Search Results (52)

Accurate brain tumor segmentation in multimodal magnetic resonance imaging (MRI) scans is essential for quantitative neuro-oncology analysis, treatment planning, and longitudinal disease monitoring. However, automatic delineation remains challenging because glioma subregions exhibit heterogeneous appearance across T1-weighted (T1), contrast-enhanced T1-weighted (T1ce), T2-weighted (T2), and fluid-attenuated inversion recovery (FLAIR) sequences; weak and irregular lesion boundaries; and severe scale variation between edema, tumor core, and enhancing tumor regions. This paper presents NeuroSPFNet, a vision–language semantic prior-guided frequency-enhanced segmentation framework for tumor subregion delineation in multimodal MRI volumes. NeuroSPFNet integrates four complementary components: a Clinical Vision–Language Prior Component (CVLPC) for category-level semantic guidance; a Frequency-Enhanced Boundary Module (FEBM) for boundary-sensitive representation learning; a Multi-Scale Lesion-Dominated Fusion Module (MLDFM) for hierarchical lesion aggregation; and a Lesion-Semantic Boundary Consistency Loss (LSBCL) for jointly constraining region overlap, boundary quality, and semantic alignment. The task-specific contribution lies in the closed coupling of shared BraTS-region semantic prototypes, gated frequency responses, lesion-weighted multiscale features, and semantic-boundary supervision within one volumetric segmentation pipeline. On the Brain Tumor Segmentation (BraTS) 2021 benchmark, NeuroSPFNet achieved Dice scores of 91.94%, 88.84%, and 85.38% for whole tumor, tumor core, and enhancing tumor, respectively, with a mean Dice of 88.72% and a mean 95th-percentile Hausdorff distance (HD95) of 3.58 mm. Ablation studies show distinct effects across subregions and metrics: semantic priors primarily strengthen tumor-core and enhancing-tumor discrimination, whereas frequency and boundary constraints primarily reduce contour error. NeuroSPFNet also attains competitive performance under the current setting, with moderate computational overhead of 42.58 million parameters, 78.60 billion floating-point operations (GFLOPs), and 265 ms per volume. Full article

Accurate, automated delineation of adult diffuse gliomas from multi-parametric magnetic resonance imaging (mpMRI) is central to quantitative neuro-oncology. Volumetric 3D networks dominate the BraTS leaderboard but require expensive GPUs, long training cycles, and provide diminishing returns relative to their compute budget. Slice-wise 2D models, by contrast, discard inter-slice context that is informative for thin tumor rims and small enhancing foci. We introduce GDNet, a 2.5D multimodal MRI segmentation framework for adult glioma evaluated on the BraTS 2024 cohort. GDNet consumes a stack of three adjacent axial slices from the four standard BraTS modalities (T1, T1ce, T2, FLAIR) as a 12-channel input to a compact U-shaped encoder–decoder with Group Normalization and predicts whole tumor (WT), tumor core (TC), and enhancing tumor (ET) masks for the central slice. The training pipeline pairs the 2.5D backbone with: (i) Exponential Moving Average (EMA) of model weights with decay 0.999, (ii) mixed tumor-aware slice sampling (p_tumor = 0.50), (iii) a compound Cross-Entropy + Soft-Dice loss, and (iv) AdamW with warm-up plus cosine annealing under Automatic Mixed Precision. We performed a systematic, step-by-step ablation covering a 2D baseline, EMA + mixed sampling, tumor-centered crop fine-tuning, a GDNet-inspired architectural integration, a region-aware loss, 3-slice and 5-slice 2.5D inputs, and connected-component post-processing, and we report multi-seed results to quantify reproducibility. On the held-out BraTS 2024 test partition, the final 3-slice 2.5D GDNet achieved positive-only Dice scores of 0.791 ± 0.000 (WT), 0.736 ± 0.003 (TC), 0.654 ± 0.004 (ET), and a mean foreground positive-only Dice of 0.820 ± 0.000 across seeds; the all-slice mean foreground Dice exceeded 0.927 ± 0.000. Validation positive-only scores were 0.805 ± 0.002 (WT), 0.757 ± 0.004 (TC), 0.683 ± 0.009 (ET). The inter-seed standard deviation was small for every region (≤0.01 Dice points), indicating low inter-seed variance across the two seeds evaluated; with only two seeds, we regard this as preliminary evidence of training stability rather than a strong reproducibility claim. The ablation isolated EMA + mixed tumor sampling and the 2.5D context window as the dominant sources of improvement; notably, a GDNet-style architectural integration with a region-aware loss did not outperform the simpler 2.5D U-Net on positive-only WT/TC/ET, and light post-processing improved only all-slice Dice. A failure-mode audit found that the residual catastrophic predictions are concentrated on a small minority of diffuse, infiltrative tumors with mass effect. Conclusions: Carefully engineered training strategies, tumor-aware sampling, EMA stabilization, and a modest 2.5D context window recover a substantial fraction of the accuracy of much heavier 3D networks at a fraction of the compute, are reproducible across seeds, and outperform a heavier GDNet-inspired architectural variant on the same data. GDNet is therefore a practical and, pending external validation, potentially clinically deployable framework for multimodal glioma segmentation on workstation-class GPU hardware. Full article

