BMC Medical Imaging[JOURNAL]
Sun
200 papers
BCP-YOLO: a lightweight and interpretable YOLOv8-based framework for real-time coloreactal polyp detection.
Fei X, Bao Y, Dai C … +1 more , Hu K
BMC Med Imaging · 2026 Jun · PMID 42365237 · Full text
BACKGROUND: Colorectal cancer is one of the most prevalent malignant tumors worldwide. Early screening relies on accurate polyp detection during colonoscopy. Polyps in colonoscopic images exhibit diverse morphologies, in...
BACKGROUND: Colorectal cancer is one of the most prevalent malignant tumors worldwide. Early screening relies on accurate polyp detection during colonoscopy. Polyps in colonoscopic images exhibit diverse morphologies, indistinct boundaries, and low contrast. Specular reflections, fold occlusions, and imaging artifacts further complicate detection, which fail to meet the requirements of real-time clinical assistance. METHODS: To address these challenges, we propose BCP-YOLO (You Only Look Once), a high-precision, relatively lightweight polyp detection framework built upon an improved YOLOv8 architecture, designed to achieve a well-balanced trade-off between detection accuracy and computational efficiency. First, to mitigate complex background interference and improve small polyp detection, a BiFormer module is integrated into the backbone network to enhance focus on salient polyp regions while suppressing noise. To alleviate boundary ambiguity, the CARAFE content-aware upsampling operator is incorporated into the feature fusion stage, to refine lesion boundaries and spatial details. PConv module is employed to optimize network efficiency, reducing computational cost while maintaining detection performance. RESULTS: Experimental results on the Kvasir-SEG and CVC-ClinicDB datasets demonstrate that BCP-YOLO achieves a mean average precision (mAP) of 88.5% on Kvasir-SEG, representing a 3.4% improvement over the YOLOv8 baseline. Precision and recall increase by 5.5% and 1.3%, respectively. The model contains 11.7 M parameters and achieves an inference speed of 104.1 frames per second (FPS). Five-fold cross-validation on both datasets validates its strong generalization capability and robustness. CONCLUSION: The method provides a high-accuracy and deployable solution for computer-aided diagnosis in real-time colonoscopy, offering significant potential to improve the reliability and efficiency of early colorectal cancer screening.
A two-stage deep learning model with segmentation-guided top-K slice selection for patient-level PAS prediction on MRI.
Song C, Chen S, Ye W … +2 more , Guo Y, Li Y
BMC Med Imaging · 2026 Jun · PMID 42363169 · Full text
OBJECTIVE: Placenta accreta spectrum (PAS) can cause severe obstetric complications, and accurate preoperative assessment is critical. However, heterogeneous MRI manifestations and the high cost of fine placental annotat...
OBJECTIVE: Placenta accreta spectrum (PAS) can cause severe obstetric complications, and accurate preoperative assessment is critical. However, heterogeneous MRI manifestations and the high cost of fine placental annotations limit deep learning applications. This study developed and validated a two-stage segmentation-guided MRI framework for automated placental localization, ROI selection, and patient-level PAS prediction. METHODS: A total of 170 third-trimester placental MRI cases were included in the internal cohort, comprising 87 PAS and 83 non-PAS cases. Data were split at the patient level into training, internal validation, and internal test sets. An external validation cohort of 54 patients, including 36 PAS and 18 non-PAS cases, was additionally used. A 2D U-Net-based segmentation model first generated placental probability maps for ROI extraction. Selected ROIs were then classified using a ResNet-18 model, and slice-level probabilities were aggregated into patient-level predictions. The operating threshold was determined using the Youden index on the internal validation set. RESULTS: For patient-level PAS classification, the internal validation set achieved an AUC of 0.85 (95% CI: 0.68-0.97), and the Youden-derived operating threshold was 0.61. In the held-out internal test set, the model achieved an AUC of 0.75 (95% CI: 0.53-0.92). In the external validation cohort of 54 patients, including 36 PAS and 18 non-PAS cases, the model achieved an AUC of 0.72 (95% CI: 0.54-0.87), sensitivity of 0.78 (95% CI: 0.62-0.88), specificity of 0.56 (95% CI: 0.34-0.75), and accuracy of 0.70 (95% CI: 0.57-0.81). CONCLUSION: This two-stage segmentation-guided MRI framework achieved reliable placental segmentation and moderate external performance for patient-level PAS prediction. The model may serve as an auxiliary screening and risk stratification tool, but further prospective multicenter validation and optimization are required before clinical implementation.
F-FDG PET/CT in the evaluation of femoral head osteonecrosis in patients with lymphoma.
Song L, Li H, Zhu A … +1 more , Zhang W
BMC Med Imaging · 2026 Jun · PMID 42351067 · Full text
OBJECTIVE: To describe the imaging features of osteonecrosis of the femoral head (ONFH) on F-fluorodeoxyglucose (F-FDG) positron emission tomography (PET)/CT and explore whether PET/CT can raise suspicion for ONFH in pat...
OBJECTIVE: To describe the imaging features of osteonecrosis of the femoral head (ONFH) on F-fluorodeoxyglucose (F-FDG) positron emission tomography (PET)/CT and explore whether PET/CT can raise suspicion for ONFH in patients with lymphoma. METHODS: A retrospective analysis was conducted on the clinical data, PET/CT, and MRI findings of 17 patients with lymphoma and ONFH. The F-FDG uptake of ONFH was recorded, and the maximum standardized uptake value (SUV) of ONFH was measured. The staging and extent of ONFH, and other bone involvements, were visually assessed. Lymphoma disease status was evaluated using the Deauville criteria. RESULTS: A total of 31 femoral heads were involved (2 stage 1, 24 stage 2, 4 stage 3, 1 stage 4). The median SUV was higher in stage 3-4 lesions (5.27) than in stage 1-2 lesions (1.37) (P = 0.002). Regarding F-FDG uptake patterns, all stage 1 lesions showed normal uptake; among stage 2 lesions, 11 showed increased uptake, 5 showed peripheral increased with central decreased uptake, 2 showed decreased uptake, and 6 showed normal uptake; all stage 3-4 lesions demonstrated increased uptake. On CT, the extent of all stage 2-4 lesions matched MRI findings. Nine patients had osteonecrosis in other bones, including the humeral heads (n = 9), bilateral iliac bones (n = 6), and vertebrae/ribs/scapulae (n = 1). CONCLUSIONS: In patients with lymphoma, ONFH exhibits variable degrees of F-FDG uptake and may be accompanied by involvement of other bones. PET/CT may raise suspicion for ONFH and detect multiple bone involvements during lymphoma assessment.
