BMC Medical Imaging[JOURNAL]
Sun
200 papers
Multiparametric MRI quantitative metrics for grading and staging graves' ophthalmopathy.
Wu S, Yang L, Cai Y … +4 more , Liang J, Wen J, Huang H, Yu S
BMC Med Imaging · 2026 May · PMID 42106660 · Full text
OBJECTIVE: To explore the application of multiparametric magnetic resonance imaging (MRI) combined with clinical parameters in the assessment of Graves' ophthalmopathy (GO). METHODS: This retrospective study included 71...
OBJECTIVE: To explore the application of multiparametric magnetic resonance imaging (MRI) combined with clinical parameters in the assessment of Graves' ophthalmopathy (GO). METHODS: This retrospective study included 71 patients with GO (142 affected eyes) categorized into inactive (75) vs. active (67) and mild (70) vs. moderate-severe (72) subgroups. Clinical parameters were collected, and extraocular muscle (EOM) parameters [signal intensity ratio (SIR), T1 relaxation times (T1RT), T2 relaxation times (T2RT), and fat fraction (FF)] were measured (max/mean values). Predicted probabilities were derived from the generalized linear mixed model. Multivariate Logistic regression (based on the Akaike Information Criterion) was used to develop combined diagnostic models, and the marginal area under the receiver operating characteristic (ROC) curve (AUC) was employed to assess their efficacy. RESULTS: Compared with the inactive and mild groups, the active and moderate-severe GO eyes showed higher mean EOM-SIR, mean EOM-T1RT, and mean/max EOM-T2RT, as well as lower mean/max EOM-FF and female proportion (all P < 0.05). The active group had a shorter disease duration, whereas the moderate-severe group had higher thyrotropin receptor antibody (TRAb) levels (P < 0.05). For differentiating GO activity and severity, the combined staging Model 3 (EOM-FFmax + EOM-T1RTmean + EOM-T2RTmean + sex + thyroid functional status) and grading Model 6 (EOM-FFmax + EOM-T1RTmean + EOM-T2RTmean + sex + TRAb + thyroid functional status) yielded the best diagnostic efficacy, with AUCs of 0.912 and 0.947, respectively. CONCLUSION: The combined model of quantitative multiparametric MRI and clinical parameters enables a more accurate assessment of GO activity and severity.
Application of multimodal ultrasound radiomics in the diagnosis of superficial lymph node tuberculosis.
Su D, Chen P, Zhang Y … +2 more , Wang Y, Yang G
BMC Med Imaging · 2026 May · PMID 42106657 · Full text
OBJECTIVE: To develop preoperative diagnostic models for superficial lymph node tuberculosis (LNTB) using radiomic features extracted from multimodal ultrasound imaging, including gray-scale ultrasound (US), ultrasound e...
OBJECTIVE: To develop preoperative diagnostic models for superficial lymph node tuberculosis (LNTB) using radiomic features extracted from multimodal ultrasound imaging, including gray-scale ultrasound (US), ultrasound elastography (UE), and contrast-enhanced ultrasound (CEUS), in conjunction with various machine learning algorithms. METHODS: A retrospective study was conducted on 222 patients with lymphadenopathy. The patients were randomly divided into a training group (n = 156) and a validation group (n = 66) in a 7:3 ratio. 837 radiomics features were extracted from images of each modality (US, UE and CEUS). After initial screening by hypothesis testing, the least absolute shrinkage and selection operator (LASSO) regression with five-fold cross-validation was used for feature dimensionality reduction and selection. After feature selection, five machine learning models-logistic regression, decision tree, random forest, support vector machine, and AdaBoost-were used to construct radiomic-based models. Receiver operating characteristic (ROC) curves were generated, and the area under the curve (AUC) was computed to assess the performance of each model in predicting superficial LNTB. Clinical decision curve (DCA) is used to measure the net benefits under various probability thresholds. The diagnostic performance of ultrasound physicians was also compared with that of the best-performing machine learning model. RESULTS: Among the models generated by different algorithms, the decision tree model exhibited the best performance, achieving an AUC of 0.909 (95% CI, 0.789-0.949) in the training cohort and 0.866 (95% CI, 0.774-0.958) in the validation cohort. The AUC, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for ultrasound physicians were 0.693 (95% CI, 0.568-0.818), 0.698, 0.689, 0.722, and 0.664, respectively. The DCA shows that the decision tree model has the best net income in the range of clinical relevance threshold of 0.6-0.8. Delong test showed that decision tree model was superior to ultrasonic doctor's diagnosis (Z = 2.98, p<0.0029). CONCLUSION: The radiomic model constructed from US, UE, and CEUS demonstrated robust diagnostic performance for superficial LNTB, with the decision tree model yielding the best results.
Recovery of thalamic damage after decompression by H-MRS in combination with Diffusion Tensor Imaging(DTI) in patients with cervical spondylotic myelopathy.
Zheng J, Zhang Y, Zhao B … +3 more , Wang N, Gao T, Zhang L
BMC Med Imaging · 2026 May · PMID 42106652 · Full text
BACKGROUND: To access the changes of thalamic metabolites before and after surgery in patients with Cervical Spondylotic Myelopathy (CSM) using Hydrogen Proton Magnetic Resonance Spectroscopy (H-MRS) combined with Diffus...
