Search

To investigate whether the combination scheme of deep learning score (DL-score) and radiomics can improve preoperative diagnosis in the presence of micropapillary/solid (MPP/SOL) patterns in lung adenocarcinoma (ADC). A retrospective cohort of 514 confirmed pathologically lung ADC in 512 patients after surgery was enrolled. The clinicoradiographic model (model 1) and radiomics model (model 2) were developed with logistic regression. The deep learning model (model 3) was constructed based on the deep learning score (DL-score). The combine model (model 4) was based on DL-score and R-score and clinicoradiographic variables. The performance of these models was evaluated with area under the receiver operating characteristic curve (AUC) and compared using DeLong's test internally and externally. The prediction nomogram was plotted, and clinical utility depicted with decision curve. The performance of model 1, model 2, model 3 and model 4 was supported by AUCs of 0.848, 0.896, 0.906, 0.921 in the Internal validation set, that of 0.700, 0.801, 0.730, 0.827 in external validation set, respectively. These models existed statistical significance in internal validation (model 4 vs model 3, P = 0.016; model 4 vs model 1, P = 0.009, respectively) and external validation (model 4 vs model 2, P = 0.036; model 4 vs model 3, P = 0.047; model 4 vs model 1, P = 0.016, respectively). The decision curve analysis (DCA) demonstrated that model 4 predicting the lung ADC with MPP/SOL structure would be more beneficial than the model 1and model 3 but comparable with the model 2. The combined model can improve preoperative diagnosis in the presence of MPP/SOL pattern in lung ADC in clinical practice. [ABSTRACT FROM AUTHOR]

To evaluate the value of texture analysis based on dynamic contrast enhanced MRI (DCE-MRI) in the differential diagnosis of thymic carcinoma and thymic lymphoma. Sixty-nine patients with pathologically confirmed (thymic carcinoma, n = 32; thymic lymphoma, n = 37) were enrolled in this retrospective study. Ktrans, Kep and Ve maps were automatically generated, and texture features were extracted, including mean, median, 5th/95th percentile, skewness, kurtosis, diff-variance, diff-entropy, contrast and entropy. The differences in parameters between the two groups were compared and the diagnostic efficacy was calculated. The Ktrans-related significant features yielded an area under the curve (AUC) of 0.769 (sensitivity 90.6%, specificity 51.4%) for the differentiation between thymic carcinoma and thymic lymphoma. The Kep-related significant features yielded an AUC of 0.780 (sensitivity 87.5%, specificity 62.2%). The Ve-related significant features yielded an AUC of 0.807 (sensitivity 75.0%, specificity 78.4%). The combination of DCE-MRI textural features yielded an AUC of 0.962 (sensitivity 93.8%, specificity 89.2%). Five parameters were screened out, including age, Ktrans-entropy, Kep-entropy, Ve-entropy, and Ve-P95. The combination of these five parameters yielded the best discrimination efficiency (AUC of 0.943, 93.7% sensitivity, 81.1% specificity). Texture analysis of DCE-MRI may be helpful to distinguish thymic carcinoma from thymic lymphoma. [ABSTRACT FROM AUTHOR]

Purpose: The purpose of this study is to compare the detection performance of the 3-dimensional convolutional neural network (3D CNN)-based computer-aided detection (CAD) models with radiologists of different levels of experience in detecting pulmonary nodules on thin-section computed tomography (CT). Patients and Methods: We retrospectively reviewed 1109 consecutive patients who underwent follow-up thin-section CT at our institution. The 3D CNN model for nodule detection was re-trained and complemented by expert augmentation. The annotations of a consensus panel consisting of two expert radiologists determined the ground truth. The detection performance of the re-trained CAD model and three other radiologists at different levels of experience were tested using a free-response receiver operating characteristic (FROC) analysis in the test group. Results: The detection performance of the re-trained CAD model was significantly better than that of the pre-trained network (sensitivity: 93.09% vs 38.44%). The re-trained CAD model had a significantly better detection performance than radiologists (average sensitivity: 93.09% vs 50.22%), without significantly increasing the number of false positives per scan (1.64 vs 0.68). In the training set, 922 nodules less than 3 mm in size in 211 patients at high risk were recommended for follow-up CT according to the Fleischner Society Guidelines. Fifteen of 101 solid nodules were confirmed to be lung cancer. Conclusion: The re-trained 3D CNN-based CAD model, complemented by expert augmentation, was an accurate and efficient tool in identifying incidental pulmonary nodules for subsequent management. [ABSTRACT FROM AUTHOR]