Selection of proper artificial intelligence techniques developed for CT scan image analysis of liver cancer using fuzzy AHP-TOPSIS.
Lu G, Wang Y, Li C … +1 more , Wu C
BMC Med Imaging · 2026 Jun · PMID 42351048 · Full text
OBJECTIVE: Comprehensive identification and prioritization of developed artificial intelligence methods for applications of radiology can help to select proper techniques. This study aimed to introduce the best artificia...
OBJECTIVE: Comprehensive identification and prioritization of developed artificial intelligence methods for applications of radiology can help to select proper techniques. This study aimed to introduce the best artificial intelligence techniques developed for CT scan image analysis of liver cancer using fuzzy AHP-TOPSIS. MATERIALS AND METHODS: To identify the artificial intelligence techniques developed, a systematic search was performed in five reliable databases. The developed methods were categorized into four groups based on their application type. After that, the Delphi method was applied in two rounds to determine the proper criteria for selecting the best artificial intelligence techniques. To estimate the relative weights of the criteria also, the fuzzy analytical hierarchy process (FAHP) method was used. In the next step, to prioritize the identified artificial intelligence techniques, the technique for order of preference by similarity to the ideal solution (TOPSIS) method was applied. RESULTS: 260 artificial intelligence techniques were identified. Seven selection criteria of validity, accuracy, comprehensiveness, processing time, cost, simplicity, and executive capability were introduced. The deep learning-based image reconstruction model (weight = 0.896) in the group of detection and diagnosis, dual-energy CT deep learning radiomics model (weight = 0.862) in the group of prediction, prognosis, and registration, hybrid densely connected UNet (H-DenseUNet) technique (weight = 0.888) in the group of segmentation and classification, and deep learning, radiomics, and clinical (DLRC) model (weight = 0.956) in the group of treatment and therapy were selected as the proper methods. CONCLUSIONS: These findings represent a total approach to the developed techniques which can be used for designing methods with better performance in the future.
Prognostic value of epicardial adipose tissue assessed by cardiac magnetic resonance in patients with ST-elevation myocardial infarction after successful revascularization.
Wei L, Zhao CX, Dong JX … +4 more , Qiu XH, Lee APW, Ge H, Pu J
BMC Med Imaging · 2026 Jun · PMID 42343325 · Full text
BACKGROUND: Epicardial adipose tissue (EAT) around the myocardium may play a role in coronary artery disease, yet its prognostic value in patients with ST-segment elevation myocardial infarction (STEMI) remains uncertain...
BACKGROUND: Epicardial adipose tissue (EAT) around the myocardium may play a role in coronary artery disease, yet its prognostic value in patients with ST-segment elevation myocardial infarction (STEMI) remains uncertain. METHODS: Data from 540 STEMI patients enrolled in a registry study (NCT03768453) were analyzed. Cardiac magnetic resonance (CMR) cine sequences were used to measure EAT volume. The primary endpoint was a composite of major adverse cardiovascular events (MACE). The Cox proportional hazards model was used to assess the association between EAT volume and MACE. Correlation analyses were performed to explore relationships between EAT volume and myocardial injury markers. In addition, paired t-tests were applied to compare changes in EAT volume over the 6-month follow-up. RESULTS: During a median follow-up of 4.5 years, MACE occurred in 65 patients (12.0%). Patients who experienced MACE had a significantly higher EAT volume (39.6 vs. 33.4 mL, P < 0.001) than those without MACE. EAT volume was an independent predictor of MACE after adjustment for clinical factors (HR = 1.08, 95% CI: 1.02-1.16, P = 0.037) and CMR parameters (HR = 1.04, 95% CI: 1.02-1.07, P < 0.001). EAT volume was significantly correlated with myocardial injury markers, including infarct size (r = 0.15, P = 0.001) and microvascular obstruction (r = 0.15, P = 0.001). EAT volume remained stable at the 6-month follow-up (34.6 vs. 35.6 mL, P = 0.181). CONCLUSIONS: EAT measured non-invasively by CMR is an independent prognostic biomarker for adverse outcomes in STEMI patients. It may serve as a promising imaging marker for risk stratification in this patient population.
Application of 3D Slicer-based three-dimensional visual surgical planning in endoscopic optic canal decompression surgery: a prospective randomised controlled study.
Jin Q, Long Q, Yu J … +5 more , Hu G, Liu Y, Gan P, Ren Z, Liao H
BMC Med Imaging · 2026 Jun · PMID 42343260 · Full text
OBJECTIVE: To evaluate the value of three-dimensional visualization preoperative planning based on the open-source software 3D Slicer in endoscopic endonasal optic nerve decompression for traumatic optic neuropathy. METH...
