Insights into Imaging最新文献

Objective: To develop an automatic segmentation model to delineate the adnexal masses and construct a machine learning model to differentiate between low malignant risk and intermediate-high malignant risk of adnexal masses based on ovarian-adnexal reporting and data system (O-RADS).

Methods: A total of 663 ultrasound images of adnexal mass were collected and divided into two sets according to experienced radiologists: a low malignant risk set (n = 446) and an intermediate-high malignant risk set (n = 217). Deep learning segmentation models were trained and selected to automatically segment adnexal masses. Radiomics features were extracted utilizing a feature analysis system in Pyradiomics. Feature selection was conducted using the Spearman correlation analysis, Mann-Whitney U-test, and least absolute shrinkage and selection operator (LASSO) regression. A nomogram integrating radiomic and clinical features using a machine learning model was established and evaluated. The SHapley Additive exPlanations were used for model interpretability and visualization.

Results: The FCN ResNet101 demonstrated the highest segmentation performance for adnexal masses (Dice similarity coefficient: 89.1%). Support vector machine achieved the best AUC (0.961, 95% CI: 0.925-0.996). The nomogram using the LightGBM algorithm reached the best AUC (0.966, 95% CI: 0.927-1.000). The diagnostic performance of the nomogram was comparable to that of experienced radiologists (p > 0.05) and outperformed that of less-experienced radiologists (p < 0.05). The model significantly improved the diagnostic accuracy of less-experienced radiologists.

Conclusions: The segmentation model serves as a valuable tool for the automated delineation of adnexal lesions. The machine learning model exhibited commendable classification capability and outperformed the diagnostic performance of less-experienced radiologists.

Critical relevance statement: The ultrasound radiomics-based machine learning model holds the potential to elevate the professional ability of less-experienced radiologists and can be used to assist in the clinical screening of ovarian cancer.

Key points: We developed an image segmentation model to automatically delineate adnexal masses. We developed a model to classify adnexal masses based on O-RADS. The machine learning model has achieved commendable classification performance. The machine learning model possesses the capability to enhance the proficiency of less-experienced radiologists. We used SHapley Additive exPlanations to interpret and visualize the model.

Objectives: To investigate the image quality and diagnostic performance with ultra-low dose dual-layer detector spectral CT (DLSCT) by various reconstruction techniques for evaluation of pulmonary nodules.

Materials and methods: Between April 2023 and December 2023, patients with suspected pulmonary nodules were prospectively enrolled and underwent regular-dose chest CT (RDCT; 120 kVp/automatic tube current) and ultra-low dose CT (ULDCT; 100 kVp/10 mAs) on a DLSCT scanner. ULDCT was reconstructed with hybrid iterative reconstruction (HIR), electron density map (EDM), and virtual monoenergetic images at 40 keV and 70 keV. Quantitative and qualitative image analysis, nodule detectability, and Lung-RADS evaluation were compared using repeated one-way analysis of variance, Friedman test, and weighted kappa coefficient.

Results: A total of 249 participants (mean age ± standard deviation, 50.0 years ± 12.9; 126 male) with 637 lung nodules were included. ULDCT resulted in a significantly lower mean radiation dose than RDCT (0.3 mSv ± 0.0 vs. 3.6 mSv ± 0.8; p < 0.001). Compared with RDCT, ULDCT EDM showed significantly higher signal-noise-ratio (44.0 ± 77.2 vs. 4.6 ± 6.6; p < 0.001) and contrast-noise-ratio (26.7 ± 17.7 vs. 5.0 ± 4.4; p < 0.001) with qualitative scores ranked higher or equal to the average. Using the regular-dose images as a reference, ULDCT EDM images had a satisfactory nodule detection rate (84.6%) and good inter-observer agreements compared with RDCT (κw > 0.60).

Conclusion: Ultra-low dose dual-layer detector CT with 91.2% radiation dose reduction achieves sufficient image quality and diagnostic performance of pulmonary nodules.

Critical relevance statement: Dual-layer detector spectral CT enables substantial radiation dose reduction without impairing image quality for the follow-up of pulmonary nodules or lung cancer screening.

Key points: Radiation dose is a major concern for patients requiring pulmonary nodules CT screening. Ultra-low dose dual-layer detector spectral CT with 91.2% dose reduction demonstrated satisfactory performance. Dual-layer detector spectral CT has the potential for lung cancer screening and management.

