RESUMEN
Objective@#To validate the performance of 3T spin-echo echo-planar imaging (SE-EPI) magnetic resonance elastography (MRE) for staging hepatic fibrosis in a large population, using surgical specimens as the reference standard. @*Materials and Methods@#This retrospective study initially included 310 adults (155 undergoing hepatic resection and 155 undergoing donor hepatectomy) with histopathologic results from surgical liver specimens. They underwent 3T SE-EPI MRE ≤ 3 months prior to surgery. Demographic findings, underlying liver disease, and hepatic fibrosis pathologic stage according to METAVIR were recorded. Liver stiffness (LS) was measured by two radiologists, and inter-reader reproducibility was evaluated using the intraclass correlation coefficient (ICC). The mean LS of each fibrosis stage (F0–F4) was calculated in total and for each etiologic subgroup. Comparisons among subgroups were performed using the Kruskal–Wallis test and Conover post-hoc test. The cutoff values for fibrosis staging were estimated using receiver operating characteristic (ROC) curve analysis. @*Results@#Inter-reader reproducibility was excellent (ICC, 0.98; 95% confidence interval, 0.97–0.99). The mean LS values were 1.91, 2.41, 3.24, and 5.41 kPa in F0–F1 (n = 171), F2 (n = 26), F3 (n = 38), and F4 (n = 72), respectively. The discriminating cutoff values for diagnosing ≥ F2, ≥ F3, and F4 were 2.18, 2.71, and 3.15 kPa, respectively, with the ROC curve areas of 0.97–0.98 (sensitivity 91.2%–95.9%, specificity 90.7%–99.0%). The mean LS was significantly higher in patients with cirrhosis (F4) of nonviral causes, such as primary biliary cirrhosis (9.56 kPa) and alcoholic liver disease (7.17 kPa) than in those with hepatitis B or C cirrhosis (4.28 and 4.92 kPa, respectively). There were no statistically significant differences in LS among the different etiologic subgroups in the F0–F3 stages. @*Conclusion@#The 3T SE-EPI MRE demonstrated high interobserver reproducibility, and our criteria for staging hepatic fibrosis showed high diagnostic performance. LS was significantly higher in patients with non-viral cirrhosis than in those with viral cirrhosis.
RESUMEN
Objective@#To evaluate the diagnostic performance of a deep learning algorithm for the automated detection of developmental dysplasia of the hip (DDH) on anteroposterior (AP) radiographs. @*Materials and Methods@#Of 2601 hip AP radiographs, 5076 cropped unilateral hip joint images were used to construct a dataset that was further divided into training (80%), validation (10%), or test sets (10%). Three radiologists were asked to label the hip images as normal or DDH. To investigate the diagnostic performance of the deep learning algorithm, we calculated the receiver operating characteristics (ROC), precision-recall curve (PRC) plots, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) and compared them with the performance of radiologists with different levels of experience. @*Results@#The area under the ROC plot generated by the deep learning algorithm and radiologists was 0.988 and 0.988–0.919, respectively. The area under the PRC plot generated by the deep learning algorithm and radiologists was 0.973 and 0.618– 0.958, respectively. The sensitivity, specificity, PPV, and NPV of the proposed deep learning algorithm were 98.0, 98.1, 84.5, and 99.8%, respectively. There was no significant difference in the diagnosis of DDH by the algorithm and the radiologist with experience in pediatric radiology (p = 0.180). However, the proposed model showed higher sensitivity, specificity, and PPV, compared to the radiologist without experience in pediatric radiology (p < 0.001). @*Conclusion@#The proposed deep learning algorithm provided an accurate diagnosis of DDH on hip radiographs, which was comparable to the diagnosis by an experienced radiologist.
RESUMEN
Objective@#To evaluate the diagnostic performance of a deep learning algorithm for the automated detection of developmental dysplasia of the hip (DDH) on anteroposterior (AP) radiographs. @*Materials and Methods@#Of 2601 hip AP radiographs, 5076 cropped unilateral hip joint images were used to construct a dataset that was further divided into training (80%), validation (10%), or test sets (10%). Three radiologists were asked to label the hip images as normal or DDH. To investigate the diagnostic performance of the deep learning algorithm, we calculated the receiver operating characteristics (ROC), precision-recall curve (PRC) plots, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) and compared them with the performance of radiologists with different levels of experience. @*Results@#The area under the ROC plot generated by the deep learning algorithm and radiologists was 0.988 and 0.988–0.919, respectively. The area under the PRC plot generated by the deep learning algorithm and radiologists was 0.973 and 0.618– 0.958, respectively. The sensitivity, specificity, PPV, and NPV of the proposed deep learning algorithm were 98.0, 98.1, 84.5, and 99.8%, respectively. There was no significant difference in the diagnosis of DDH by the algorithm and the radiologist with experience in pediatric radiology (p = 0.180). However, the proposed model showed higher sensitivity, specificity, and PPV, compared to the radiologist without experience in pediatric radiology (p < 0.001). @*Conclusion@#The proposed deep learning algorithm provided an accurate diagnosis of DDH on hip radiographs, which was comparable to the diagnosis by an experienced radiologist.