BACKGROUND: To access the changes of thalamic metabolites before and after surgery in patients with Cervical Spondylotic Myelopathy (CSM) using Hydrogen Proton Magnetic Resonance Spectroscopy (H-MRS) combined with Diffusion Tensor Imaging (DTI) and to investigate its association with improvement in neurological function. METHODS: Forty-eight CSM patients who underwent cervical decompression surgery from December 2022 to June 2023 were included, and 33 healthy volunteers were recruited. All subjects underwent bilateral thalamic H-MRS and DTI scans before the surgical procedure, and subsequently again 6 months later. Neurological function was assessed pre-operatively and post-operatively (6 months) in all patients with CSM using the modified Japanese Orthopaedic Association (mJOA). The changes of mJOA (△mJOA = postoperative mJOA - preoperative mJOA) were employed as an indicator of neurological improvement. The pro- and postoprative N-acetylaspartate/creatine (NAA/Cr), choline/creatine (Cho/Cr), myo-inositol /creatine (mI/Cr), glutamate and glutamine complex/creatine (Glx/Cr) fractional anisotropy (FA), apparent diffusion coefficient (ADC) were statistically compared in CSM patients and healthy controls (HCs). A correlation analysis was conducted to determine the relationship between alterations in pre- and postoperative metabolite ratios (△NAA/Cr, △Cho/Cr, △mI/Cr, △Glx/Cr), △FA, △ADC and ΔmJOA. RESULTS: Results Compared to the HCs, patients with CSM showed significantly pre- and post-operative NAA/Cr (t = -4.988, P < 0.001; t = -3.562, P = 0.001), Cho/Cr (t = -5.946, P < 0.001; t = -2.764, P = 0.007), mI/Cr (t = -3.988, P < 0.001; t = -2.079, P = 0.041) and FA (t = -4.884, P < 0.001; t = -3.813, P < 0.001). There was no difference in Glx/Cr and ADC between patients in patients with CSM, either preoperatively or postoperatively, compared to HCs. Post-operative NAA/Cr (t = -2.805, P = 0.007), mI/Cr (t = -3.285, P = 0.003) and FA (t = -2.690, P = 0.007) were increased in CSM patients compared to pre-operative NAA/Cr and mI/Cr. In CSM patients, ΔmI/Cr correlated significantly with ΔmJOA (r = 0.4782, P < 0.001). CONCLUSION: The preliminary findings indicate that metabolites in the thalamus of CSM patients exhibit changes following surgery. Additionally, it has been demonstrated that elevated post operiative mI correlates with improvements in neurological function.
A study on the correlation between cardiac magnetic resonance characteristics and myocardial fibrosis in overweight and obese individuals.
Liu L, Yao Y, Yu H … +1 more , Zhai J
BMC Med Imaging · 2026 May · PMID 42106645 · Full text
BACKGROUND: Overweight and obesity are major risk factors for adverse cardiovascular outcomes. Cardiac magnetic resonance (CMR) can detect subclinical cardiac dysfunction through myocardial strain and quantify epicardial...
BACKGROUND: Overweight and obesity are major risk factors for adverse cardiovascular outcomes. Cardiac magnetic resonance (CMR) can detect subclinical cardiac dysfunction through myocardial strain and quantify epicardial adipose tissue volume (EATV). This study aimed to investigate the differences in cardiac structure, function, myocardial strain, and EATV among individuals with normal weight, overweight, and obesity using CMR, and to evaluate their value for myocardial fibrosis (LGE positivity). METHODS: In this study, 192 subjects undergoing CMR were classified into control (n = 77), overweight (n = 37), and obesity (n = 78) groups based on body mass index (BMI). CMR-derived parameters included cardiac function, left ventricular global/segment strain (radial, circumferential, longitudinal), left atrial strain, EATV, and late gadolinium enhancement (LGE). Differences among groups were compared. Spearman correlation analyzed relationships between BMI, EATV, and strain parameters. Logistic regression and ROC curve analyses assessed factors independently associated with LGE positivity. RESULTS: Compared to the control group, the overweight and obesity groups showed significantly increased left ventricular volumes, mass, and EATV, but decreased left ventricular ejection fraction (LVEF), global longitudinal strain (GLS), and left atrial reservoir strain (εs), with the most pronounced changes in the obesity group. EATV was highly correlated with BMI (r = 0.72). Both BMI and EATV showed low-to-moderate inverse correlations with impaired myocardial strain parameters (|r| = 0.22-0.31). Multivariate logistic regression identified higher BMI (OR = 1.22, 95% CI:1.07-1.40), larger EATV (OR = 1.03, 95% CI:1.01-1.04), and lower left ventricular apical segment radial strain (A-PRS) (OR = 0.93, 95% CI:0.87-0.98) as factors independently associated with LGE positivity. CONCLUSIONS: Overweight and obesity are associated with adverse cardiac remodeling, subclinical systolic dysfunction, impaired myocardial strain, and increased EATV. BMI, EATV, and A-PRS are factors independently associated with myocardial fibrosis. These findings underscore the presence of early cardiac alterations in the overweight stage, highlighting the need for early cardiovascular risk assessment in this population.
The influence of fine-needle aspiration on dual-layer detector spectral CT parameters of papillary thyroid carcinoma: a propensity-match analysis.
Yu J, Lv L, Song Z … +4 more , Zhou B, Zhang X, Peng J, Zhang D
BMC Med Imaging · 2026 May · PMID 42098667 · Full text
BACKGROUND: Dual-layer detector spectral computed tomography (DLCT) and fine-needle aspiration (FNA) are commonly used in papillary thyroid carcinoma (PTC). However, whether FNA affects DLCT quantitative parameters remai...
BACKGROUND: Dual-layer detector spectral computed tomography (DLCT) and fine-needle aspiration (FNA) are commonly used in papillary thyroid carcinoma (PTC). However, whether FNA affects DLCT quantitative parameters remains uncertain. This study aimed to investigate the effect of FNA on DLCT parameters in PTC and to determine the optimal timing for DLCT examination. METHODS: This retrospective study included 689 patients with PTC, who were categorized into pre-FNA (n = 222) and post-FNA (n = 467) groups according to the sequence of FNA and DLCT. Propensity score matching (PSM) was applied to balance confounding variables. Sensitivity analyses were performed using inverse probability of treatment weighting and multivariate regression. DLCT parameters, including virtual monoenergetic images (VMIs) at 40, 70, and 100 keV; spectral attenuation curve slope (λ); iodine concentration (IC); normalized IC (NIC); effective atomic number (Z); and normalized Zeff (NZ), were compared between groups in arterial (AP) and venous phases (VP). Post-FNA time intervals were ranked and divided into deciles. For each DLCT parameter, standardized differences from the pre-FNA baseline were plotted across deciles to propose temporal cutoffs, which were validated by comparing post-FNA subgroups with the pre-FNA group. RESULTS: Covariates were well balanced after PSM, with all standardized mean differences below 0.25. Sensitivity analyses confirmed consistent direction of effect estimates across methods. Two clinically relevant exploratory temporal landmarks were observed around 6 and 18 days. Within 6 days after FNA, significant differences were observed in AP VMIs (40, 70, 100 keV), λHU, IC, NIC, and VP VMI at 100 keV. Between 7 and 18 days, only VP-IC remained significantly altered. Beyond 18 days, no parameters differed from pre-FNA baselines. CONCLUSIONS: DLCT parameters showed a dynamic post-FNA temporal pattern, with early and later candidate recovery windows centered around approximately 6 and 18 days. From a clinical perspective, DLCT may be considered before FNA or after about 6 days, whereas a later interval around 18 days may be preferable when minimizing biopsy-related perturbation is particularly important.
