OBJECTIVE: To assess the predictive value of intratumoral and peritumoral radiomics based on Dual-energy CT urography (DECTU) multi-images for preoperatively predicting the muscle invasion status of bladder cancer (BCa).
METHODS: This retrospective analysis involved 202 BCa patients who underwent DECTU. DECTU-derived quantitative parameters were identified as risk factors through stepwise regression analysis to construct a DECT model. The radiomic features from the intratumoral and 3 mm outward peritumoral regions were extracted from the 120 kVp-like, 40 keV, 100 keV, and iodine-based material-decomposition (IMD) images in the venous-phase and were screened using Mann-Whitney U test, Spearman correlation analysis, and LASSO. Radiomics models were developed using the Multilayer Perceptron for the intratumoral, peritumoral and intra- and peritumoral (IntraPeri) regions. Subsequently, a nomogram was created by integrating the multi-image IntraPeri radiomics and DECT model. Model performance was evaluated using area-under-the-curve (AUC), accuracy, sensitivity, and specificity.
RESULTS: Normalized iodine concentration (NIC) was identified as an independent predictor for the DECT model. The IntraPeri model demonstrated superior performance compared to the intratumoral and peritumoral models both in 40 keV (0.830 vs. 0.766 vs. 0.763) and IMD images (0.881 vs. 0.840 vs. 0.821) in the test cohort. In the test cohort, the nomogram exhibited the best predictability (AUC=0.886, accuracy=0.836, sensitivity=0.737, and specificity=0.881), outperformed the DECT model (AUC=0.763, accuracy=0.754, sensitivity=0.632, and specificity=0.810) in predicting muscle invasion status of BCa with a statistically significant difference (p < 0.05).
CONCLUSIONS: The nomogram, incorporating IntraPeri radiomics and NIC, serves as a valuable and non-invasive tool for preoperatively assessing the muscle invasion status of BCa.

OBJECTIVE: To develop a prediction model based on spectral computed tomography (CT) to evaluate axillary lymph node (ALN) with macrometastases in clinical T1/2N0 invasive breast cancer.
METHODS: A total of 217 clinical T1/2N0 invasive breast cancer patients who underwent spectral CT scans were retrospectively enrolled and categorized into a training cohort (n = 151) and validation cohort (n = 66). These patients were classified into ALN nonmacrometastases (stage pN0 or pN0 [i+] or pN1mi) and ALN macrometastases (stage pN1-3) subgroups. The morphologic criteria and quantitative spectral CT parameters of the most suspicious ALN were measured and compared. Least absolute shrinkage and selection operator (Lasso) was used to screen predictive indicators to build a logistic model. The receiver operating characteristic (ROC) curve and decision curve analysis (DCA) were used to evaluate the models.
RESULTS: The combined arterial-venous phase spectral CT model yielded the best diagnostic performance in discrimination of ALN nonmacrometastases and ALN macrometastases with the highest AUC (0.963 in the training cohort and 0.945 in validation cohorts). Among single phase spectral CT models, the venous phase spectral CT model showed the best performance (AUC = 0.960 in the training cohort and 0.940 in validation cohorts). There was no significant difference in AUCs among the 3 models (DeLong test, P > .05 for each comparison).
CONCLUSIONS: A Lasso-logistic model that combined morphologic features and quantitative spectral CT parameters based on contrast-enhanced spectral imaging potentially be used as a noninvasive tool for individual preoperative prediction of ALN status in clinical T1/2N0 invasive breast cancers.