OBJECTIVE: To evaluate the value of three-dimensional visualization preoperative planning based on the open-source software 3D Slicer in endoscopic endonasal optic nerve decompression for traumatic optic neuropathy. METHODS: A prospective randomized controlled study was conducted. A total of 48 patients with traumatic optic neuropathy who underwent endoscopic endonasal optic nerve decompression at the Affiliated Eye Hospital of Nanchang University between January 2023 and June 2025 were enrolled and randomly assigned to an experimental group (3D Slicer-based three-dimensional visualization planning, n = 25) and a control group (conventional two-dimensional CT planning, n = 23). The primary outcome measures included operative time, intraoperative misjudgment rate, complication rate, surgeon's subjective score (5-point Likert scale), and visual improvement at 3 months postoperatively. Continuous data were analyzed using independent-sample t-test or Mann-Whitney U test, and categorical data were analyzed using Fisher's exact test. RESULTS: There were no statistically significant differences in baseline characteristics or preoperative visual acuity grade between the two groups (P > 0.05). The operative time was significantly shorter in the experimental group than in the control group (129.4 ± 11.6 min vs. 150.7 ± 20.4 min, P < 0.001). Postoperative visual acuity improved significantly compared with preoperative values in both groups (experimental group: P < 0.001; control group: P = 0.003), but the intergroup difference in the grade of visual improvement was not statistically significant (P = 0.439). The intraoperative misjudgment rate was 0% (0/25) in the experimental group and 13.0% (3/23) in the control group, with no statistically significant difference (P = 0.24). The incidence of cerebrospinal fluid leakage was 4.0% (1/25) in the experimental group and 21.7% (5/23) in the control group, and the difference did not reach statistical significance (P = 0.08). The surgeon's subjective score was significantly higher in the experimental group than in the control group [5 (5,5) vs. 4 (4,4), P < 0.001]. No severe complications such as major vessel injury or direct optic nerve injury occurred in either group. CONCLUSION: Three-dimensional visualization preoperative planning based on 3D Slicer can significantly shorten the operative time of endoscopic endonasal optic nerve decompression, enhance the surgeon's confidence, and a potential reduction in cerebrospinal fluid leakage that warrants further investigation. This approach is low-cost, highly generalizable, and has good clinical application value.
The role of magnetic resonance imaging (MRI) in the detection of intracellular lipid content in adrenal adenomas.
Yalçın M, Kahvecioğlu N, Kılıç KK … +2 more , Yapar D, Gürses C
BMC Med Imaging · 2026 Jun · PMID 42337707 · Full text
BACKGROUND: Accurate differentiation of adrenal adenomas from non-adenomatous lesions is critical for patient management. Chemical shift MRI (CSI) detects microscopic intracellular lipid in adenomas through quantitative...
BACKGROUND: Accurate differentiation of adrenal adenomas from non-adenomatous lesions is critical for patient management. Chemical shift MRI (CSI) detects microscopic intracellular lipid in adenomas through quantitative metrics, including chemical shift ratio (CSR) and signal intensity index (SII). However, optimal cutoff values remain heterogeneous across populations, and confounding pathologies such as pheochromocytomas and non-suppressing adenomas challenge diagnostic reliability. PURPOSE: To establish population-specific CSI thresholds and evaluate the incremental value of paraspinal muscle signal intensity in a histopathologically validated Turkish cohort, with particular attention to lipid-poor adenomas and diagnostic confounders. MATERIALS AND METHODS: This retrospective study included 67 consecutive patients (mean age 52.3 ± 14.8 years; 36 adenomas, 31 non-adenomas) with histopathologically confirmed adrenal lesions who underwent preoperative 3.0T CSI MRI. Chemical shift ratio (CSR), signal intensity index (SII), and paraspinal muscle in-phase and out-phase signal intensity (PM-IP & OP SI) were measured using standardized ROI protocols. Interobserver reproducibility was assessed in 25 lesions using intraclass correlation coefficients. ROC curve analysis determined optimal cutoffs, with diagnostic performance calculated at literature-recommended and population-optimized thresholds. RESULTS: CSR demonstrated superior discriminative performance (AUC 0.822, 95% CI: 0.718-0.926, p < 0.001) with an optimal cutoff of ≤ 0.895 (sensitivity 83.3%, specificity 80.6%), significantly exceeding literature standards (≤ 0.71). SII exhibited suboptimal sensitivity (63.9%) at cutoff ≥ 15.1% due to a 13.9% prevalence of non-suppressing adenomas. Paraspinal muscle signal intensity emerged as an ancillary discriminator (AUC 0.729, cutoff ≤ 671.5 AU, p < 0.001), with significantly lower values in adenomas (452.5 vs. 768.0 AU). Interobserver agreement was excellent for all parameters (ICC > 0.84). Notably, 19.4% of non-adenomas (predominantly pheochromocytomas and myelolipomas) exhibited false-positive CSR patterns. CONCLUSION: In surgical referral populations enriched with lipid-poor adenomas and confounding pathologies, higher CSR cutoffs (≤ 0.895) are required. The modest SII sensitivity highlights the challenge of non-suppressing adenomas, while paraspinal muscle signal intensity provides valuable ancillary discrimination. A multiparametric, population-aware diagnostic algorithm integrating CSR, SII, and PM-IP & OP SI is recommended to optimize adrenal mass characterization and reduce unnecessary interventions.
Diagnostic reliability of O-RADS score based on non-dynamic contrast-enhanced MRI and apparent diffusion coefficient in characterization of adnexal masses.
Elged BA, Abdelzaher DG, Elraouf GHA … +3 more , Elmokadem AH, Hassan A, Saleh GA
BMC Med Imaging · 2026 Jun · PMID 42337503 · Full text
OBJECTIVE: To evaluate the diagnostic performance and inter-rater reliability of (O-RADS) MRI based on non-dynamic contrast-enhanced (DCE) MRI and the value of apparent diffusion coefficient (ADC) in the characterization...