Utilizing imaging features of preoperative gadoxetic acid-enhanced MRI for predicting lymphovascular invasion in colorectal cancer liver metastases and exploring its impact on survival.
Xing Q, Cui Y, Gu XL … +2 more , Li XT, Sun YS
BMC Med Imaging · 2026 May · PMID 42098665 · Full text
PURPOSE: To construct an imaging model based on preoperative gadoxetic acid- enhanced MRI for predicting lymphovascular invasion (LVI) in colorectal cancer liver metastases (CRLM) and explore its impact on survival. METH...
PURPOSE: To construct an imaging model based on preoperative gadoxetic acid- enhanced MRI for predicting lymphovascular invasion (LVI) in colorectal cancer liver metastases (CRLM) and explore its impact on survival. METHOD: A total of 91 patients with CRLM were retrospectively enrolled in this study. The liver lesions were categorized into two groups, with LVI and without LVI by pathological examination. The long diameter, vascular penetration sign, peritumoral hepatobiliary phase (HBP) hypointensity and other qualitative signs were evaluated. The mean value and standard deviation (SD) value of signal intensity (SI) of the liver lesion and the region with 5-mm tumor expansion were recorded. The relative enhancement rate (RER) was calculated on each phase. Univariate and multivariate logistic regression were used to construct the imaging model (LVI ) for predicting LVI in CRLM. Each liver lesion in all the patients was assessed using the LVI to classify patients into predicted LVI-negative or predicted LVI-positive groups. RESULTS: The vascular penetration sign (odds ratio [OR] = 30.052, p<0.001) and SD of tumor on HBP (SD)(OR = 1.004, p = 0.026) were used to construct the imaging model for predicting LVI and the AUC of the model was 0.874 (95%CI:0.747-1.000). The liver recurrence-free survival (LRFS) of 22 predicted LVI-positive patients was significantly lower than that of 69 predicted LVI-negative patients (median 9.0 vs. 28.0 months, p = 0.028). The overall survival (OS) of predicted LVI-positive patients was significantly lower than that of predicted LVI-negative patients (median 21.0 vs. 63.0 months, p = 0.001). CONCLUSION: The imaging model based on vascular penetration sign and SD has good efficiency in predicting lymphovascular invasion and survival in patients with CRLM.
Integration of glymphatic system function and hippocampal radiomics for diagnosis and conversion prediction of Alzheimer's disease.
Mao X, Zhang D, Ying D … +3 more , Yu J, Ge Y, Jia Z
BMC Med Imaging · 2026 May · PMID 42098640 · Full text
BACKGROUND: Glymphatic system (GS) function and hippocampal microstructural changes are promising imaging markers of Alzheimer's disease (AD). This study aims to investigate the effectiveness of combining diffusion tenso...
BACKGROUND: Glymphatic system (GS) function and hippocampal microstructural changes are promising imaging markers of Alzheimer's disease (AD). This study aims to investigate the effectiveness of combining diffusion tensor image analysis along the perivascular space (DTI-ALPS) with hippocampal radiomics for diagnosing AD, and to develop an innovative multivariable model integrating hippocampal radiomics and clinical biomarkers for predicting mild cognitive impairment (MCI) progression. METHODS: We included three cohorts from two databases retrospectively, using an internal (n = 210) and an external dataset (n = 430) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The ALPS index was employed to measure GS function, and 3D-T1WI hippocampal radiomics features were extracted to construct machine learning models for classifying and diagnosing AD. Conversion of MCI to AD was assessed through integrating the hippocampal radiomics features, ALPS index, and AD-related clinical biomarkers. RESULTS: The ALPS index was lower in patients with AD than in healthy controls (HCs) in both the internal and external cohorts (p < 0.001). The combined hippocampal radiomics features and ALPS index model demonstrated good performance in AD classification. The multivariable prediction model of MCI progression to AD achieved an area under the curve of 0.97 and 0.92 for the training and testing cohorts, respectively. CONCLUSIONS: Integrated ALPS index and hippocampal-based radiomics features can improve diagnostic performance in patients with AD, showing predictive capability for identifying the MCI conversion.
Pediatric lung ground glass nodules: a real-world, large-scale CT cohort analysis.
Duan YN, Guo YF, Wen JZ … +3 more , Lin X, Zhu YQ, Qin J
BMC Med Imaging · 2026 May · PMID 42098619 · Full text
BACKGROUND: Increasing detection of pediatric ground-glass nodules (GGNs) presents a clinical dilemma lacking robust evidence and guidelines. We aimed to evaluate the short-term natural course of incidental pediatric GGN...