Objectives: To evaluate the value of contrast-enhanced CT in diagnosing ultrasonography-unspecified adnexal torsion (AT).

Methods: Surgically confirmed patients with painful pelvic masses (n = 165) were retrospectively collected from two institutes. Two senior radiologists independently reviewed the CT images and determined the Hounsfield unit difference between non-contrast vs portal venous phases (ΔHU_PV-NC) in both derivation and validation samples. The cutoff value, sensitivity, specificity, predictivity, and reproducibility of the ΔHU_PV-NC and other visually assessed CT signs were analyzed and compared using the receiver-operating characteristic curve, multivariable regression, and inter-rater agreement assays, respectively.

Results: Women with twisted (n = 73 [47 ± 19 years]) or untwisted (n = 92 [40 ± 15 years]) adnexal lesions were reviewed. The ΔHU_PV-NC ≤ 17.5 HU (AUC: 0.91 [95% CI: 0.86, 0.96]; sensitivity: 95% [95% CI: 87, 98]; and specificity: 88% [95% CI: 80, 94]) was the independent predictor of AT (OR: 137 [95% CI: 39, 481], p < 0.001). After training in ΔHU_PV-NC measurement, the agreement between two junior residents and the consensus increased from fair (resident-1: 0.29 [95% CI: 0.17, 0.41]; resident-2: 0.24 [95% CI: 0.1, 0.39]) to substantial (resident-1: 0.75 [95% CI: 0.65, 0.85]; resident-2: 0.72 [95% CI: 0.62, 0.83]). The post-training diagnostic accuracy (both residents: 81% [95% CI: 74, 87]) was higher than the pre-training accuracy (resident-1: 67% [95% CI: 59, 74], p = 0.007; resident-2: 66% [95% CI: 58, 73], p = 0.002).

Conclusion: The sign of ΔHU_PV-NC ≤ 17.5 HU in contrast-enhanced CT can be used to predict the ultrasonography-unspecified AT.

Critical relevance statement: The decreased attenuation difference between non-contrast vs portal venous phases, a quantitative measurement-based CT sign, highlights the value of using contrast-enhanced CT as a second-line imaging approach after an equivocal ultrasonographic examination to diagnose AT in emergency settings.

Key points: The value of contrast-enhanced CT in diagnosing ultrasonography-unspecified AT is underestimated. The ΔHU_PV-NC ≤ 17.5 HU is the only predictor to diagnose the ultrasonography-unspecified AT. Contrast-enhanced CT can be used as a second-line imaging approach after an equivocal ultrasonographic examination.

Purpose: To assess the efficacy of bolus injections of landiolol hydrochloride as premedication in coronary artery CT angiography (CCTA).

Methods: The study population consisted of 37 patients (17 female; median age, 56 years; IQR, 19 years; range, 19-88 years) who underwent CCTA after intravenous injection of landiolol hydrochloride due to a heart rate > 60 bpm. Landiolol hydrochloride was administered in a stepwise manner until a heart rate of ≤ 60 bpm was achieved or a maximum dose of 60 mg was reached after six injections. Heart rates routinely displayed continuously on the CT scanner before the start of the landiolol hydrochloride injection (HR_PRE), after each partial dose (HR_1-6), during the CT scan (HR_CT), and after the examination before moving from the CT table (HR_POST) were recorded. Furthermore, the blood pressure routinely measured before (BP_PRE) and after the examination before moving from the CT table (BP_POST) was recorded.

Results: A HR_CT of ≤ 60 bpm was achieved in 13 patients (35%) and a HR_CT ≤ 65 bpm was achieved in 25 patients (68%). The mean difference (± SD) between HR_PRE and HR_CT was -11 ± 9 bpm in total, -14 ± 10 bpm in patients without oral beta-blocker premedication and -6 ± 5 bpm in patients with oral Beta-blocker premedication.

Conclusions: Landiolol hydrochloride enables a reduction of the heart rate in patients with and without oral beta-blocker premedication, whereby the use of serial partial doses is a simple and effective approach in clinical routine.