OBJECTIVE: To evaluate the diagnostic performance and inter-rater reliability of (O-RADS) MRI based on non-dynamic contrast-enhanced (DCE) MRI and the value of apparent diffusion coefficient (ADC) in the characterization of adnexal masses. METHODS: Retrospective analysis was done for 148 patients with 191 adnexal masses who underwent non-DCE MRI and diffusion-weighted imaging (DWI). Two independent radiologists classified the masses into five categories according to O-RADS-MRI and measured ADCmean values for solid and cystic components. The final diagnoses were determined by postoperative histopathology. Logistic regression analysis was performed to detect predictors of malignancy. RESULTS: There was excellent inter-rater agreement in assessing O-RADS categories (k=0.901 and 95% CI = 0.863 to 0.939). Solid tissue enhancement more than the myometrium at 30-40 s at non-DCE MRI was statistically significantly higher in malignant lesions (p < 0.001). There was also excellent reliability (absolute inter-rater agreement) in measuring ADCmean values. The ADCmean of solid and cystic components at cutoff values of ≤ 1.3 × 10 - 3 mm2/s and > 2.07 × 10 -3 mm²/s are perfect and excellent discriminators for differentiating malignant and benign adnexal masses (AUC = 1.000 and 0.952) respectively. Multivariate regression analysis revealed that the presence of simple fluid, low DW signal of cystic component, and ADCmean of cystic component > 2.07 × 10 -3 mm²/s were statistically significant independent predictors of malignancy in an O-RADS-MRI score > 3 adnexal lesions. CONCLUSIONS: O-RADS-MRI score utilizing non-DCE MRI is a reliable system with high PPV. ADCmean value is a non-invasive excellent discriminator for differentiating adnexal lesions that could improve O-RADS-MRI score performance and treatment plan.
Advancing precision risk stratification in adult diffuse gliomas through DSC-MRI-based habitat analysis.
Liang W, Zhou J, Huang Y … +3 more , Zou P, Xu D, Xu H
BMC Med Imaging · 2026 Jun · PMID 42337463 · Full text
OBJECTIVES: Diffuse gliomas exhibit substantial molecular and spatial heterogeneity. This study aimed to evaluate the ability of habitat-based radiomics models derived from dynamic susceptibility contrast MRI (DSC-MRI) a...
OBJECTIVES: Diffuse gliomas exhibit substantial molecular and spatial heterogeneity. This study aimed to evaluate the ability of habitat-based radiomics models derived from dynamic susceptibility contrast MRI (DSC-MRI) and conventional MRI to identify aggressive diffuse glioma phenotypes associated with integrated histologic-molecular risk. METHODS: This retrospective study included 197 adult patients with histopathologically confirmed diffuse gliomas. Multiparametric MRI data were preprocessed and segmented into tumor and peritumoral edema. K-means clustering was used to identify imaging-defined habitats reflecting spatial hemodynamic heterogeneity. A total of 855 radiomic features were extracted from each habitat and reduced through a sequential selection process involving univariate statistical testing, correlation filtering, recursive feature elimination, and LASSO regression. Random forest classifiers were developed to predict high-risk molecular subtypes, including IDH wildtype and other aggressive genetic alterations, and validated in internal held-out testing cohort. RESULTS: Habitat-based models significantly outperformed whole-region analyses (AUC 0.949 (0.858, 0.989) vs. 0.931(0.833, 0.980), p = 0.013). Crucially, models derived from hemodynamic features (CBF + MTT) demonstrated superior predictive accuracy compared to conventional anatomical sequences (T1C+T2FLAIR) in both tumor (AUC 0.944 (0.852, 0.987) vs. 0.895 (0.787, 0.960)) and edema habitats (AUC 0.932 (0.835, 0.981) vs. 0.819(0.698, 0.907)). The optimal model relied solely on hemodynamic features from combined habitats (AUC 0.949 (0.858, 0.989)). Multimodal fusion failed to improve performance, suggesting that hemodynamic parameters may provide the most discriminative imaging information. CONCLUSION: DSC-MRI-based habitat analysis provides significant value over conventional imaging by resolving perfusion heterogeneity. These findings highlight that hemodynamic features serve as a promising tool for prediction of integrated histologic-molecular risk status of adult diffuse gliomas, potentially serving as a promising imaging biomarker for preoperative assessment.
Diagnostic accuracy of fetal cardiac magnetic resonance imaging in cardiac and thoracic vascular anomalies: a meta-analysis.
Bhargavi C, Shenoy RD, Rassaf A … +5 more , Jonouchi D, Anunwa E, Rao SS, Kamath N, Acuna JM
BMC Med Imaging · 2026 Jun · PMID 42337462 · Full text
BACKGROUND: Fetal echocardiography (FE) is the primary modality for screening cardiac anomalies. Diagnostic limitations persist in complex lesions, thoracic vascular anomalies, late gestation, and cases with poor acousti...
BACKGROUND: Fetal echocardiography (FE) is the primary modality for screening cardiac anomalies. Diagnostic limitations persist in complex lesions, thoracic vascular anomalies, late gestation, and cases with poor acoustic windows. Fetal cardiac magnetic resonance imaging (FCMRI) complements FE by providing additional anatomical details of cardiac and thoracic vascular anomalies, although its diagnostic test accuracy has not been systematically evaluated. This meta-analysis aimed to determine the diagnostic accuracy of FCMRI in detecting cardiac and thoracic vascular anomalies and to evaluate its potential role in prenatal decision-making. METHODS: Observational studies evaluating FCMRI for cardiac and thoracic vascular anomalies were analyzed at the lesion-description level. Summary statistics were estimated using a bivariate random-effects and hierarchical summary receiver operating characteristic models. Subgroup analyses were performed across anatomical lesion categories, image acquisition methods, and acquisition eras. RESULTS: Seven studies involving 449 cardiac and thoracic vascular anomalies were analyzed. FCMRI demonstrated a sensitivity of 0.80 (95% CI: 0.66-0.89) and specificity of 0.98 (95% CI: 0.92-0.99), indicating acceptable accuracy, with a 95% prediction ellipse area of 0.29, suggesting moderate between-study heterogeneity. Pooled sensitivity [0.88 (95%CI: 0.79-0.93)] was highest for complex heart defects, with a tightly clustered prediction ellipse (0.04). Thoracic vascular anomalies showed lower sensitivity [0.75 (95%CI: 0.56-0.88)] and greater heterogeneity (0.39), reflecting diagnostic challenges. Only two studies showed incremental clinical utility of FCMRI in decision-making that varied by specific cardiac and thoracic vascular anomaly. The analysis was limited by the small number of studies, methodological heterogeneity, and the underrepresentation of true-negative cases. CONCLUSIONS: Fetal CMRI can provide clinically useful anatomical information complementary to FE, particularly in late gestation and selected high-risk pregnancies. Although it demonstrated high specificity and moderate sensitivity overall, diagnostic performance varied across lesion subtypes, with aortic and aberrant vascular anomalies remaining the most challenging to diagnose prenatally. META ANALYSIS REGISTRATION: PROSPERO: CRD420251133115.