BACKGROUND: Increasing detection of pediatric ground-glass nodules (GGNs) presents a clinical dilemma lacking robust evidence and guidelines. We aimed to evaluate the short-term natural course of incidental pediatric GGNs through real-world observation. METHODS: This retrospective, single-center, real-world study screened children (0-18 years) undergoing low-dose chest CT between January 1, 2010, and December 15, 2025. Patients with GGNs were included, excluding those with malignancy, immune dysfunction, specific infections, mean diameter < 3 mm or > 30 mm, artificial intelligence recognition failure, or poor image quality. Baseline characteristics, clinical presentation, and CT imaging features were collected and analyzed, with subgroup analyses performed. For patients with follow-up CT, nodule evolution was assessed. RESULTS: Among 14,106 children, 901 (6.4%) had GGNs. After exclusions, 602 patients were included, with a median age of 15 (14, 17) years, 58.6% were male. From these patients, 602 most suspicious GGNs were analyzed, comprising 43 (7.1%) mixed GGNs and 559 (92.9%) pure GGNs. Mixed GGNs showed significantly larger size and higher attenuation than pure GGNs (P < 0.01). Children aged > 12 years had GGNs with larger volume and lower attenuation compared to younger children (P < 0.05). Among the follow-up subgroup (n = 78), with a median follow-up period of 268.5 days, 32 GGNs regressed, 45 remained stable, and only 1 increased in size (pathologically confirmed adenocarcinoma in situ). Smaller GGNs at baseline were more likely to regress (P < 0.05). CONCLUSIONS: GGNs are not uncommon in children on chest CT. In our cohort, most GGNs remained stable or regressed over short-term follow-up. These observations suggest a relatively indolent short-term natural course and may support a conservative management strategy for incidentally detected GGNs in children. Given the limited follow-up duration, these findings should be interpreted with caution. Further studies with longer follow-up durations and larger sample sizes are warranted to elucidate the long-term natural course of pediatric GGNs.
Deep learning for automated diagnosis and differentiation of otitis media on temporal bone CT.
Tang Y, Fan W, Cheng Z … +6 more , Leng Y, Liu Y, Wu T, Su S, Fei J, Li L
BMC Med Imaging · 2026 May · PMID 42098614 · Full text
BACKGROUND: The accurate diagnosis of otitis media (OM) has long been a challenge for clinicians (especially for less experienced clinicians) due to the variety of types and the complex anatomical structures of the middl...
BACKGROUND: The accurate diagnosis of otitis media (OM) has long been a challenge for clinicians (especially for less experienced clinicians) due to the variety of types and the complex anatomical structures of the middle ear. Although deep learning (DL) based on different examination methods (mostly otoscopy) has been applied to the diagnosis of single species OM in previous studies, DL using temporal bone computed tomography (TBCT) images to diagnose OM and simultaneously differentiate between chronic otitis media (COM) and otitis media with effusion (OME) has not been investigated in depth. This study aimed to develop and evaluate a DL framework for the automated diagnosis of OM and identifying OME and COM with or without cholesteatoma using TBCT images. METHODS: Our team created a unique large dataset of 2011 TBCT images from 1200 patients who were diagnosed with OM, which was determined the regions of interest (ROI) for middle ear (ME) by experienced experts. Then, a DL model was trained to detect the MEs in TBCT images and determine the OM status with this dataset of pre-processed images. Five-fold cross-validation was utilized for training and selecting the models. Finally, we evaluated the model using 406 images and verified the effectiveness of model-assisted diagnosis for different levels of clinicians in a comparative study. RESULTS: In the detection of the ME, the DL model achieved a detection ratio of 98.53%. The model showed satisfying performance in the classification of normal middle ear (NME), OME, and COM with an accuracy of 0.9238. With the assistance of the DL, the diagnostic accuracies were significantly improved from 81.53% to 93.60% (junior clinician) and from 87.93% to 95.57% (senior clinician), respectively. CONCLUSIONS: The findings suggested that the DL model could accurately identify MEs in TBCT images and classify NME, OME, and COM with satisfying accuracy. DL could also effectively assist clinicians in TBCT interpretation for OM diagnosis.
Predicting MRI-derived total brain volume from DXA-derived head composition in middle-aged and older adults: WASEDA'S Health Study.
Tsutsui T, Torii S, Tanisawa K … +12 more , Takahashi T, Usui K, Nakamura N, Midorikawa T, Nakagawa K, Ohkuma R, Kumano H, Ishii K, Suzuki K, Sakamoto S, Higuchi M, Oka K
BMC Med Imaging · 2026 May · PMID 42092792 · Full text
BACKGROUND: Total brain volume (TBV) derived from brain MRI is an important marker of brain structural health in middle-aged and older adults, but MRI is resource-intensive and not always feasible in largescale or repeat...
BACKGROUND: Total brain volume (TBV) derived from brain MRI is an important marker of brain structural health in middle-aged and older adults, but MRI is resource-intensive and not always feasible in largescale or repeated assessments. We examined whether dual-energy X-ray absorptiometry (DXA)-derived head composition measures can estimate MRI-derived TBV in middle-aged and older adults. METHODS: This study included 314 participants (≥ 40 years) who underwent whole-body DXA (head ROI manually defined using a sub-region tool) and 3T brain MRI within 1 year. MRI-derived TBV was defined as the sum of gray and white matter volumes. We developed multivariable linear regression models using either DXA-derived head lean-and-fat mass or head fat mass as the primary predictor. Nested models were fitted: Model 1 (predictor only), Model 2 (+ age and sex), and Model 3 (+ BMI). Apparent model performance was summarized using R² and RMSE, and internal validation was performed using 1,000 bootstrap resamples to obtain optimism-corrected performance estimates. Calibration was evaluated using calibration-in-the-large (CITL) and calibration slope. Agreement between observed and predicted TBV was assessed using Bland-Altman analysis. Sensitivity analyses additionally adjusted for the MRI-DXA measurement interval and evaluated sex-stratified performance. RESULTS: Model 3 was treated as the prespecified primary model because it was the fully adjusted model including clinically relevant covariates. In Model 3, both head lean-and-fat mass and head fat mass were positively associated with TBV, whereas age was negatively associated and male sex was associated with larger TBV. Across the nested models, optimism-corrected bootstrap validation showed broadly similar performance, with numerically slightly higher R² values and lower RMSE values for Model 3. Calibration was favorable in both predictor-based primary models (CITL approximately 0; calibration slope approximately 1.00). Bland-Altman analyses showed small mean bias with evidence of proportional bias across the TBV range. Bootstrap validation indicated stable performance. Sensitivity analyses yielded similar results after accounting for measurement interval and across sex strata. CONCLUSIONS: DXA-derived head composition measures can provide a practical approximation of MRI-derived TBV in middle-aged and older adults, with good calibration and stable internal validation performance.
