BMC Medical Imaging最新文献

Background: This study aimed to describe the color Doppler flow features of papillary thyroid carcinoma (PTC) and to further investigate the associations between these features and lymph node metastasis (LNM).

Methods: A retrospective analysis of the clinical data of 287 PTC patients confirmed by postoperative pathology at the Second Affiliated Hospital of Xi'an Jiaotong University from January 2022 to April 2023 was conducted. The Adler grading system and novel blood flow patterns were used to analyze the vascularity of the PTC lesions on color Doppler images. Univariate and multivariate logistic regression analyses were conducted to evaluate the independent effects of blood flow characteristics on LNM, and a logistic regression model was established to assess their predictive value for PTC-related LNM.

Results: In all, 287 PTC lesions were analyzed using color Doppler ultrasonography, which identified five main reference patterns: avascular (26.13%), dot-line (24.74%), branching (14.29%), garland (11.50%), and rich-disorganized (23.34%). The Adler blood flow grading was as follows: 0 (32.75%), I (18.82%), II (19.16%), and III (29.27%). A univariate analysis revealed that the Adler grade was not significantly associated with LNM (P > 0.05), whereas the garland pattern was significantly associated with LNM (P < 0.05). A multivariate analysis revealed that the garland pattern was an independent protective factor for LNM (OR [95% CI] = 0.386 [0.156-0.893]). The incorporation of the garland pattern into the model improved the predictive accuracy for LNM in PTC patients, and the AUC increased from 0.727 [95% CI: 0.669-0.786] to 0.767 [95% CI: 0.731-0.821].

Conclusions: This study classifies PTC into five types on the basis of color Doppler flow features and highlights the garland pattern as a potential predictor of LNM risk.

Background: Considering the potential association between radiation-induced hypothyroidism (RHT) and the thyroid subregions as well as the received radiation dose in each subregion, this study aims to develop a subregional prediction model for RHT.

Methods: CT images and dose images of 128 patients with nasopharyngeal carcinoma were collected retrospectively. The thyroid subregion was obtained by clustering thyroid voxels and voxel entropy. After extracting 1781 radiomics features and 1767 dosiomics features, a subregional RHT prediction model was established, and its performance was compared with that of the whole thyroid model. The phenotype and dosimetry parameters of each subregion were analyzed by AUC, T test and Delong test.

Results: Three subregions (S1, S2, S3) were identified. The subregional prediction model was constructed based on 34 radiomics and dosiomics features. According to the Delong test, the prediction performance of the subregional model was significantly superior than that of the whole thyroid model (0.813 VS 0.624, p = 0.038). Subregional analysis suggests that S1 and S3 regions may have higher radiosensitivity than S2 regions.

Conclusions: In this study, a subregional model for predicting RHT was established and the radiosensitivity of the relevant subregions was evaluated. The subregion-based RHT prediction model may help to improve radiotherapy plan design for better thyroid function protection.

Objective: To evaluate the effectiveness of the variable rate bolus tracking technique combined with third-generation dual-source CT dual-energy scanning in enhancing the quality of head and neck vascular CT angiography (CTA).

Methods: We conducted a retrospective analysis of 202 patients who underwent head and neck vascular CTA using a third-generation dual-source CT with dual-energy scanning. Patients were divided based on the contrast injection method into two groups: the variable-rate bolus tracking group (Group A, n = 100) and the fixed flow rate group (Group B, n = 102). We compared subjective image quality, venous artifacts, and objective image quality parameters between the two groups.

Results: The amount of contrast agent used in Group A was significantly lower than in Group B. Additionally, mean attenuation values of arterial segments in Group A were markedly lower than those in Group B. Compared to Group B, attenuation values of the intracranial venous sinuses, right jugular vein, superior vena cava, right subclavian vein, and left jugular vein in Group A showed significant reductions. No significant difference was observed in the subjective image quality between the two groups. However, venous artifact in the right subclavian vein was significantly diminished in Group A.

Conclusion: The application of the variable rate bolus tracking technique alongside third-generation dual-source CT dual-energy scanning in head and neck vascular CTA can achieve high-quality imaging while reducing contrast agent dosage. It enhances the attenuation contrast of intracranial arteries and veins and minimizes residual contrast and artifacts in the right subclavian vein.

