Background: This study was performed to compare the mammographic, sonographic, and magnetic resonance imaging (MRI) characteristics of phyllodes tumors and fibroadenomas, which may resemble each other. Methods: Preoperative mammograms, B-mode and Doppler sonograms, and dynamic breast MRIs of 72 patients with pathologically proven fibroadenomas and 70 patients with pathologically proven phyllodes tumor were evaluated in this retrospective study. Statistical significance was evaluated using chi-square and Fisher's exact tests. Correlations in lesion size among radiological methods were examined by Pearson's correlation analysis. Results: The features that differed on mammogram were size, shape, and margin of the mass. Sonograms showed significant differences in size, shape, margin, echo pattern, and vascularization of the mass. Pearson's correlation analysis showed strong agreement among radiological methods in terms of assessment of size. Tumor size ≥ 3 cm, irregular shape, microlobulated margins, complex internal echo pattern, and hypervascularity were significant findings of phyllodes tumors. Internal cystic areas on MRI were frequently associated with phyllodes tumors. Conclusion: Mammographic, sonographic, and MRI findings of fibroadenomas and phyllodes tumors could help radiologists to ascertain imaging-histological concordance and guide clinicians in their decision making regarding adequate follow-up or the necessity of biopsy.

To compare the accuracy and repeatability of emerging machine learning based (i.e. deep) automatic segmentation algorithms with those of well-established semi-automatic (interactive) methods for determining liver volume in living liver transplant donors at computerized tomography (CT) imaging. A total of 12 (6 semi-, 6 full-automatic) methods are evaluated. The semi-automatic segmentation algorithms are based on both traditional iterative models including watershed, fast marching, region growing, active contours and modern techniques including robust statistical segmenter and super-pixels. These methods entail some sort of interaction mechanism such as placing initialization seeds on images or determining a parameter range. The automatic methods are based on deep learning and they include three framework templates (DeepMedic, NiftyNet and U-Net) the first two of which are applied with default parameter sets and the last two involve adapted novel model designs. For 20 living donors (6 training and 12 test datasets), a group of imaging scientists and radiologists created ground truths by performing manual segmentations on contrast material-enhanced CT images. Each segmentation is evaluated using five metrics (i.e. volume overlap and relative volume errors, average/RMS/maximum symmetrical surface distances). The results are mapped to a scoring system and a final grade is calculated by taking their average. Accuracy and repeatability were evaluated using slice by slice comparisons and volumetric analysis. Diversity and complementarity are observed through heatmaps. Majority voting and Simultaneous Truth and Performance Level Estimation (STAPLE) algorithms are utilized to obtain the fusion of the individual results. The top four methods are determined to be automatic deep models having 79.63, 79.46 and 77.15 and 74.50 scores. Intra-user score is determined as 95.14. Overall, deep automatic segmentation outperformed interactive techniques on all metrics. The mean volume of liver of ground truth is found to be 1409.93 mL ± 271.28 mL, while it is calculated as 1342.21 mL ± 231.24 mL using automatic and 1201.26 mL ± 258.13 mL using interactive methods, showing higher accuracy and less variation on behalf of automatic methods. The qualitative analysis of segmentation results showed significant diversity and complementarity enabling the idea of using ensembles to obtain superior results. The fusion of automatic methods reached 83.87 with majority voting and 86.20 using STAPLE that are only slightly less than fusion of all methods that achieved 86.70 (majority voting) and 88.74 (STAPLE). Use of the new deep learning based automatic segmentation algorithms substantially increases the accuracy and repeatability for segmentation and volumetric measurements of liver. Fusion of automatic methods based on ensemble approaches exhibits best results almost without any additional time cost due to potential parallel execution of multiple models.

Dir Journal Diagnostic and Interventional Radiology

SUMMARY Comparison of radiological scoring systems, clinical scores, neutrophil-lymphocyte ratio and serum C-reactive protein level for severity and mortality in acute pancreatitis BACKGROUND/AIMS To compare radiological scoring systems, clinical scores, serum C-reactive protein (CRP) levels and the neutrophil-lymphocyte ratio (NLR) for predicting the severity and mortality of acute pancreatitis (AP). MATERIALS AND METHODS Demographic, clinical, and radiographic data from 80 patients with AP were retrospectively evaluated. The harmless acute pancreatitis score (HAPS), systemic inflammatory response syndrome (SIRS), bedside index for severity in acute pancreatitis (BISAP), Ranson score, Balthazar score, modified computed tomography severity index (CTSI), extrapancreatic inflammation on computed tomography (EPIC) score and renal rim grade were recorded. The prognostic performance of radiological and clinical scoring systems, NLR at admission, and serum CRP levels at 48 hours were compared for severity and mortality according to the revised Atlanta Criteria. The data were evaluated by calculating the receiver operator characteristic (ROC) curves and area under the ROC (AUROC). RESULTS Out of 80 patients, 19 (23.8%) had severe AP, and 9 (11.3%) died. The AUROC for the BISAP score was 0.836 (95%CI: 0.735-0.937), with the highest value for severity. With a cut-off of BISAP ≥2, sensitivity and specificity were 68.4% and 78.7%, respectively. The AUROC for NLR was 0.915 (95%CI: 0.790-1), with the highest value for mortality. With a cut-off of NLR >11.91, sensitivity and specificity were 76.5% and 94.1%, respectively. Of all the radiological scoring systems, the EPIC score had the highest AUROC, i.e., 0.773 (95%CI: 0.645-0.900) for severity and 0.851 (95%CI: 0.718-0.983) for mortality, with a cut-off value ≥6. CONCLUSION The BISAP score and NLR might be preferred as early determinants of severity and mortality in AP. The EPIC score might be suggested from the current radiological scoring systems.