Pre-treatment prediction of microsatellite instability in colon cancer: a nomogram model combining clinicopathological features and pre-treatment CT-based radiomics.
Wei M, Jia C, Zhang Y … +2 more , You P, Chen W
BMC Med Imaging · 2026 May · PMID 42087105 · Full text
BACKGROUND: Determining microsatellite instability (MSI) status in colon cancer is crucial for selecting treatment strategies in advanced stages. Thus, accurately identifying MSI status before treatment is essential. OBJ...
BACKGROUND: Determining microsatellite instability (MSI) status in colon cancer is crucial for selecting treatment strategies in advanced stages. Thus, accurately identifying MSI status before treatment is essential. OBJECTIVE: This study aims to evaluate the utility of nomogram model that integrates clinicopathological indicators and pre-treatment CT-based radiomics features for predicting DNA mismatch repair deficiency (dMMR) /microsatellite instability (MSI) status in colon cancer prior to treatment. METHODS: A total of 201 colon cancer patients who had undergone preoperative contrast-enhanced CT scans were categorized into the dMMR/MSI group or the proficient Mismatch Repair (pMMR)/Microsatellite Stable (MSS) group based on surgical pathology results. Multivariate logistic regression was applied to identify independent clinical predictors. The least absolute shrinkage and selection operator (LASSO) regression was applied for dimensionality reduction of radiomics features. Clinical, radiomics, and nomogram models were established through logistic regression analysis based on the risk clinicopathological predictors and radiomics features. RESULTS: Multivariate logistic regression identified patient age, pericentric lymph node metastasis, and CA72-4 levels as significant (P < 0.05). Four radiomic features were selected to construct the radiomics model. In the training set, the AUC values for the clinical model, Rad score, and combined model were 0.86, 0.89, and 0.94, respectively, and in the validation set, 0.81, 0.89, and 0.91, respectively. The Delong test showed the nomogram model outperformed both the clinical model and Rad score (P < 0.05). The calibration curve confirmed good consistency between predicted and actual outcomes for dMMR/MSI colon cancer using the combined model. CONCLUSION: The nomogram model, which combines clinicopathological features with pre-treatment CT-based radiomics features, demonstrates greater predictive accuracy for dMMR/MSI colon cancer than the standalone clinical and radiomics models.
Prediction of tumor regression grading in rectal cancer neoadjuvant chemoradiotherapy: a habitat radiomics analysis of imaging biomarker.
Sha X, Dou X, Ma L … +6 more , Qiu Q, Li Z, Li T, Cui Y, Shu H, Yin Y
BMC Med Imaging · 2026 May · PMID 42087101 · Full text
BACKGROUND: Tumor regression grading (TRG) is a core prognostic predictor of treatment outcomes in rectal cancer. Conventional TRG assessment methods are limited in capturing the full complexity of intratumoral heterogen...
BACKGROUND: Tumor regression grading (TRG) is a core prognostic predictor of treatment outcomes in rectal cancer. Conventional TRG assessment methods are limited in capturing the full complexity of intratumoral heterogeneity. Advances in medical imaging, particularly radiomics and habitat-based analysis, hold promise the improve TRG prediction by quantitatively characterizing subregional tumor features. This study aimed to evaluate the performance of habitat radiomics in preoperatively predicting TRG in rectal cancer patients receiving neoadjuvant chemoradiotherapy (nCRT). METHODS: Computed tomography (CT) images were analyzed to compare the predictive performance of conventional radiomics features and habitat-based analysis. Tumor regions of interest (ROIs) were segmented, extracting local imaging features. Voxel-level clustering was employed to identify distinct intratumoral subregions. Machine learning algorithms, including ExtraTrees, support vector machine (SVM), and Random Forest, were applied to predict TRG. RESULTS: For the conventional radiomics model, the ExtraTrees algorithm yielded superior performance, with AUCs of 0.912 and 0.817 in training and testing cohorts, respectively, outperforming SVM and Random Forest. The habitat model outperformed conventional radiomics model, while the combined model integrating habitat features and clinical variables yielded the optimal efficacy (training AUC = 0.916, test AUC = 0.833). In the binary classification task of TRG0 (pathologic complete response, pCR) vs. TRG1-2, the Habitat model achieved a test AUC of 0.884, and the combined model further reached 0.929. SHAP analysis identified that features from the H1 subregion and wavelet-transformed features were the top predictive contributors. CONCLUSION: Habitat-based radiomics, especially when integrated with clinical data, significantly improves the preoperative prediction of TRG in rectal cancer patients undergoing nCRT, providing a powerful tool to advance personalized oncology. Further validation in large-scale, multicenter, independent cohorts is warranted to facilitate the clinical translation of this approach.
The effect of head coil configuration and channel count on the quality of double inversion recovery (DIR) MRI images.
Alahmadi A, Alshehri RA, Gasem RA … +15 more , Aljuhani A, Bedaiwi A, Malaih AA, Alghamdi J, Alsalamah A, Alkhateeb SM, Waggass G, Khalil M, Alhasan MS, Alshamrani KM, Hendi AM, Aldusary N, Alsharif W, Hakami NY, Kanbayti IH
BMC Med Imaging · 2026 May · PMID 42087085 · Full text
BACKGROUND: Double inversion recovery (DIR) MRI provides high sensitivity for detecting white matter abnormalities but suffers from reduced signal-to-noise ratio (SNR) due to simultaneous suppression of multiple tissue s...