Shear wave elastography and ultrasonic multi-parameter score for the identification of moderate-to-severe chronic kidney disease.
Guo L, Ran L, Wu Q … +5 more , Zhang Y, Gao Y, Zhou X, Li J, Li J
BMC Med Imaging · 2026 Jun · PMID 42332634 · Full text
OBJECTIVE: This study aimed to investigate the value of shear wave elastography (SWE) and an ultrasonic multi-parameter score in identifying chronic kidney disease (CKD) stage ≥ III (moderate-to-severe CKD) in patients w...
OBJECTIVE: This study aimed to investigate the value of shear wave elastography (SWE) and an ultrasonic multi-parameter score in identifying chronic kidney disease (CKD) stage ≥ III (moderate-to-severe CKD) in patients with known CKD. METHODS: A single-centre cross-sectional study was conducted, enrolling 152 patients with CKD (staged I-V according to estimated glomerular filtration rate). Conventional ultrasound parameters (renal length, parenchymal thickness, echogenicity) and SWE measurements (Young's modulus [YM] in kPa) of the renal cortex and medulla were obtained. The ultrasonic multi-parameter score was calculated using a standardised rubric. Receiver operating characteristic curve analysis was performed to evaluate the diagnostic performance of SWE cortical YM and the multi-parameter score for identifying CKD stage ≥ III. RESULTS: The study included 23 patients with stage I CKD, 45 with stage II, 44 with stage III, 15 with stage IV and 25 with stage V. Shear wave elastography cortical YM (area under the curve [AUC]: 0.894) and the multi-parameter score (AUC: 0.870) showed promising discrimination for stage ≥ III CKD. Cortical YM demonstrated a specificity of 0.912 and a sensitivity of 0.671. CONCLUSION: Shear wave elastography cortical YM has high specificity, supporting its potential role in the rule-in of moderate-to-severe CKD. The ultrasonic multi-parameter score also shows clinical value for severity stratification. However, the relatively low sensitivity limits its utility for rule-out purposes. CLINICAL TRIAL NUMBER: Not applicable.
Characteristics of pulmonary perfusion and ventilation in healthy adults: a prospective observational study with phase-resolved functional lung magnetic resonance imaging.
Liu A, Ni Y, Wang J … +15 more , Xi L, Yang H, Wang H, Du J, Zhang L, Dai J, Huang K, Ren Y, Wang S, Xia J, An J, Grimm R, Voskrebenzev A, Vogel-Claussen J, Liu M
BMC Med Imaging · 2026 Jun · PMID 42332603 · Full text
BACKGROUND: Phase-resolved functional lung magnetic resonance imaging (PREFUL-MRI) enables simultaneous, free-breathing, radiation-free assessment of regional lung perfusion and ventilation. This study aimed to provide p...
BACKGROUND: Phase-resolved functional lung magnetic resonance imaging (PREFUL-MRI) enables simultaneous, free-breathing, radiation-free assessment of regional lung perfusion and ventilation. This study aimed to provide preliminary reference data for lung perfusion and ventilation using PREFUL-MRI in healthy adults, and to characterize the physiological dependencies of these metrics on two breathing patterns, sex, and age. METHODS: In this prospective observational study, 87 healthy adults underwent PREFUL-MRI at 1.5 T during both normal and deep-slow breathing. Perfusion- and ventilation-related metrics were quantified via an automated pipeline. Paired comparisons between breathing states were performed using Wilcoxon signed-rank tests; unpaired comparisons between sexes and age groups (< 45 vs. ≥45 years) used Mann-Whitney U tests, with Holm-Bonferroni and Benjamini-Hochberg false-discovery-rate corrections applied to control for multiple comparisons. RESULTS: Mean perfusion (7.7% vs. 6.0%, Holm-Bonferroni adjusted p < 0.001) and ventilation defects (8.6% vs. 5.1%, Holm-Bonferroni adjusted p = 0.010) were decreased, and mean ventilation (15.8% vs. 48.3%, Holm-Bonferroni adjusted p < 0.001) and perfusion defects (1.9% vs. 7.9%, Holm-Bonferroni adjusted p = 0.005) increased during deep breathing compared with normal breathing. Twenty-eight participants had increased lung perfusion while 59 had reduced perfusion during deep breathing relative to normal breathing. During normal breathing, men exhibited higher mean ventilation than women (20.2% vs. 14.2%, Holm-Bonferroni adjusted p = 0.018). During deep breathing, men demonstrated higher total perfusion defect percentage and matched ventilation-perfusion defects than women (FDR adjusted q < 0.05). Total perfusion defect percentage was lower in participants aged ≥ 45 years than in those aged < 45 years (1.8% vs. 2.7%, FDR adjusted q = 0.036). Mean flow-volume loop correlations were similar between breathing patterns, sexes, and age groups after multiple comparison correction (p > 0.05). CONCLUSIONS: PREFUL-MRI can captures physiological variations related to breathing pattern, sex, and age in healthy adults. These findings provide a framework for distinguishing normal physiological heterogeneity from pathological change in clinical PREFUL-MRI interpretation. CLINICAL TRIAL NUMBER: Not applicable.
Construction of a small-sample brain imaging data augmentation and explainable diagnostic model for autism based on generative adversarial networks.