Background/Objectives: Accurate automatic segmentation of multimodal magnetic resonance imaging (MRI) is essential for neurosurgical planning and image-guided procedures. However, existing three-dimensional segmentation models often struggle with low lesion-to-tissue contrast, ambiguous tumor boundaries, small enhancing tumor regions, and performance degradation caused by missing imaging modalities. This study aimed to develop a robust segmentation framework that improves cross-modal representation learning, boundary recovery, and segmentation performance under incomplete-input conditions. Methods: We propose PF-CMNet, a Progressive Frequency-Aware Cross-Modal Network with Missing-Modality Distillation for three-dimensional brain tumor segmentation. The network introduces a Cross-Modal Selective Frequency Attention module in the early encoder stage to model modality-specific frequency responses and spatially adaptive cross-modal correlations. A Progressive Cross-Scale Detail Fusion decoder is further employed to aggregate multilevel semantic features and refine high-resolution boundary details. To enhance robustness under missing-modality conditions, a teacher–student distillation strategy transfers full-modality predictions and shallow feature knowledge to a student network trained with random modality dropout. Results: On the MSD Task01_BrainTumour dataset, PF-CMNet achieved an average Dice score of 84.3%, with Dice scores of 79.6%, 82.8%, and 90.4% for enhancing tumor, tumor core, and whole tumor, respectively. On the BraTS2021 dataset, the model achieved an average Dice score of 88.2% and the lowest average 95th percentile Hausdorff distance among the compared methods. In predefined complete-modality absence stress tests, where unavailable MRI sequences were zero-masked to model the absence of input modalities rather than partial image degradation, the distilled model maintained average Dice scores of 78.64%, 82.58%, 58.39%, 82.03%, and 79.29% when FLAIR, T1, T1ce, T2, and T1 + T2 were unavailable, respectively. Conclusions: PF-CMNet provides a unified framework for multimodal brain tumor segmentation, improving full-modality segmentation accuracy, boundary consistency, and robustness to incomplete MRI inputs while maintaining a favorable accuracy–efficiency trade-off. Full article

Objectives: Contrast-enhanced magnetic resonance imaging (CE-MRI) plays an important role in the diagnosis, treatment monitoring, and follow-up of brain tumors. However, the use of gadolinium-based contrast agents (GBCAs) is limited in patients with contraindications, such as severe renal impairment or situations requiring repeated examinations. This study aimed to develop a diffusion model-based Difference-Aware Guided Control Network (DAGCN) for synthesizing high-quality contrast-enhanced T1-weighted MRI (T1-CE) from non-contrast T1-weighted images in combination with an auxiliary sequence. Methods: Using the BraTS 2021 dataset, we proposed a two-stage generative framework that first localizes lesion-related enhancement cues and then guides image synthesis. In the first stage, a Difference-Aware Fusion and Prediction (DAFP) module was designed to extract complementary information from non-contrast T1-weighted images and an auxiliary sequence (T2-weighted or FLAIR) through dual-branch feature extraction and cross-modal channel attention fusion, followed by prediction of a lesion-related discrepancy map. In the second stage, the predicted discrepancy map was concatenated with the original T1-weighted images and introduced into a ControlNet-guided diffusion model to constrain the reverse denoising process and generate the target T1-CE image. Model performance was evaluated by visual comparison, quantitative metrics including peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), visual information fidelity (VIF), and normalized cross-correlation (NCC), as well as blinded radiologist scoring of image quality (IQ), clinical replaceability (IC), contrast enhancement (CE), and lesion conformity (CF). Results: DAGCN generated synthetic T1-CE images with preserved global anatomical structure and faithful local lesion enhancement without the need for contrast agent administration. Compared with baseline methods, DAGCN achieved the highest PSNR and NCC under both T1 + T2 and T1 + FLAIR settings, while showing competitive SSIM and VIF performance. Visual comparison and radiologist-based subjective evaluation further indicated improved lesion-focused enhancement fidelity and reduced false-positive enhancement. Among the two auxiliary sequence settings, the T1 + FLAIR configuration provided more specific lesion localization and cleaner background suppression than the T1 + T2 configuration, particularly by reducing interference from cerebrospinal fluid signals. Conclusions: The proposed DAGCN framework enables the synthesis of clinically informative contrast-enhanced-like MRI from non-contrast multi-sequence inputs and may provide a promising alternative for patients in whom gadolinium administration is contraindicated or should be avoided. In particular, the FLAIR-guided setting showed advantages in lesion specificity, background cleanliness, and overall diagnostic quality. Full article