BACKGROUND: Double inversion recovery (DIR) MRI provides high sensitivity for detecting white matter abnormalities but suffers from reduced signal-to-noise ratio (SNR) due to simultaneous suppression of multiple tissue signals. Head-coil configuration and channel count may influence the resulting image quality. METHODS: Seventeen healthy subjects underwent DIR imaging on a 3-T MRI system using both 64-channel and 20-channel head/neck coils. Quantitative image quality was assessed using SNR and contrast-to-noise ratio (CNR) measurements across multiple brain regions, with comparisons performed using paired t-tests. Structural Similarity Index Measure (SSIM) was additionally computed between registered 64-channel and 20-channel DIR images to quantify inter-coil structural image similarity. Qualitative image quality was evaluated by three experienced neuroradiologists using a 5-point rating scale for contrast, spatial resolution, and noise; inter-rater agreement was assessed using Kendall's coefficient of concordance (Kendall's W). RESULTS: Quantitative analysis demonstrated significantly higher SNR and CNR values for the 64-channel coil compared with the 20-channel coil across all assessed regions (p < 0.0001). Qualitative evaluation showed that images acquired with the 64-channel coil received marginally higher mean scores for contrast, spatial resolution, and noise from all raters; inter-rater agreement was moderate-to-strong across all domains (Kendall's W = 0.33-0.89). CONCLUSION: At 3 T, the use of a 64-channel head/neck coil provides significant quantitative improvements in DIR image quality compared with a 20-channel coil, with small but consistent advantages also observed in qualitative assessments. These findings support the use of higher-channel-count coils to mitigate SNR limitations inherent to DIR imaging. However, qualitative differences between coil configurations were modest and inter-rater agreement was moderate-to-strong by Kendall's W (W = 0.33-0.89). The clinical benefit of the 64-channel coil in pathological conditions such as multiple sclerosis or cortical dysplasia requires further investigation in patient-based studies.
Comparison of elastosonographic changes of the tibial nerve and Achilles tendon in patients with type II diabetes mellitus.
Yang H, Huang J, Chen S … +4 more , Lin J, Cai K, Feng Q, He Y
BMC Med Imaging · 2026 May · PMID 42070019 · Full text
BACKGROUND: To compare the elastosonographic changes of the tibial nerve (TN) and Achilles tendon (AT) in patients with type 2 diabetes mellitus (T2DM) and explore their relationship and respective relevant factors. METH...
BACKGROUND: To compare the elastosonographic changes of the tibial nerve (TN) and Achilles tendon (AT) in patients with type 2 diabetes mellitus (T2DM) and explore their relationship and respective relevant factors. METHODS: This case-control study enrolled 165 subjects, comprising 126 patients with T2DM and 39 healthy controls matched for age and gender. The patients were further divided into those with and without diabetic peripheral neuropathy (PN-DM and NPN-DM groups). Clinical and laboratory data were collected. Conventional ultrasound and elastography were performed to assess the changes in the morphology and elasticity of the bilateral TN and AT. Sonographic features were compared across the three groups, relevant factors affecting the stiffness of TN and AT were analyzed, respectively. RESULTS: Diabetic patients exhibited significantly higher levels of HbA1C and a higher rate of smoking than healthy controls (P < 0.01 and P = 0.02, respectively). Their levels of body mass index (BMI) and total cholesterol have a significant difference between the NPN-DM group and healthy controls (both P = 0.02). The incidence of other microvascular complications in the NPN-DM group was significantly lower among diabetic patients (P = 0.04). Compared with healthy controls, the cross-sectional area (CSA) and transverse diameter of TN in diabetic patients were significantly larger (both P < 0.01), and CSA and anteroposterior diameter of AT were notably greater (P = 0.02 and P < 0.01). Besides, the stiffness of TN in the longitudinal section was significantly higher (P < 0.01), and the stiffness of AT in the cross-section was remarkably lower (P < 0.01). There was no significant difference in the morphology or elastography of TN or AT between NPN-DM and PN-DM groups. Furthermore, the stiffness of TN was not linearly related to that of AT, but independently correlated with age, HbA1C, and other microvascular complications (P < 0.05). The stiffness of AT was only independent of age (P < 0.01). CONCLUSIONS: The size of both TN and AT in diabetic patients was significantly larger. The stiffness of TN increased, and that of AT decreased; however, these changes were independent of each other. CLINICAL TRIAL NUMBER: Not applicable.
A multicenter-validated interpretable transformer model for pituitary microadenoma detection on non-contrast multiparametric MRI.
Kang S, Yang W, Yu Y … +4 more , Wang K, Yuan W, Jiang Y, Zhang J
BMC Med Imaging · 2026 May · PMID 42069529 · Full text
BACKGROUND: Detecting pituitary microadenomas using non-contrast multi-parametric magnetic resonance imaging (MRI) is challenging yet essential for clinical decisions. This study aimed to develop a transformer deep learn...
BACKGROUND: Detecting pituitary microadenomas using non-contrast multi-parametric magnetic resonance imaging (MRI) is challenging yet essential for clinical decisions. This study aimed to develop a transformer deep learning (DL) model for detecting pituitary microadenomas based on non-contrast multiparametric MRI and explore the explainability techniques to enhance transparency in convolutional neural network (CNN)-based classification. The primary research question addressed is how to improve the accuracy, generalization, and interpretability of CNNs for microadenomas detection. METHODS: Non-contrast multiparametric MRI sella area scans of 590 patients were retrospectively collected from three hospitals. The development and comparison of 2D_DL, 2.5D_DL, 2D_multichannel, and transformer models for classification. By incorporating Explainable AI (XAI), including Gradient-weighted Class Activation Mapping(Grad-CAM) and SHapley Additive exPlanations (SHAP), we improve model interpretability. RESULTS: The performance of the 2D_multichannel model, with an area under the curve (AUC) of 0.893, was better to that of the 2D_T1SAG_DL, 2D_T1COR_DL, 2D_T2COR_DL (AUC, 0.884, 0.779, and 0.846, respectively). The performance of the transformer model, with an area under the curve (AUC) of 0.985, was superior to that of the 2.5D_T1SAG_DL, 2.5D_T1COR_DL, 2.5D_T2COR_DL (AUC, 0.763, 0.863, and 0.835, respectively). The non-contrast MRI-based 2.5D_DL transformer model all shows outperforming performance in the internal and two external test sets (AUC, 0.874, 0.829, and 0.819, respectively). CONCLUSIONS: Given its robust diagnostic performance and enhanced interpretability, this model demonstrates significant potential for clinical translation as a decision-support tool in the detection of pituitary microadenomas.