Critical relevance statement: In cardiac CT, weight-independent, stepwise landiolol hydrochloride injection up to 40 mg reduces heart rate by -14 bpm without and -5 bpm with oral beta-blocker premedication, and achieves heart rates of ≤ 65 bpm in a significant proportion of patients.

Key points: The ideal heart rate for cardiac CT is ≤ 60-65 bpm, which improves image quality and reduces radiation dose. In cardiac CT, landiolol hydrochloride intravenously reduces heart rate by -14 bpm. Heart rate of ≤ 65 bpm can be achieved in a significant proportion of patients.

Objective: To determine the feasibility, yield, and safety of fluoroscopic-guided aspiration of the acutely dislocated total hip arthroplasty (AD-THA).

Materials and methods: IRB-approved, retrospective review of fluoroscopic-guided aspirations of AD-THA (January 2005-December 2023) was performed. Data from electronic charts and fluoroscopy images/reports were obtained. Positive yield was defined as spontaneous aspirate or saline rinse adequate for microbiology analysis. Sub-analysis by needle target (acetabular cup or femur) was performed for spontaneous aspiration rate, aspirate volume and fluoroscopy time. Differences between groups were analyzed with unpaired, t-test (2-tail) and between proportions with Fisher's exact test, with significance p < 0.05.

Results: Aspiration of 20 AD-THA in 19 patients (12 female, mean age (SD) of 73 years (16)) targeted the acetabular cup in 45% (9/20) or femur in 55% (11/20) of cases. Positive yield was obtained in 95% (19/20), with spontaneous aspirate in 75% (15/20) and saline rinse in 20% (4/20) of cases; in 5% (1/20), no diagnostic sample was obtained. Spontaneous aspirate mean volume (SD, range) for all cases was 8.3 mL (6.9, 0.2-25), and higher when targeting the acetabular cup 11.2 mL (6.9, 5-25) versus the femur 4.0 mL (4.4, 0.2-12) (p = 0.026). The rate of spontaneous aspiration was higher for the acetabular cup 100% (9/9) versus the femur 55% (6/11) (p = 0.038). The mean fluoroscopy time (SD, range) for all cases was 43 s (25, 19-102), and shorter for targeting the acetabular cup 32 s (16, 19-75) versus the femur 56 s (28, 28-102) (p = 0.034). No immediate complications occurred in all aspirations.

Conclusion: Fluoroscopy-guided aspiration of AD-THA is a feasible, high-yield, and safe procedure. Targeting the acetabular cup results in a higher rate of spontaneous aspirate, larger aspiration volume, and lower fluoroscopy time.

Critical relevance statement: Although technically more challenging, radiologists should feel confident aspirating the acutely dislocated total hip arthroplasty (AD-THA) under fluoroscopic guidance.

Key points: Total hip arthroplasty (THA) infection can be evaluated with synovial fluid aspiration. Fluoroscopic-guided aspiration of the dislocated THA is feasible, high-yield, and safe. Targeting of the acetabular cup is recommended over the femoral prosthetic component. Acetabular cup targeting gives larger, spontaneous aspirates with lower fluoroscopy time.

Introduction: A large number of middle-aged and elderly patients have an insufficient understanding of osteoporosis and its harm. This study aimed to establish and validate a convolutional neural network (CNN) model based on unenhanced chest computed tomography (CT) images of the vertebral body and skeletal muscle for opportunistic screening in patients with osteoporosis.

Materials and methods: Our team retrospectively collected clinical information from participants who underwent unenhanced chest CT and dual-energy X-ray absorptiometry (DXA) examinations between January 1, 2022, and December 31, 2022, at four hospitals. These participants were divided into a training set (n = 581), an external test set 1 (n = 229), an external test set 2 (n = 198) and an external test set 3 (n = 118). Five CNN models were constructed based on chest CT images to screen patients with osteoporosis and compared with the SMI model to predict the performance of osteoporosis patients.

Results: All CNN models have good performance in predicting osteoporosis patients. The average F1 score of Densenet121 in the three external test sets was 0.865. The area under the curve (AUC) of Desenet121 in external test set 1, external test set 2, and external test set 3 were 0.827, 0.859, and 0.865, respectively. Furthermore, the Densenet121 model demonstrated a notably superior performance compared to the SMI model in predicting osteoporosis patients.