Background/Objectives: Glioblastoma (GBM) comprises a contrast-enhancing (CE) mass and peritumoral hyperintensity on fluid-attenuated inversion recovery (FLAIR) on magnetic resonance imaging (MRI). Although a smaller residual FLAIR lesion volume (FLV) after gross total resection (GTR) is associated with longer survival, the clinical observation of rapid FLV reduction within the first postoperative month remains unknown. We evaluated the independent prognostic contributions of early and late FLV reduction in patients with newly diagnosed GBM, isocitrate dehydrogenase (IDH)-wildtype, who underwent GTR. Methods: In a retrospective cohort of 51 adults with GBM, IDH-wildtype, who underwent GTR and standard chemoradiotherapy, semi-automated FLV was measured on MRI preoperatively, immediately postoperatively (FLV0), at 1 month (FLV1), and at 3 months (FLV3). FLV changes were expressed as percentage changes relative to the preceding timepoint (ΔFLV0-1, ΔFLV1-3). Using a 3-month landmark analysis, multivariable Cox models for overall survival (OS) were fitted, adjusted for age, O6-methylguanine-DNA-methyltransferase promoter (pMGMT) methylation, and postoperative Karnofsky Performance Status (KPS). Results: In the multivariable 3-month landmark analysis, each 10-percentage-point early FLV reduction (∆FLV0-1^[10-pp]) was associated with lower mortality risk (hazard ratio [HR] 0.91, 95% confidence interval [CI] 0.82–0.99, p = 0.037) whereas a late FLV reduction (∆FLV1-3^[10-pp]) showed a smaller association with OS (HR 0.99, 95% CI 0.98–1.00, p = 0.043). In an exploratory dichotomized analysis at a 20% reduction threshold, early responders (∆FLV0-1 ≥ 20%) had markedly better OS (HR 0.33, p = 0.010). Conclusions: In selected patients with GBM, IDH-wildtype, who achieved radiographic GTR and received standard chemoradiotherapy, early postoperative FLV reduction at 1 month independently predicted longer OS, with a substantially larger effect than late reduction. Prospective multi-center validation is required before clinical implementation. Full article

Accurate brain tumor segmentation in magnetic resonance imaging (MRI) is essential for diagnosis, treatment planning, and therapy monitoring. Conventional deep learning models often struggle with large variations in tumor shape, size, and contrast, as well as severe foreground–background imbalance. To address these challenges, this study presents Deep-MiSR, an enhanced encoder–decoder framework built upon DeepLabV3+ with a MobileNetV2 backbone, tailored for single-modality contrast-enhanced T1-weighted (T1CE) MRI segmentation. Three complementary components are integrated into the architecture: mixed depthwise convolution (MixConv) with heterogeneous kernels within the atrous spatial pyramid pooling module for multi-scale feature aggregation, a squeeze-and-excitation block for adaptive channel recalibration, and R-Drop regularization that enforces prediction consistency via symmetric Kullback–Leibler divergence. The model was evaluated on 3064 T1CE slices from 233 patients drawn from the publicly available Nanfang Hospital brain MRI dataset. Deep-MiSR achieved a Dice similarity coefficient of 0.9281, a mean intersection-over-union of 0.8738, a precision of 0.8839, and a 95th-percentile Hausdorff distance of 7.69 mm, demonstrating consistent improvements over both the DeepLabV3+ baseline and all prior methods evaluated on the same data. Ablation studies confirmed that each component contributes independently, with R-Drop providing the largest individual gain. These findings demonstrate that combining multi-scale convolution, channel attention, and consistency regularization constitutes an effective and computationally practical strategy for robust single-modality brain tumor segmentation. Full article