Objectives: To explore the feasibility of a one-beat protocol and ultra-low tube voltage of 60 kVp in coronary CT angiography (CCTA).

Methods: This prospective study enrolled 107 patients (body mass index ≤ 26 kg/m²) undergoing CCTA examinations. Specifically, the conventional group (n = 52) underwent 100 kVp scanning with 45 ml iodine contrast agent and 4 ml/s injection rate, and the low-dose group (n = 55) underwent 60 kVp scanning with 28 ml iodine contrast agent and 2.5 ml/s injection rate. The CT value, signal-noise-ratio (SNR), contrast-noise-ratio (CNR) and subjective image quality score of two groups in aorta (AO), right coronary artery (RCA), left anterior descending (LAD) and left circumflex (LCX) are analyzed in this study. Three types of radiation doses [i.e., volume CT dose index (CTDIvol), dose length product (DLP), effective dose (ED)] of two groups are also compared.

Results: The quantitative results indicated that the low-dose group achieved higher CT values, SNR and CNR results of the AO than the conventional group (P values < 0.001). Both groups had similar CT values, SNR and CNR results in RCA, LAD, and LCX (P values > 0.05). A good agreement is noted with respect to subjective image quality scores in both groups, while the Cohen's kappa value is 0.815 in the low-dose group and 0.825 in the conventional group, respectively. In addition, the radiation dose of the low-dose group is significantly lower than the conventional group in terms of CTDIvol, DLP and ED values, and the contrast dose in the low-dose group is also significantly reduced compared to the conventional group (P values < 0.001).

Conclusions: One-beat protocol with an ultra-low tube voltage of 60 kVp could provide improved coronary image quality, reduced radiation dose and reduced iodine contrast dose.

Background: Pulmonary diseases such as COVID-19 and pneumonia, are life-threatening conditions, that require prompt and accurate diagnosis for effective treatment. Chest X-ray (CXR) has become the most common alternative method for detecting pulmonary diseases such as COVID-19, pneumonia, and lung opacity due to their availability, cost-effectiveness, and ability to facilitate comparative analysis. However, the interpretation of CXRs is a challenging task.

Methods: This study presents an automated deep learning (DL) model that outperforms multiple state-of-the-art methods in diagnosing COVID-19, Lung Opacity, and Viral Pneumonia. Using a dataset of 21,165 CXRs, the proposed framework introduces a seamless combination of the Vision Transformer (ViT) for capturing long-range dependencies, DenseNet201 for powerful feature extraction, and global average pooling (GAP) for retaining critical spatial details. This combination results in a robust classification system, achieving remarkable accuracy.

Results: The proposed methodology delivers outstanding results across all categories: achieving 99.4% accuracy and an F1-score of 98.43% for COVID-19, 96.45% accuracy and an F1-score of 93.64% for Lung Opacity, 99.63% accuracy and an F1-score of 97.05% for Viral Pneumonia, and 95.97% accuracy with an F1-score of 95.87% for Normal subjects.

Conclusion: The proposed framework achieves a remarkable overall accuracy of 97.87%, surpassing several state-of-the-art methods with reproducible and objective outcomes. To ensure robustness and minimize variability in train-test splits, our study employs five-fold cross-validation, providing reliable and consistent performance evaluation. For transparency and to facilitate future comparisons, the specific training and testing splits have been made publicly accessible. Furthermore, Grad-CAM-based visualizations are integrated to enhance the interpretability of the model, offering valuable insights into its decision-making process. This innovative framework not only boosts classification accuracy but also sets a new benchmark in CXR-based disease diagnosis.

Background: Acute-on-chronic liver failure (ACLF) is a life-threatening hepatic syndrome. Therefore, this study aimed to develop a comprehensive model combining extracellular liver volume derived from spectral CT (ECV_IC-liver) and sarcopenia, for the early prediction of short-term (90-day) disease progression in ACLF.

Materials and methods: A retrospective cohort of 126 ACLF patients who underwent hepatic spectral CT scans was included. According to the Asia-Pacific Association for the Study of the Liver (APASL) criteria, patients were divided into the progression group (n = 70) and the stable group (n = 56). ECV_IC-liver was measured on the equilibrium period (EP) images of spectral CT, and L3-SMI was measured on unenhanced CT images, with sarcopenia assessed. A comprehensive model was developed by combining independent predictors. Model performance was evaluated using receiver operating characteristic (ROC) curve analysis, calibration curves, and decision curve analysis (DCA).