Conclusions: The CNN model based on unenhanced chest CT vertebral and skeletal muscle images can opportunistically screen patients with osteoporosis. Clinicians can use the CNN model to intervene in patients with osteoporosis and promptly avoid fragility fractures.

Critical relevance statement: The CNN model based on unenhanced chest CT vertebral and skeletal muscle images can opportunistically screen patients with osteoporosis. Clinicians can use the CNN model to intervene in patients with osteoporosis and promptly avoid fragility fractures.

Key points: The application of unenhanced chest CT is increasing. Most people do not consciously use DXA to screen themselves for osteoporosis. A deep learning model was constructed based on CT images from four institutions.

Objectives: To develop and validate the performance of CT-based radiomics models for predicting the prognosis of acute pancreatitis.

Methods: All 344 patients (51 ± 15 years, 171 men) in a first episode of acute pancreatitis (AP) were retrospectively enrolled and randomly divided into training (n = 206), validation (n = 69), and test (n = 69) sets with the ratio of 6:2:2. The patients were dichotomized into good and poor prognosis subgroups based on follow-up CT and clinical data. The radiomics features were extracted from contrast-enhanced CT. Logistic regression analysis was applied to analyze clinical-radiological features for developing clinical and radiomics-derived models. The predictive performance of each model was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis (DCA).

Results: Eight pancreatic and six peripancreatic radiomics features were identified after reduction and selection. In the training set, the AUCs of clinical, pancreatic, peripancreatic, radiomics, and combined models were 0.859, 0.800, 0.823, 0.852, and 0.899, respectively. In the validation set, the AUCs were 0.848, 0.720, 0.746, 0.773, and 0.877, respectively. The combined model exhibited the highest AUC among radiomics-based models (pancreatic, peripancreatic, and radiomics models) in both the training (0.899) and validation (0.877) sets (all p < 0.05). Further, the AUC of the combined model was 0.735 in the test set. The calibration curve and DCA indicated the combined model had favorable predictive performance.

Conclusions: CT-based radiomics incorporating clinical features was superior to other models in predicting AP prognosis, which may offer additional information for AP patients at higher risk of developing poor prognosis.

Critical relevance statement: Integrating CT radiomics-based analysis of pancreatic and peripancreatic features with clinical risk factors enhances the assessment of AP prognosis, allowing for optimal clinical decision-making in individuals at risk of severe AP.

Key points: Radiomics analysis provides help to accurately assess acute pancreatitis (AP). CT radiomics-based models are superior to the clinical model in the prediction of AP prognosis. A CT radiomics-based nomogram integrated with clinical features allows a more comprehensive assessment of AP prognosis.

Lung cancer is the leading cause of cancer-related deaths worldwide, with invasive non-mucinous adenocarcinoma (INMA) being the most common type and carrying a poor prognosis. In 2020, the International Association for the Study of Lung Cancer (IASLC) pathology committee proposed a new histological grading system, which offers more precise prognostic assessments by combining the proportions of major and high-grade histological patterns. Accurate identification of lung INMA grading is crucial for clinical diagnosis, treatment planning, and prognosis evaluation. Currently, non-invasive imaging methods (such as CT, PET/CT, and MRI) are increasingly being studied to predict the new grading of lung INMA, showing promising application prospects. This review outlines the establishment and prognostic efficiency of the new IASLC grading system, highlights the application and latest progress of non-invasive imaging techniques in predicting lung INMA grading, and discusses their role in personalized treatment of lung INMA and future research directions. CRITICAL RELEVANCE STATEMENT: The new IASLC grading system has important prognostic implications for patients with lung invasive non-mucinous adenocarcinoma (INMA), and non-invasive imaging methods can be used to predict it, thereby improving patient prognoses. KEY POINTS: The new IASLC grading system more accurately prognosticates for patients with lung INMA. Preoperative prediction of the new grading is challenging because of the complexity of INMA subtypes. It is feasible to apply non-invasive imaging methods to predict the new IASLC grading system.

Objectives: This study aimed to investigate the diagnostic value of spectral parameters of dual-layer spectral detector computed tomography (DLCT) in distinguishing between low- and high-grade bladder cancer (BCa).