Xu W
BMC Med Imaging · 2026 Jun · PMID 42332578 · Full text
BACKGROUND: Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized primarily by social communication deficits and repetitive stereotyped behaviors. Its objective diagnosis has long relied on clinic...
BACKGROUND: Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized primarily by social communication deficits and repetitive stereotyped behaviors. Its objective diagnosis has long relied on clinical scale assessments, lacking automated tools based on brain imaging. METHOD: This study proposes an ASD auxiliary diagnostic framework integrating conditional generative adversarial network (conditional GAN, cGAN) data augmentation, multimodal feature fusion, and explainable deep learning, based on the ABIDE I/II multi-center public datasets. First, functional connectivity matrices of AAL-116 brain regions were extracted from resting-state functional magnetic resonance imaging (rs-fMRI), and cortical morphological features were derived from structural magnetic resonance imaging (sMRI). Multi-site scanning biases were corrected using the ComBat method. On this basis, minority class samples were augmented using class-conditional GAN, followed by multimodal information fusion via a dual-branch encoder and cross-attention mechanism, ultimately outputting classification decisions between ASD and typically developing (TD) subjects. RESULTS: Experimental results demonstrate that, under stratified five-fold cross-validation, the proposed method achieved an AUC of 0.871 ± 0.016 and a balanced accuracy of 0.797 ± 0.012 on the full multimodal sample set, representing improvements of 13.2% and 9.2% over single sMRI and rs-fMRI modalities, respectively. Leave-one-site-out (LOSO) cross-validation yielded an average AUC of 0.783 ± 0.041, validating the model's cross-center generalization capability. CONCLUSION: Explainability analysis based on Integrated Gradients revealed that the default mode network and social brain regions are key decision bases for distinguishing ASD from TD, highly consistent with existing neurobiological evidence. CLINICAL TRIAL: Not applicable.
FD-MSP: feature decoupling network with multi-scale prototypes for domain-adaptive polyp segmentation.
Zhong W, Yang X, Liu S … +2 more , Wang Y, Im SK
BMC Med Imaging · 2026 Jun · PMID 42321673 · Full text
Polyp segmentation in colonoscopy images plays a critical role in the early detection and treatment of colorectal cancer. Although deep learning-based segmentation models have achieved remarkable success, their performan...
Polyp segmentation in colonoscopy images plays a critical role in the early detection and treatment of colorectal cancer. Although deep learning-based segmentation models have achieved remarkable success, their performance is often limited by variable imaging protocols or unseen datasets from different endoscopic devices. To address this challenge, unsupervised domain adaptation (UDA) methods have emerged as a promising solution, leveraging labeled source data and unlabeled target data to bridge the domain gap. However, existing UDA models suffer from two key limitations. First, they entangle domain-invariant and domain-specific features, making it difficult to identify which representations should transfer across domains. Second, they rely on single-scale features that fail to capture the extreme size variations observed in polyps, which range from tiny flat lesions to large protruding masses. As such, we propose FD-MSP, a feature decoupling network with multi-scale prototypes for cross-domain polyp segmentation. The core contribution is a structural redesign of the UDA pipeline that explicitly decouples feature distribution alignment from multi-scale semantic prototype learning, rather than entangling them in a single objective. This redesign is realized through four tightly-coupled components: (i) a dual-stream shallow-encoder decoupling module that separates domain-invariant polyp signatures from equipment-dependent variations through the joint effect of adversarial alignment on both low-level features and output predictions and a pixel-wise orthogonality regularizer between source-fed and target-fed shallow features; (ii) an online pseudo-fusion adapter that blends offline soft labels with current full-image predictions under class-wise confidence gating; (iii) a multi-scale grouped prototype layer (three groups of seven prototypes) that captures polyp patterns at multiple granularities through dilated convolutions with progressively expanding receptive fields; and (iv) a dual-branch gated head that adaptively fuses convolutional and prototype predictions at inference. Experimental results on three public colonoscopy datasets showed that FD-MSP offers consistent improvements over existing UDA methods for polyp segmentation performance in the target domain.
Diagnostic value of ultrasonographic measurement of peritoneal, mesenteric and intestinal wall thickness in children with peritoneal dialysis-related peritonitis: a preliminary clinical study.
Dong L, Lin X, Zhu S … +3 more , Liu Y, Zhang Y, Luo H
BMC Med Imaging · 2026 Jun · PMID 42316080 · Full text
BACKGROUND: Peritonitis is a common serious complication of peritoneal dialysis. Although peritoneal thickening is commonly observed in long-term peritoneal dialysis patients, there have been relatively few studies on me...
BACKGROUND: Peritonitis is a common serious complication of peritoneal dialysis. Although peritoneal thickening is commonly observed in long-term peritoneal dialysis patients, there have been relatively few studies on mesenteric and intestinal wall alterations through ultrasonography, particularly in pediatric patients. We aim to determine the value of ultrasonographic measurement in children with peritoneal dialysis-related peritonitis. METHODS: We recruited two groups of pediatric patients, of whom 40 had peritonitis (study group) and 42 had no peritonitis (control group). We measured peritoneal, mesenteric and intestinal wall thickness by trans-abdominal ultrasonography. To identify risk factors and develop a predictive nomogram for peritonitis, we employed multivariate logistic regression analysis. Subsequently, the performance of the nomogram model was assessed through the ROC curve, calibration curve, and decision curve analysis (DCA). RESULTS: The ultrasonographic measurements of the peritoneal thickness, mesenteric thickness, jejunal wall and ileal wall thickness in peritonitis group were significantly thicker than the non-peritonitis group. The peritonitis group had significantly higher NLR (neutrophil-to-lymphocyte ratio) and significantly lower albumin and uric acid levels than the non-peritonitis group. Multivariate Logistic analysis showed that dialysis duration, mesenteric thickness, jejunal wall thickness and NLR were potential associated factors. ROC curve showed that the integrated predictive model exhibited superior diagnostic accuracy (AUC = 0.873) over that of any single predictive parameter (P < 0.05). The nomogram model showed good calibration and clinical benefit. CONCLUSIONS: High-frequency ultrasound effectively supports the diagnosis of peritoneal dialysis-related peritonitis. The integrated predictive model based on ultrasonographic measurements is effective for predicting the recurrent PD-related peritonitis in children and offering practical ultrasonographic criteria.
