystoscopy (observer 1, 95%; observer 2, 90%) than with multiplanar reconstruction (78% and 60%) and source CT (68% and 65%) images (p < 0.05), whereas the specificity by bladder site and the sensitivity and specificity by patient did not differ with the three methods (p > 0.05). For determining the presence or absence of lesion at each site, virtual cystoscopy was more accurate than multiplanar reconstruction and source CT images for both observers (p < 0.05). CONCLUSION. Virtual cystoscopy is more accurate than multiplanar reconstruction and source CT images for the detection of lesions in the bladder.