Results: In the univariate analysis, BMI, WBC, PLT, PTA, L3-SMI, IC-EP, Z-EP, K₁₄₀-EP, NIC-EP, ECV_IC-liver, and Sarcopenia demonstrated associations with disease progression status at 90 days in ACLF patients. In multivariate logistic regression, white blood cell count (WBC) (OR = 1.19, 95% CI: 1.02-1.40; P = 0.026), ECV_IC-liver (OR = 1.27, 95% CI: 1.15-1.40; P < 0.001), sarcopenia (OR = 4.15, 95% CI: 1.43-12.01; P = 0.009), MELD-Na score (OR = 1.06, 95%CI: 1.01-1.13;P = 0.042), and CLIF-SOFA score (OR = 1.37, 95%CI:1.15-1.64; P<0.001) emerged as independent risk factors for ACLF progression. The combined model exhibited superior predictive performance (AUCs = 0.910, sensitivity = 80.4%, specificity = 90.0%, PPV = 0.865, NPV = 0.851) compared to CLIF-SOFA, MELD-Na, MELD and CTP scores(both P < 0.001). Calibration curves and DCA confirmed the high clinical utility of the combined model.

Conclusions: Patients without sarcopenia and/or with a lower ECV_IC-liver have a better prognosis, and the integration of WBC, ECV_IC-liver, Sarcopenia, CLIF-SOFA and MELD-Na scores in a composite model offers a concise and effective tool for predicting disease progression in ACLF patients.

Trial registration: Not Applicable.

Purposes: This retrospective study was designed to explore the relationship between right ventricular fibroblast activation measured by fibroblast activation protein inhibitor (FAPI) imaging and myocardial deformation measured by Speckle‑tracking Echocardiography (STE) in patients with pulmonary hypertension (PH).

Methods: Clinical data of PH patients were collected [15 chronic thromboembolic pulmonary hypertension (CTEPH), 4 PAH, 1 PH with unclear and/or multifactorial mechanisms]. All of patients underwent FAPI imaging and echocardiography within one month. FAPI activity of right ventricle higher than that in the blood pool was defined as abnormal. The global and segmental maximum standardised uptake values (SUV_max) of the right ventricle were measured and further expressed as target-to-background ratio (TBR) with blood pool activity as background. right ventricular global longitudinal strain (RVGLS) and right ventricular free wall longitudinal strain (RVFWLS) including the basal-, mid-, and apical-segments were measured by STE.

Results: Eighteen patients with PH showed abnormal FAPI uptake in right ventricle. No significant differences were found between CTEPH and other types of PH. TBR of right ventricle had negative correlations with RVGLS (r = -0.597, P = 0.005) and RVFWLS (r = -0.586, P = 0.007) at global level. While, at regional level, significant correlation was only demonstrated between TBR of right ventricle free wall and RVFWLS in apical region (r = -0.530, P = 0.016) and middle region (r = -0.457, P = 0.043). Among the traditional Echocardiography parameters, TBR of right ventricle were positively associated with thickness of right ventricular anterior wall (RVAW) (r_s = 0.475, P = 0.034), and inversely with right ventricular systolic function [RVFAC (r = -0.586, P = 0.007) and TAPSE (r = -0.565, P = 0.009)].

Conclusion: FAPI imaging can partially reflect the right ventricular strain reduction in patients with PH.

Purposes: To develop a predictive model for recurrent intussusception based on abdominal ultrasound (US) images and abdominal radiographs.

Methods: A total of 3665 cases of intussusception were retrospectively collected from January 2017 to December 2022. The cohort was randomly assigned to training and validation sets at a 6:4 ratio. Two types of images were processed: abdominal grayscale US images and abdominal radiographs. These images served as inputs for the deep learning algorithm and were individually processed by five detection models for training, with each model predicting its respective categories and probabilities. The optimal models were selected individually for decision fusion to obtain the final predicted categories and their probabilities.