UNASSIGNED: The commonly used methods for removing metal-induced beam hardening artifacts often rely on the use of high energy photons with either high tube voltage or high energy virtual monoenergetic images in dual-energy computed tomography (CT), the radiation dose was usually relatively high in order to generate adequate signals. This retrospective study is designed to evaluate the application of a metal artifact reduction (MAR) algorithm in reducing pedicle screw metal-caused beam hardening artifacts in post-surgery pediatric low radiation dose spine CT images.
UNASSIGNED: Seventy-seven children (3-15 years) who had undergone a low dose CT with 140 or 100 kV were enrolled. The radiation dose was 1.40 mGy for the 3-8 years old and 2.61 mGy for 9-15 years old children. There were 116 pedicle screws evaluated. The raw data were reconstructed with adaptive statistical iterative reconstruction-V (ASIR-V) at 50% strength, ASIR-V with MAR (AV-MAR), deep learning image reconstruction (DLIR) at high strength and DLIR with MAR (DL-MAR). The image quality concerning pedicle screws was evaluated objectively in terms of the length of beam-hardening artifact (LHA) and artifact index (AI), and subjectively using a 4-point scale (4 points: best, 3 points: acceptable).
UNASSIGNED: Both AV-MAR and DL-MAR significantly reduced metal-induced beam hardening artifacts with smaller LHA (15.76±10.12 mm, a reduction of 57.24% and 15.66±10.49 mm, a reduction of 57.40%, respectively), and AI value (62.50±33.51, a reduction of 64.65% and 61.03±32.61, a reduction of 65.01%, respectively) compared to ASIR-V and DLIR (all P<0.01), The subjective image quality scores concerning the screws were 3.37±0.49 and 3.47±0.50 with AV-MAR and DL-MAR, respectively, higher than the respective value of 1.73±0.44 and 1.76±0.43 without MAR (all P<0.01).
UNASSIGNED: MAR significantly reduces the low-density artifacts caused by metal screws in post-surgery pediatric low-dose spine CT images, across different tube voltages, radiation dose levels and reconstruction algorithms. Combining DL-MAR further improves the overall image quality under low radiation dose conditions.

BACKGROUND: Percutaneous Endoscopic Lumbar Discectomy (PELD) has emerged as routine treatment for lumbar disc herniation (LDH) due to its minimal invasiveness and quick recovery. However, PELD demands high precision from the surgeon, as the risk of intraoperative complications is substantial, including potential damage to the nerve root and dura, and a higher likelihood of recurrence post-surgery. Thus, preoperative planning utilizing CT and MRI imaging is essential.
METHODS: In this study, the clinical data of 140 patients treated with PELD for LDH from January 2021 to December 2023 were retrospectively analyzed. Patients were categorized into two groups based on whether CT and MRI registration (CMR) was employed for surgical planning: a CMR group (n=68) and a control group (n=72). Data collected included surgery time, hospital stay duration, and scores from the Visual Analog Scale (VAS) for low back and leg pain, as well as the Japanese Orthopaedic Association Lumbar Spine Score (JOA). Differences between the two groups were assessed using the Student\'s t-test.
RESULTS: No significant difference was found in hospital stay length between the groups (P=0.277). Surgery time was significantly shorter in the CMR group (P<0.001). Prior to surgery, no significant differences in VAS scores for leg and low back pain were observed between the groups (P=0.341 and P=0.131, respectively); however, at 2 months postoperatively, both scores were significantly lower in the CMR group (P<0.001 and P=0.002, respectively). Similarly, no difference in preoperative JOA scores was noted (P=0.750), but at 2 months postoperative, the CMR group exhibited significantly higher scores (P<0.001).
CONCLUSIONS: Compared with the traditional PELD, the preoperative use of CMR has shown to reduce surgery time, alleviate leg and low back pain, and increase the lumbar JOA score at 2 months after surgery, underscoring its efficacy in enhancing surgical outcomes.

OBJECTIVE: The accurate detection and precise segmentation of lung nodules on computed tomography are key prerequisites for early diagnosis and appropriate treatment of lung cancer. This study was designed to compare detection and segmentation methods for pulmonary nodules using deep-learning techniques to fill methodological gaps and biases in the existing literature.
METHODS: This study utilized a systematic review with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, searching PubMed, Embase, Web of Science Core Collection, and the Cochrane Library databases up to May 10, 2023. The Quality Assessment of Diagnostic Accuracy Studies 2 criteria was used to assess the risk of bias and was adjusted with the Checklist for Artificial Intelligence in Medical Imaging. The study analyzed and extracted model performance, data sources, and task-focus information.
RESULTS: After screening, we included nine studies meeting our inclusion criteria. These studies were published between 2019 and 2023 and predominantly used public datasets, with the Lung Image Database Consortium Image Collection and Image Database Resource Initiative and Lung Nodule Analysis 2016 being the most common. The studies focused on detection, segmentation, and other tasks, primarily utilizing Convolutional Neural Networks for model development. Performance evaluation covered multiple metrics, including sensitivity and the Dice coefficient.
CONCLUSIONS: This study highlights the potential power of deep learning in lung nodule detection and segmentation. It underscores the importance of standardized data processing, code and data sharing, the value of external test datasets, and the need to balance model complexity and efficiency in future research.
CONCLUSIONS: Deep learning demonstrates significant promise in autonomously detecting and segmenting pulmonary nodules. Future research should address methodological shortcomings and variability to enhance its clinical utility.
CONCLUSIONS: Deep learning shows potential in the detection and segmentation of pulmonary nodules. There are methodological gaps and biases present in the existing literature. Factors such as external validation and transparency affect the clinical application.