Methods: This single-center retrospective study included pathologically confirmed BCa patients who underwent preoperative contrast-enhanced DLCT. Patients were divided into low- and high-grade groups based on pathology. We measured and calculated the following spectral CT parameters: iodine density (ID), normalized ID (NID), arterial enhancement fraction (AEF), extracellular volume (ECV) fraction, virtual non-contrast (VNC), slope of the attenuation curve, and Z effective (Z_eff). Univariate and multivariable logistic regression analyses were used to determine the best predictive factors in differentiating between low- and high-grade BCa. We used receiver operating characteristic curve analysis to assess diagnostic performance and decision curve analysis to determine the net benefit.

Results: The study included 64 patients (mean age, 64 ± 11.0 years; 46 men), of whom 42 had high-grade BCa and 22 had low-grade BCa. Univariate analysis revealed that differences in ID and NID in the corticomedullary phase, AEF, ECV, VNC, and Z_eff images were statistically significant (p = 0.001-0.048). Multivariable analysis found that AEF was the best predictor of high-grade tumors (p = 0.006). With AEF higher in high-grade BCa, AEF results were as follows: area under the curve (AUC), 0.924 (95% confidence interval, 0.861-0.988); sensitivity, 95.5%; specificity, 81.0%; and accuracy, 85.9%. The cutoff valve of AEF for predicting high-grade BCa was 67.7%.

Conclusion: Using DLCT AEF could help distinguish high-grade from low-grade BCa.

Critical relevance statement: This research demonstrates that the arterial enhancement fraction (AEF), a parameter derived from dual-layer spectral detector CT (DLCT), effectively distinguishes between high- and low-grade bladder cancer, thereby aiding in the selection of appropriate clinical treatment strategies.

Key points: This study investigated the value of dual-layer spectral detector CT in the assessment of bladder cancer (BCa) histological grade. The spectral parameter arterial enhancement fraction could help determine BCa grade. Our results can help clinicians formulate initial treatment strategies and improve prognostications.

Purpose: This study compares the diagnostic efficacy of non-contrast abbreviated MRI protocols with Gadoxetic acid-enhanced abbreviated MRI for detecting colorectal liver metastasis (CRLM), focusing on lesion characterization and surveillance.

Methods: Ninety-four patients, including 55 with pathologically verified CRLM, were enrolled, totaling 422 lesions (287 metastatic, 135 benign). Two independent readers assessed three MRI protocols per patient: Protocol 1 included non-contrast sequences (T2-weighted turbo spin-echo, T1-weighted Dixon, diffusion-weighted imaging (DWI), and ADC mapping). Protocol 2 included gadoxetic acid enhancement with hepatobiliary phase imaging, T2 TSE, DWI, and ADC maps. Protocol 3 utilized the standard Gadoxetic Acid-enhanced MRI sequence, which included pre-contrast T1-weighted imaging, T1-weighted Dixon sequences, post-contrast T1-weighted imaging (including arterial, portal venous, transitional and hepatobiliary phases), and additional T2-weighted and DWI sequences. Diagnoses were scored on a 5-point scale (benign = 1; malignant = 5), with scores ≥ 3 indicating CRLM. ROC curves analyzed diagnostic accuracy, comparing area under the curve (AUC) values across protocols.

Results: No significant difference in AUCs was observed between Protocol 1 (0.899-0.909) and Protocol 2 (0.906-0.931) versus Protocol 3 (0.935-0.939) (p = 0.091-0.195). For lesions ≤ 10 mm, Protocol 1 was slightly inferior to Protocol 3 (p = 0.002-0.032), while Protocol 2 remained comparably effective (p = 0.096-0.179). These findings held when using a threshold of ≥ 4 to define CRLM.

Conclusion: The non-enhanced abbreviated MRI protocol is as effective as the gadoxetic acid-enhanced protocol in identifying CRLM. The proposed Ab-MRI approach may be a viable alternative for CRLM surveillance.

Critical relevance statement: The non-enhanced abbreviated MRI (Ab-MRI) protocol is as effective as the gadoxetic acid-enhanced protocol in identifying colorectal liver metastasis (CRLM). The proposed Ab-MRI approach may be a viable alternative for CRLM surveillance.

Key points: Two abbreviated protocols are proposed for colorectal liver metastasis (CRLM) surveillance. The non-enhanced protocol showed equivalent efficacy and was more cost-effective. The non-enhanced protocol may be a viable alternative for CRLM surveillance.