Results: With US, the VGG11 model showed the best performance, achieving an area under the receiver operating characteristic curve (AUC) of 0.669 (95% CI: 0.635-0.702). In contrast, with radiographs, the ResNet18 model excelled with an AUC of 0.809 (95% CI: 0.776-0.841). We then employed two fusion methods. In the averaging fusion method, the two models were combined to reach a diagnostic decision. Specifically, a soft voting scheme was used to average the probabilities predicted by each model, resulting in an AUC of 0.877 (95% CI: 0.846-0.908). In the stacking fusion method, a meta-model was built based on the predictions of the two optimal models. This approach notably enhanced the overall predictive performance, with LightGBM emerging as the top performer, achieving an AUC of 0.897 (95% CI: 0.869-0.925). Both fusion methods demonstrated excellent performance.

Conclusions: Deep learning algorithms developed using multimodal medical imaging may help predict recurrent intussusception.

Clinical trial number: Not applicable.

Background: Single intrauterine fetal death (sIUFD) will lead to an increased risk of adverse events such as fetal brain abnormalities in the survivor. However, how to detect these anomalies in the early stages remains to be explored.

Objective: To compare apparent diffusion coefficient (ADC) values of fetal brain in cases of single intrauterine fetal death (sIUFD) with twins control and singleton control using diffusion weighted imaging (DWI), and to perform follow-up study to reveal the underlying cerebral microstructure changes.

Materials and methods: In this prospective MRI-based cohort study, we compared 43 surviving fetuses of sIUFD (18 following selective fetal reduction, 2 following laser ablation treatment for twin-to-twin transfusion syndrome, and 23 spontaneous) with 2 control cohorts ( 43 healthy twin fetuses, 43 singletons). All fetuses underwent fetal brain MRI. DWI was performed and ADC map was reconstructed. ADC values of certain regions were compared among the three groups.

Results: ADC values were lower in bilateral white matter of frontal, parietal, temporal lobes and cerebellum in surviving fetuses compared with twins control and singleton control, respectively. ADC values of bilateral basal ganglia, thalamus and cerebellum in surviving fetuses, that of bilateral frontal lobes, cerebellum in twins control and that of right temporal lobe, left basal ganglia, and bilateral cerebellum in singleton control, were negatively correlated with gestational age. ADC values of left cerebellum in surviving fetuses were positively correlated with interval time.

Conclusions: DWI is a very useful sequence for detecting underlying changes. ADC value might be a effective indicator of subtle anomalies in surviving fetuses.

Objective: By discussing the difference, stability and classification ability of tumor contour extracted by artificial intelligence and doctors, can a more stable method of tumor contour extraction be obtained?

Methods: We propose a novel framework for the automatic segmentation of lung tumor contours and the differential diagnosis of downstream tasks. This framework integrates four key modules: tumor segmentation, extraction of radiomic features, feature selection, and the development of diagnostic models for clinical applications. Using this framework, we conducted a study involving a cohort of 1,429 patients suspected of lung cancer. Four automatic segmentation methods (RNN, UNET, WFCM, and SNAKE) were evaluated against manual segmentation performed by three radiologists with varying levels of expertise. We further studied the consistency of radiomic features extracted from these methods and evaluates their diagnostic performance across three downstream tasks: benign vs. malignant classification, lung adenocarcinoma infiltration, and lung nodule density classification.

Results: The Dice coefficient of RNN is the highest among the four automatic segmentation methods (0.803 > 0.751, 0.576, 0.560), and all P < 0.05. In the consistency comparison of the seven contour-extracted radiomic features, that the features extracted by RNN and S1 (the senior radiologist) showed the highest similarity which was higher than the other automatic segmentation methods and doctors with low seniority. In all three downstream tasks, the radiomic features extracted from RNN segmentation contours showed the highest diagnostic discrimination. In the classification of benign and malignant nodules, the RNN method performed slightly better than the S1 method, with an AUC of 0.840 ± 0.01 and 0.824 ± 0.015, respectively, and significantly better than the other five methods. Similarly, the RNN method had an AUC value of 0.946 in lung adenocarcinoma infiltration, and a kappa value of 0.729 in lung nodule density classification, both of which were better than the other six methods.

Conclusions: Our findings suggest that AI-driven tumor segmentation methods can enhance clinical decision-making by providing reliable and reproducible results, ultimately emphasizing the auxiliary role of automated tumor contouring in clinical practice. The findings will have important implications for the application of radiomics in clinical practice.