OBJECTIVE: This study aimed to develop and validate a predictive model for osteoporotic vertebral fractures (OVFs) risk by integrating demographic, bone mineral density (BMD), CT imaging, and deep learning radiomics features from CT images.
METHODS: A total of 169 osteoporosis-diagnosed patients from three hospitals were randomly split into OVFs (n = 77) and Non-OVFs (n = 92) groups for training (n = 135) and test (n = 34). Demographic data, BMD, and CT imaging details were collected. Deep transfer learning (DTL) using ResNet-50 and radiomics features were fused, with the best model chosen via logistic regression. Cox proportional hazards models identified clinical factors. Three models were constructed: clinical, radiomics-DTL, and fusion (clinical-radiomics-DTL). Performance was assessed using AUC, C-index, Kaplan-Meier, and calibration curves. The best model was depicted as a nomogram, and clinical utility was evaluated using decision curve analysis (DCA).
RESULTS: BMD, CT values of paravertebral muscles (PVM), and paravertebral muscles\' cross-sectional area (CSA) significantly differed between OVFs and Non-OVFs groups (P < 0.05). No significant differences were found between training and test cohort. Multivariate Cox models identified BMD, CT values of PVM, and CSAPS reduction as independent OVFs risk factors (P < 0.05). The fusion model exhibited the highest predictive performance (C-index: 0.839 in training, 0.795 in test). DCA confirmed the nomogram\'s utility in OVFs risk prediction.
CONCLUSIONS: This study presents a robust predictive model for OVFs risk, integrating BMD, CT data, and radiomics-DTL features, offering high sensitivity and specificity. The model\'s visualizations can inform OVFs prevention and treatment strategies.

UNASSIGNED: To construct and validate a computed tomography (CT) radiomics model for differentiating lung neuroendocrine neoplasm (LNEN) from lung adenocarcinoma (LADC) manifesting as a peripheral solid nodule (PSN) to aid in early clinical decision-making.
UNASSIGNED: A total of 445 patients with pathologically confirmed LNEN and LADC from June 2016 to July 2023 were retrospectively included from five medical centers. Those patients were split into the training set (n = 316; 158 LNEN) and external test set (n = 129; 43 LNEN), the former including the cross-validation (CV) training set and CV test set using ten-fold CV. The support vector machine (SVM) classifier was used to develop the semantic, radiomics and merged models. The diagnostic performances were evaluated by the area under the receiver operating characteristic curve (AUC) and compared by Delong test. Preoperative neuron-specific enolase (NSE) levels were collected as a clinical predictor.
UNASSIGNED: In the training set, the AUCs of the radiomics model (0.878 [95% CI: 0.836, 0.915]) and merged model (0.884 [95% CI: 0.844, 0.919]) significantly outperformed the semantic model (0.718 [95% CI: 0.663, 0.769], p both<.001). In the external test set, the AUCs of the radiomics model (0.787 [95% CI: 0.696, 0.871]), merged model (0.807 [95%CI: 0.720, 0.889]) and semantic model (0.729 [95% CI: 0.631, 0.811]) did not exhibit statistical differences. The radiomics model outperformed NSE in sensitivity in the training set (85.3% vs 20.0%; p <.001) and external test set (88.9% vs 40.7%; p = .002).
UNASSIGNED: The CT radiomics model could non-invasively, effectively and sensitively predict LNEN and LADC presenting as a PSN to assist in treatment strategy selection.