Fusion attention-based nasopharyngeal carcinoma segmentation model in predicting the clinical outcome of cervical lymph node residue after IMRT.
Liu Y, Xie J, Li J … +6 more , Chen H, Dong S, Zhou S, Liang S, Qin C, Xiao L
BMC Med Imaging · 2026 May · PMID 42069522 · Full text
BACKGROUND: Deep learning methods have made great progress in the automatic segmentation of nasopharyngeal carcinoma, but challenges remain. PURPOSE: Computer-aided automatic segmentation of nasopharyngeal cancer primary...
BACKGROUND: Deep learning methods have made great progress in the automatic segmentation of nasopharyngeal carcinoma, but challenges remain. PURPOSE: Computer-aided automatic segmentation of nasopharyngeal cancer primary area is of great significance for automatic outlining of nasopharyngeal cancer target areas and accurate prediction of responsiveness and prognosis of metastatic lymph nodes in the neck after radiotherapy. In this paper, we use deep learning methods to construct an automatic segmentation network for gross target volume of nasopharynx, combine clinical factors and radiomics features to establish a radiomics nomogram model, which will then predict the final outcome of metastatic lymph nodes that have not achieved complete remission after radical radiotherapy. METHODS: Clinical and IMRT radiotherapy plan CT data were retrospectively collected from 69 patients who received intensity-modulated radiation therapy between July 2014 and December 2016. These patients exhibited residual metastatic lymph node lesions without residual primary lesions on the first follow-up MRI and had continuous follow-up records. The median follow-up was 53 months (IQR 39.75-62.37), with 30 patients eventually regressing and 39 patients persisting or progressing. The ct images of 69 radiotherapy plans were randomly divided into training and test sets according to 8:2, and a fusion attention-based model was trained for automatic nasopharyngeal carcinoma segmentation. Based on the unet framework, a fusion attention model was proposed, and a 2·5 d convolutional neural network was used to deal with the anisotropy. An improved channel and spatial attention module is fused in the codec 4 layer to enable the network to focus on small targets. 2d interlaced sparse self-attention module is extended to 3d to better extract the feature information of the tumor target area and solve the problem of low contrast between the target area and the surrounding soft tissues, thus optimizing the overall segmentation effect. The performance of the segmentation model was evaluated using the mean dice coefficient, relative volume error (RVE), average symmetric surface distance (ASSD) and hausdorff distance (HD), using the target area of the primary lesion of nasopharyngeal carcinoma manually outlined by a senior radiation therapy specialist as the gold standard. Radiomics features were extracted using the pyradiomics package, and the classification performance of the radiomics model was assessed by the area under the curve of the receiver operating curve (ROC). RESULTS: The average dice coefficient, RVE, ASSD and HD of our model for nasopharyngeal carcinoma were 75.05%, 14.63%, 2.224 mm, and 8.75 mm, respectively, which were 11.01%, 26.34%, 3.101 mm, and 52.58 mm better than the baseline 3dunet model. The radiomic features were an effective predictor of tumor outcome in nasopharyngeal carcinoma, with the highest area under the receiver operating characteristic curve (AUC) of 0.892 for the radiomic nomogram in the training set and 0.825 for the radiomic model in the test set. CONCLUSIONS: The fused attention-based segmentation network for nasopharyngeal carcinoma can effectively and reliably segment the region of the primary nasopharyngeal carcinoma, and the radiomic nomogram can effectively predict the response after treatment.
Impact of maras powder on mandibular bone microarchitecture: a fractal and radiomorphometric study.
Ararat E, Talmaç AGÖ
BMC Med Imaging · 2026 May · PMID 42069516 · Full text
BACKGROUND: The aim of this study was to examine how maras powder (MP) affects on the cortical and trabecular bone of the mandible using the radiomorphometric indexes and fractal dimension (FD). METHODS: A retrospective...
BACKGROUND: The aim of this study was to examine how maras powder (MP) affects on the cortical and trabecular bone of the mandible using the radiomorphometric indexes and fractal dimension (FD). METHODS: A retrospective analysis of radiographic records of 150 male individuals, 50 of whom used MP, 50 of whom smoked cigarettes, and 50 of whom were healthy and did not use any tobacco derivatives, was performed. Cortical bone was evaluated with mandibular cortical width (MCW) and panoramic mandibular index (PMI). Trabecular bone in mandibular anterior was evaluated by FD. The ANOVA test was used to compare normally distributed variables across the three groups, and the Kruskal Wallis test was used to compare non-normally distributed variables across the three groups. RESULTS: The mean age of MP users was 42.92 ± 10.21; in smokers, 40.46 ± 10.51; and in the healthy control group, 40 ± 15.05. When the FD measurements were examined in regions of interest (ROI) 1, ROI 2, ROI 3, and the mean ROI values, no significant difference was found between the three groups in terms of FD (p > 0.05), but the fractal dimension was found to be lower in individuals using MP. No significant difference was found between the groups in terms of histogram values and MCW and PMI measurements (p > 0.05). CONCLUSION: No significant differences were found between users of MP, smokers, and healthy individuals. However, the decreasing trend in FD values may indicate early effects of MP. Studies with larger sample sizes and advanced imaging techniques are needed.
Vascular ultrasound-based risk stratification model for atherosclerotic cardiovascular disease in patients with type 2 diabetes mellitus.
Chen C, Xu Q, Xu H … +6 more , Xu Y, He X, Lin M, Li Y, Li Y, Liu L
BMC Med Imaging · 2026 May · PMID 42069513 · Full text
BACKGROUND: This study aimed to investigate the ability of an ultrasound-based risk stratification model integrating carotid intima thickness (CIT) and carotid-femoral pulse wave velocity (cfPWV) to aid in risk stratific...