Distinguishing Ki67 stratification expression in ER-positive/HER2-negative breast cancers: comparison of advanced MRI diffusion models.
Kang Z, Du S, Zhao L … +8 more , Zhao R, Gao S, Zhang Y, Yu J, Liu X, Jiang Y, Wang Y, Zhang L
BMC Med Imaging · 2026 Jun · PMID 42316062 · Full text
PURPOSE: Identifying the stratification expression of Ki67 is crucial for directing clinical treatment strategies in ER+/HER2- breast cancers. This diagnostic study investigated the value of first-order features extracte...
PURPOSE: Identifying the stratification expression of Ki67 is crucial for directing clinical treatment strategies in ER+/HER2- breast cancers. This diagnostic study investigated the value of first-order features extracted from conventional DWI, diffusion kurtosis imaging (DKI), fractional order calculus (FROC), and continuous-time random walk (CTRW) in discriminating Ki67 expression of ER+/HER2- invasive ductal breast cancer. MATERIALS AND METHODS: This retrospective study included 121 patients who underwent DWI, DKI, FROC and CTRW and were pathologically categorized into the low (≤ 5%, n = 30), medium (> 5% to < 30%, n = 41), and high Ki67 expression group (≥ 30%, n = 50). Sixty-three diffusion parameters were computed and subsequently compared across different groups. The area under the receiver operating characteristic (ROC) curve (AUC) was used to quantify diagnostic efficacy. Multivariate logistic regression and bootstrap (1,000 samples) analyses were used to establish and evaluate, respectively, the optimal model to identify Ki67 expression. RESULTS: Twenty-three features showed statistically significant differences among the low, medium and high expression groups (all p values < 0.05). Further multivariable logistic regression analysis for discriminating the low Ki67 and non-low expression group showed that the FROC model constructed by D, µ and µ had optimal diagnostic efficacy (AUC = 0.858; 95% confidence interval, 0.783-0.915), which was significantly better than that of DWI model (AUC = 0.677; p = 0.008). The validation model showed good accuracy (AUC = 0.850; 95%CI, 0.774-0.916). CONCLUSIONS: The FROC model could help identify the low Ki67 expression (≤ 5%) and prevent unnecessary chemotherapy.
Differentiating sub-centimeter lung metastases in colorectal cancer by deep learning: a multicenter retrospective study.
Gao X, Ma X, Zhang Z … +6 more , Yuan J, Li Q, Zheng P, Lv M, Ma D, Sun J
BMC Med Imaging · 2026 Jun · PMID 42316047 · Full text
BACKGROUND: Early diagnosis of sub-centimeter lung metastases is critical for timely decision-making and improved prognosis in patients with colorectal cancer. The diagnostic evaluation of indeterminate sub-centimeter lu...
BACKGROUND: Early diagnosis of sub-centimeter lung metastases is critical for timely decision-making and improved prognosis in patients with colorectal cancer. The diagnostic evaluation of indeterminate sub-centimeter lung nodules in colorectal cancer patients remains a crucial challenge. We aim to develop and validate a deep learning model for differentiating sub-centimeter lung metastases noninvasively. METHODS: Our retrospective study included 1335 colorectal cancer patients with pretreated sub-centimeter lung metastases from 3 centers and 1335 benign lung nodules from one center. The primary cohort comprised 1194 patients, who were randomly assigned (8:2) to training and internal validation cohorts. Two external validation cohorts (EVC) consisted of 101 (EVC1) and 40 (EVC2) patients. The deep learning framework employed a fully automated VNet-based segmentation model for non-contrast computed tomography (CT) scans, integrated with a ResNet18 classifier for discriminating whether sub-centimeter lung nodules are benign or metastatic. Moreover, stepwise validation of subgroups was performed according to the maximum diameter of the nodules (10, 9, 8, 7, 6, 5, ≤ 4 mm). RESULTS: The automatic segmentation model achieved a dice coefficient of 0.825. In primary cohort, the accuracy by radiologists was 0.705. The AUC in deep learning model showed 0.953 (95%CI: 0.937-0.967), 0.906 (95%CI: 0.874-0.926), and 0.951 (95%CI: 0.938-0.965) in internal and two external validation cohorts. Furthermore, the stepwise validation demonstrated that the diagnostic performance remain stable as the nodule maximum diameter shrinks when size≥5 mm. CONCLUSION: The deep learning model can accurately and non-invasively distinguish between benign and metastatic sub-centimeter lung nodules in patients with colorectal cancer.
Super-resolution of periapical and bitewing digital radiographs using convolutional neural network.
Pourahmadiyan-Nadiki H, Alemohammad N, Mohammadi-Bassir M … +1 more , Hamze F
BMC Med Imaging · 2026 Jun · PMID 42316037 · Full text
BACKGROUND AND AIM: Due to the insufficient data and the ongoing development of machine learning (ML), this study was conducted to examine a deep learning approach for enhancing the resolution of dental Bite-wing (BW) an...