Background: To compare intravoxel incoherent motion (IVIM) and pharmacokinetic analysis dynamic contrast-enhanced MR imaging (DCE-MRI) in distinguishing lung cancer (LC) from benign solitary pulmonary lesions (SPL). Methods: This prospective study was approved by the institutional review board, and written informed consent was obtained. Eighty-one consecutive patients considered for SPL underwent DW-IVIM and DCE-3T MRI. ADC, D, D*, and f were calculated with mono- and bi-exponential models. K(trans) , kep , ve , and vp were calculated with the modified Tofts model. Receiver operating characteristic (ROC) analysis was constructed to determine the diagnostic performance of IVIM and DCE-MRI in discriminating LC from benignity. Results: There were 29 patients with a total of 48 benign SPL and 52 LCs: 4 small cell carcinomas (SCLC), 19 squamous cell carcinomas (SCC), and 29 adenocarcinomas (Adeno-Ca). Both Adeno-Ca (ADC: 1.19 ± 0.23 × 10(-3) mm(2) /s; D:1.12 ± 0.35 × 10(-3) mm(2) /s; ve :0.27 ± 0.13; K(trans) :0.24 ± 0.09 min(-1) ; kep :0.90 ± 0.45 min(-1) ) and SCC (1.13± 0.28 × 10(-3) mm(2) /s; 1.02 ± 0.32 10(-3) mm(2) /s; 0.32 ± 0.14; 0.26 ± 0.08 min(-1) ; 0.90 ± 0.48 min(-1) ) had significantly lower ADC, D, ve and larger K(trans) , kep than benignity (1.37 ± 0.38 × 10(-3) mm(2) /s; 1.34 ± 0.45 × 10(-3) mm(2) /s; 0.42 ± 0.19; 0.19 ± 0.08 min(-1) ; 0.53 ± 0.26 min(-1) ). D (72.2%) had significantly higher accuracy (72.2%) and higher sensitivity (91.3%) than other imaging indices (accuracy: 55.5-68.0%; sensitivity: 41.3-78.3%; all P < 0.01) except for accuracy in kep (70.8%; P > 0.05) in discriminating LC from benignity. K(trans) exhibited significantly higher specificity (84.6%) than the other indices (38.5-73.1%; P < 0.01). These results can be improved by combined D and K(trans) , leading to a sensitivity, specificity and accuracy of 94.2%, 92%, and 93.5%, respectively. Conclusion: IVIM-derived D and DCE-derived K(trans) are two promising parameters for differentiating LC from benignity. [ABSTRACT FROM AUTHOR]

To investigate the value of apparent diffusion coefficients (ADCs) histogram analysis for assessing World Health Organization (WHO) pathological classification and Masaoka clinical stages of thymic epithelial tumours. 37 patients with histologically confirmed thymic epithelial tumours were enrolled. ADC measurements were performed using hot-spot ROI (ADCHS-ROI) and histogram-based approach. ADC histogram parameters included mean ADC (ADCmean), median ADC (ADCmedian), 10 and 90 percentile of ADC (ADC10 and ADC90), kurtosis and skewness. One-way ANOVA, independent-sample t-test, and receiver operating characteristic were used for statistical analyses. There were significant differences in ADCmean, ADCmedian, ADC10, ADC90 and ADCHS-ROI among low-risk thymoma (type A, AB, B1; n = 14), high-risk thymoma (type B2, B3; n = 9) and thymic carcinoma (type C, n = 14) groups (all p-values <0.05), while no significant difference in skewness (p = 0.181) and kurtosis (p = 0.088). ADC10 showed best differentiating ability (cut-off value, ≤0.689 × 10−3 mm2 s–1; AUC, 0.957; sensitivity, 95.65%; specificity, 92.86%) for discriminating low-risk thymoma from high-risk thymoma and thymic carcinoma. Advanced Masaoka stages (Stage III and IV; n = 24) tumours showed significant lower ADC parameters and higher kurtosis than early Masaoka stage (Stage I and II; n = 13) tumours (all p-values <0.05), while no significant difference on skewness (p = 0.063). ADC10 showed best differentiating ability (cut-off value, ≤0.689 × 10−3 mm2 s–1; AUC, 0.913; sensitivity, 91.30%; specificity, 85.71%) for discriminating advanced and early Masaoka stage epithelial tumours. ADC histogram analysis may assist in assessing the WHO pathological classification and Masaoka clinical stages of thymic epithelial tumours. 1. ADC histogram analysis could help to assess WHO pathological classification of thymic epithelial tumours. 2. ADC histogram analysis could help to evaluate Masaoka clinical stages of thymic epithelial tumours. 3. ADC10 might be a promising imaging biomarker for assessing and characterizing thymic epithelial tumours. [ABSTRACT FROM AUTHOR]