BACKGROUND: This study aimed to investigate the ability of an ultrasound-based risk stratification model integrating carotid intima thickness (CIT) and carotid-femoral pulse wave velocity (cfPWV) to aid in risk stratification and assessment of atherosclerotic cardiovascular disease (ASCVD) in patients with type 2 diabetes mellitus (T2DM), thereby providing an objective basis for identifying high-risk individuals and informing individualized management strategies. METHODS: A total of 105 patients with T2DM were enrolled in this study. According to the 10-year ASCVD risk score, patients were further classified into T2DM patients with low-to-moderate burden of other cardiovascular risk factors and T2DM patients with high burden of other cardiovascular risk factors. CIT was measured using high-resolution ultrasound to assess vascular structure, while cfPWV was evaluated using the automatic measurement of arterial stiffness (AMAS) system to assess vascular function. Logistic regression and least absolute shrinkage and selection operator (LASSO) regression analyses were performed to identify independent risk factors of high ASCVD risk. Based on these risk factors, individual discriminative models and a nomogram were constructed. Receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA) were used to evaluate model performance, and differences among models were assessed using the DeLong test. RESULTS: CIT, cfPWV, and estimated glomerular filtration rate (eGFR) were identified as independent risk factors of high 10-year ASCVD risk in patients with T2DM. The areas under the curve (AUCs) for the CIT model, cfPWV model, eGFR model, combined CIT-cfPWV model, and the nomogram were approximately 0.781, 0.808, 0.797, 0.831, and 0.875, respectively. The constructed nomogram demonstrated excellent discrimination, calibration, and clinical applicability. CONCLUSIONS: CIT and cfPWV show strong potential for identifying T2DM patients at high ASCVD risk as estimated by the China-PAR model. Incorporating these parameters into vascular evaluation may aid in risk stratification and provide a robust basis for individualized clinical intervention strategies. Prospective studies are needed to validate their prognostic value for future ASCVD events.
Gray matter volume and structural covariance alterations in young males with childhood-onset growth hormone deficiency.
Wang S, Su W, Liu X … +8 more , Zhao Y, Zhang M, Lu L, Zhang S, Wang D, Yang W, Zhang Y, Kuai X
BMC Med Imaging · 2026 Apr · PMID 42062975 · Full text
PURPOSE: This study investigated the differences in gray matter volume (GMV), structural covariance, and structural covariance networks between adults with childhood-onset growth hormone deficiency (GHD) and healthy cont...
PURPOSE: This study investigated the differences in gray matter volume (GMV), structural covariance, and structural covariance networks between adults with childhood-onset growth hormone deficiency (GHD) and healthy controls (HCs). MATERIALS AND METHODS: A total of 70 right-handed male participants (32 GHD patients and 38 HCs) underwent high-resolution T1-weighted magnetic resonance imaging. Atlas-based morphometry was conducted to examine regional GMV differences. Structural covariance matrices were generated using inter-regional partial correlations, and graph-based network analyses were performed to assess global and regional network topological properties. RESULTS: The GHD group exhibited significantly reduced GMV in the bilateral pallidum and nucleus accumbens compared to HCs, while showing increased GMV in regions such as the frontal, parietal, temporal lobes, and thalamus (family wise error-corrected P < 0.05). Structural covariance analysis indicated increased connectivity in the GHD group between the left precentral gyrus and bilateral temporal poles, as well as between the right subgenual anterior cingulate cortex and right superior parietal gyrus (false discovery rate-corrected P < 0.001). Both groups demonstrated small-world organization in their structural brain networks, with no significant differences in global network metrics across the density range of 0.1 to 0.5. Regional analyses revealed variations in clustering coefficient, nodal degree, and betweenness centrality, predominantly in somatomotor network regions and areas associated with self-reflection and semantic processing. However, these regional findings did not survive false discovery rate correction for multiple comparisons. CONCLUSION: Despite preserved global brain network organization, childhood-onset GHD demonstrates region-specific GMV abnormalities and altered local structural connectivity, which may impact brain regions involved in motor, cognitive, and emotional regulation.
Fluoroscopic image-driven deep learning model for predicting intussusception irreducibility during air enema in children.
Zhou H, Huang J, Zhang Y … +7 more , Pan H, Li Z, Zhang R, Ma X, Li Z, Zhang Z, Yu G
BMC Med Imaging · 2026 Apr · PMID 42062943 · Full text
BACKGROUND: Accurate identification of irreducible intussusception during air enema is crucial for optimizing enema strategies. Current methods are limited by subjective interpretation and inconsistent clinical criteria....
BACKGROUND: Accurate identification of irreducible intussusception during air enema is crucial for optimizing enema strategies. Current methods are limited by subjective interpretation and inconsistent clinical criteria. We developed a deep learning (DL) framework to objectively predict irreducibility using air enema fluoroscopic images. METHODS: In this retrospective study, a hybrid ensemble DL model was developed using fluoroscopic images acquired during air enema, comprising 770 irreducible and 1214 reducible cases. Model performance was evaluated on a real-world test set (46 irreducible vs. 802 reducible cases) and an external test set (9 irreducible vs. 101 reducible cases), with benchmarking against state-of-the-art techniques. The model's performance was further compared with radiologists' interpretations, and its ability to improve diagnostic accuracy was assessed. Performance was evaluated using receiver operating characteristic (ROC) analysis and confusion matrix-derived metrics. RESULTS: The proposed model achieved areas under the ROC curves (AUCs) of 0.89 (95% CI: 0.836-0.944) and 0.883 (95% CI: 0.78-0.968) on the real-world and external test sets, respectively, outperforming comparative methods (AUC ranges: 0.823-0.877 and 0.634-0.826). The model demonstrated superior performance compared with that of the intermediate radiologist (AUC: 0.89 vs. 0.804; P < 0.001) and comparable performance to that of a senior radiologist (AUC: 0.89 vs. 0.842; P = 0.108). When used as an assistive tool, the model significantly improved radiologists' diagnostic performance (all P < 0.01), with AUC improvements of 0.095-0.072, balanced accuracy gains of 8.6-11.7%, and specificity increases of 18.7-22.6%. CONCLUSIONS: The proposed model demonstrated promising diagnostic performance in identifying irreducible intussusception and may serve as an effective decision-support tool to improve radiologists' diagnostic accuracy during air enema procedure.
About
- Frequency
- Sun
- Papers found
- 200
- RSS feed
- Subscribe