BACKGROUND AND AIM: Due to the insufficient data and the ongoing development of machine learning (ML), this study was conducted to examine a deep learning approach for enhancing the resolution of dental Bite-wing (BW) and Peri-Apical (PA) Radiographs (Rg) based on Super Resolution (SR) theory. METHODS AND MATERIALS: 1000 BW, and PA Rg were collected: 750 images for training while 250 images for test. At first step, we downscaled all High Resolution (HR) images to create Low Resolution (LR) ones using 4*4 average pooling without overlap. Thereafter, we incorporated three deep learning-based super-resolution approaches and the most efficient one (down sampled skip-connection/ Multi-scale (DSC/MS)) was chosen. After training, our ML algorithm was tested by the 250 LR images incorporated six evaluation metrics. RESULTS: After five-time repletion of our model, the mean ± S.D of R, RSME, MSE, MAE, SSIM, and PSNR was 0.90 ± 0.0006, 0.039 ± 0.001, 0.0017 ± 0.00015, 0.026 ± 0.001, 0.85 ± 0.003, 28.45 ± 0.30. All these metrics was superior comparing to conventional methods. CONCLUSION: Our SR model demonstrated significant effectiveness and the DSC/MS showed noticeably superior results comparing to linear, cubic, or nearest neighbor interpolations.
A transformer-based multimodal deep learning model for preoperative prediction of Ki-67 expression level in glioma.
Liu L, Hu F, Lai M … +4 more , Zhao X, Lei P, Yao Z, Fan B
BMC Med Imaging · 2026 Jun · PMID 42310560 · Full text
PURPOSE: The aim of this study is to develop a deep learning model using preoperative multimodal MR data to predict the Ki-67 expression level of glioma and externally validate the predictive performance of the model. ME...
PURPOSE: The aim of this study is to develop a deep learning model using preoperative multimodal MR data to predict the Ki-67 expression level of glioma and externally validate the predictive performance of the model. METHODS: This study retrospectively collected the clinical and imaging data from 421 patients with grade 2-4 gliomas who underwent surgical resection or biopsy and were pathologically diagnosed in two hospitals between January 2020 and December 2024. The 421 patients were divided into a training set (N = 217), an internal validation set (N = 94), and an external validation set (N = 110). Then, the tumor margins were delineated on contrast-enhanced T1- weighted imaging (CE-T1WI) and contrast-enhanced T2-fluid-attenuated inversion recovery (CE-T2FLAIR) to obtain the three-dimensional region of interest (3D ROI) of the tumor. Three vision transformer (ViT) models based on CE-T1WI, CE-T2FLAIR, and CE-T1WI + CE-T2FLAIR (CE-T1WI_ViT, CE-T2FLAIR_ViT and Combined_ViT) were constructed respectively. The predictive performance of the models was evaluated by the area under the receiver operating characteristic curve (AUC). Finally, to further assess the predictive performance of our transformer model, we trained and tested three convolutional neural network (CNN) models (ShuffleNet, ResNet50, and DenseNet121) on the same dataset and compared our trained Combined_ViT model with these three CNN models. RESULTS: The three models, including CE-T1WI_ViT, CE-T2FLAIR_ViT, and Combined_ViT, demonstrated high predictive accuracy for Ki-67 level in grade 2-4 gliomas, with AUC values of 0.859 (95% confidence interval [CI], 0.792-0.926), 0.825 (95%CI, 0.745-0.906), and 0.922 (95%CI, 0.868-0.976), respectively. Among the three models, the Combined_ViT model achieved the highest predictive accuracy. Furthermore, the predictive performance of the Combined_ViT model exceeded that of the three CNN models (ShuffleNet, ResNet50, and DenseNet121), with AUC values of 0.897 (95%CI, 0.839-0.954), 0.905 (95%CI, 0.843-0.966), and 0.913 (95%CI, 0.861-0.965) respectively. CONCLUSIONS: The deep learning models based on ViT can effectively predict the Ki-67 expression level of glioma, and are a feasible alternative to CNN models. CLINICAL TRIAL NUMBER: Not applicable.
Grading knee cartilage degeneration and meniscal injury with combined ACS-accelerated T1ρ and T2 mapping.
Yang S, Liu Z, Huang S … +5 more , Cheng L, Nie Z, Yu Q, Lei Z, Fan W
BMC Med Imaging · 2026 Jun · PMID 42310557 · Full text
OBJECTIVE: To evaluate the diagnostic performance of artificial intelligence-compressed sensing (ACS)-accelerated T1ρ mapping combined with conventional T2 mapping for quantitative differentiation of articular cartilage...
OBJECTIVE: To evaluate the diagnostic performance of artificial intelligence-compressed sensing (ACS)-accelerated T1ρ mapping combined with conventional T2 mapping for quantitative differentiation of articular cartilage degeneration and meniscal injury severity in knee osteoarthritis (KOA). METHODS: We prospectively enrolled 56 KOA patients and 30 age-matched healthy volunteers between August 2025 and February 2026. All underwent 3T knee MRI using ACS T1ρ and conventional T2 mapping. Relaxation times were quantified in predefined cartilage and meniscal regions. Cartilage degeneration was classified by Kellgren-Lawrence (K-L) grade: 0 (normal), I-II (mild), and III-IV (severe). Meniscal injury was graded using the Stoller system: 0 (normal), I-II (mild), and III (severe). Group and subgroup comparisons used appropriate statistical tests. RESULTS: T1ρ and T2 values were significantly elevated in KOA versus controls across all regions (P < 0.05). Within KOA, relaxation times differed significantly between mild and severe subgroups (P < 0.05). For cartilage degeneration, AUCs were 0.948 (95% CI: 0.889-1.000) for ACS T1ρ, 0.867 (95% CI: 0.760-0.975) for T2, and 0.964 (95% CI: 0.922-1.000) for the combined model. For meniscal injury, AUCs were 0.851 (95% CI: 0.779-0.922) for ACS T1ρ, 0.802 (95% CI: 0.720-0.885) for T2, and 0.855 (95% CI: 0.784-0.926) for the combined approach. CONCLUSION: Integrating ACS T1ρ and T2 mapping enables precise, quantitative stratification of cartilage and meniscal pathology across KOA severity levels, offering comprehensive biochemical assessment prior to morphological changes and highlighting its potential for early intervention and stage-based management.
About
- Frequency
- Sun
- Papers found
- 200
- RSS feed
- Subscribe