OBJECTIVE: Collateral status is a pivotal determinant of clinical outcomes in acute ischemic stroke (AIS); however, its evaluation can be challenging. We investigated the predictive value of CT perfusion (CTP) derived time and density alterations versus CTP for collateral status prediction in AIS.
METHODS: Consecutive patients with anterior circulation occlusion within 24 h were retrospectively included. Time-density curves of the CTP specified ischemic core, penumbra, and the corresponding contralateral unaffected brain were obtained. The collateral status was dichotomised into robust (4-5 scores) and poor (0-3 scores) using multiphase collateral scoring, as described by Menon et al.. Receiver operating characteristic curves and multivariable regression analysis were performed to assess the predictive ability of CTP-designated tissue time and density alterations, CTP for robust collaterals, and favourable outcomes (mRS score of 0-2 at 90 days).
RESULTS: One-hundred patients (median age, 68 years; interquartile range, 57-80 years; 61 men) were included. A smaller ischemic core, shorter peak time delay, lower peak density decrease, lower cerebral blood volume ratio, and cerebral blood flow ratio in the CTP specified ischemic core were significantly associated with robust collaterals (PFDR ≤ 0.004). The peak time delay demonstrated the highest diagnostic value (AUC, 0.74; P < 0.001) with 66.7 % sensitivity and 73.7 % specificity. Furthermore, the peak time delay of less than 8.5 s was an independent predictor of robust collaterals and favourable clinical outcomes.
CONCLUSIONS: Robust collateral status was significantly associated with the peak time delay in the ischemic core. It is a promising image marker for predicting collateral status and functional outcomes in AIS.

UNASSIGNED: To analyze the CT and MR features of Primary hepatic neuroendocrine neoplasms (PHNENs) in order to enhance the diagnostic accuracy of this disease.
UNASSIGNED: A retrospective analysis was conducted on patients diagnosed with hepatic neuroendocrine neoplasms, excluding other sites of origin through general examination and postoperative follow-up. The CT and MR signs were analyzed according to the 2018 version of Liver Imaging Reporting and Data System (LI-RADS), along with causes of misdiagnosis.
UNASSIGNED: Twelve patients, including 6 males and 6 females, were enrolled in this study. There was no significant increase in liver tumor markers among all cases. Most masses were multiple (9/12), exhibiting low attenuation on pre-contrast CT scans, T1-hypointense signal, T2-hyperintense signal, and restricted diffusion. The majority of these masses (7/10) demonstrated similar rim arterial phase hyper-enhancement as well as peripheral \"washout\" during venous portal phase and delayed phase imaging. Three cases had incomplete capsules while one case had a complete capsule. Cyst/necrosis was observed in 7 out of all cases following administration of contrast agent, with 5 mainly distributed in the periphery. All masses lacked fat, calcification, vascular or bile duct tumor thrombus formation.
UNASSIGNED: The imaging findings associated with PHNENs possess certain specificity, often presenting as multiple masses within the liver accompanied by peripheral cyst/necrosis, similar rim arterial phase hyper-enhancement during venous portal phase and delayed phase imaging.

OBJECTIVE: Diagnosing subcentimeter solid pulmonary nodules (SSPNs) remains challenging in clinical practice. Deep learning may perform better than conventional methods in differentiating benign and malignant pulmonary nodules. This study aimed to develop and validate a model for differentiating malignant and benign SSPNs using CT images.
METHODS: This retrospective study included consecutive patients with SSPNs detected between January 2015 and October 2021 as an internal dataset. Malignancy was confirmed pathologically; benignity was confirmed pathologically or via follow-up evaluations. The SSPNs were segmented manually. A self-supervision pre-training-based fine-grained network was developed for predicting SSPN malignancy. The pre-trained model was established using data from the National Lung Screening Trial, Lung Nodule Analysis 2016, and a database of 5478 pulmonary nodules from the previous study, with subsequent fine-tuning using the internal dataset. The model\'s efficacy was investigated using an external cohort from another center, and its accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were determined.
RESULTS: Overall, 1276 patients (mean age, 56 ± 10 years; 497 males) with 1389 SSPNs (mean diameter, 7.5 ± 2.0 mm; 625 benign) were enrolled. The internal dataset was specifically enriched for malignancy. The model\'s performance in the internal testing set (316 SSPNs) was: AUC, 0.964 (95% confidence interval (95%CI): 0.942-0.986); accuracy, 0.934; sensitivity, 0.965; and specificity, 0.908. The model\'s performance in the external test set (202 SSPNs) was: AUC, 0.945 (95% CI: 0.910-0.979); accuracy, 0.911; sensitivity, 0.977; and specificity, 0.860.
CONCLUSIONS: This deep learning model was robust and exhibited good performance in predicting the malignancy of SSPNs, which could help optimize patient management.