Background: The aim of the present systematic review (SR) is to compile evidence regarding the use of radiomic-based machine learning (ML) models for predicting human papillomavirus (HPV) status in oropharyngeal squamous cell carcinoma (OPSCC) patients and to assess their reliability, methodological frameworks, and clinical applicability. The SR was conducted following PRISMA 2020 guidelines and registered in PROSPERO (CRD42025640065). Methods: Using the PICOS framework, the review question was defined as follows: “Can radiomic-based ML models accurately predict HPV status in OPSCC?” Electronic databases (Cochrane, Embase, IEEE Xplore, BVS, PubMed, Scopus, Web of Science) and gray literature (arXiv, Google Scholar and ProQuest) were searched. Retrospective cohort studies assessing radiomics for HPV prediction were included. Risk of bias (RoB) was evaluated using Prediction model Risk Of Bias ASsessment Tool (PROBAST), and data were synthesized based on imaging modality, architecture type/learning modalities, and the presence of external validation. Meta-analysis was performed for externally validated models using MetaBayesDTA and RStudio. Results: Twenty-four studies including 8627 patients were analyzed. Imaging modalities included computed tomography (CT), magnetic resonance imaging (MRI), contrast-enhanced computed tomography (CE-CT), and ¹⁸F-fluorodeoxyglucose positron emission tomography (¹⁸F-FDG PET). Logistic regression, random forest, eXtreme Gradient Boosting (XGBoost), and convolutional neural networks (CNNs) were commonly used. Most datasets were imbalanced with a predominance of HPV+ cases. Only eight studies reported external validation results. AUROC values ranged between 0.59 and 0.87 in the internal validation and between 0.48 and 0.91 in the external validation results. RoB was high in most studies, mainly due to reliance on p16-only HPV testing, insufficient events, or inadequate handling of class imbalance. Deep Learning (DL) models achieved moderate performance with considerable heterogeneity (sensitivity: 0.61; specificity: 0.65). In contrast, traditional models provided higher, more consistent performance (sensitivity: 0.72; specificity: 0.77). Conclusions: Radiomic-based ML models show potential for HPV status prediction in OPSCC, but methodological heterogeneity and a high RoB limit current clinical applicability. Full article

Background: Neoadjuvant chemotherapy (NAC) is a standard preoperative intervention for early-stage breast cancer (BC). Dynamic contrast-enhanced magnetic resonance imaging (CE-MRI) has emerged as a critical tool for evaluating treatment response and pathological complete response (pCR) following NAC. Computational modeling offers a robust framework to simulate tumor growth dynamics and therapy response, leveraging patient-specific data to enhance predictive accuracy. Despite this potential, integrating imaging data with computational models for personalized treatment prediction remains underexplored. This case study presents a proof-of-concept prognostic tool that bridges oncology, radiology, and computational modeling by simulating BC behavior and predicting individualized NAC outcomes. Methods: CE-MRI scans, clinical assessments, and blood samples from three retrospective NAC patients were analyzed. Tumor growth was modeled using a system of partial differential equations (PDEs) within a reaction–diffusion mass transfer framework, incorporating patient-specific CE-MRI data. Tumor volumes measured pre- and post-treatment were compared with model predictions. A 20% error margin was applied to assess computational accuracy. Results: All cases were classified as true positive (TP), demonstrating the model’s capacity to predict tumor volume changes within the defined threshold, achieving 100% precision and sensitivity. Absolute differences between predicted and observed tumor volumes ranged from 0.07 to 0.33 cm³. Virtual biomarkers were employed to quantify novel metrics: the biological conversion coefficient ranged from 4 × 10⁻⁷ to 6 × 10⁻⁶ s⁻¹, while the pharmacodynamic efficiency coefficient ranged from 1 × 10⁻⁷ to 4 × 10⁻⁴ s⁻¹, reflecting intrinsic tumor biology and treatment effects, respectively. Conclusions: This approach demonstrates the feasibility of integrating CE-MRI and computational modeling to generate patient-specific treatment predictions. Preliminary model training on retrospective cohorts with matched BC subtypes and therapy regimens enabled accurate prediction of NAC outcomes. Future work will focus on model refinement, cohort expansion, and enhanced statistical validation to support broader clinical translation. Full article

Early diagnosis of pancreatic cancer, particularly in intraductal papillary mucinous neoplasm (IPMN), remains challenging despite advances in imaging and biomarkers. Pancreatic adenocarcinoma (PDAC) has a high mortality rate; therefore, its early detection and adequate interventions are necessary to improve the disease outcome. Most IPMNs are asymptomatic and discovered incidentally. Magnetic resonance imaging (MRI) is a preferred tool for diagnosing malignant IPMNs, with a sensitivity of 90.7–94.1% and a specificity of 84.7–87.2% in detecting mural nodules > 5 mm, a strong predictor of high-risk lesions. Radiomics further enhances diagnostic accuracy (sensitivity 91–96%, specificity 78–81%), especially when combined with CA 19-9, which has lower sensitivity (73–90%) but higher specificity (79–95%). Computed tomography (CT), though less effective for small mural nodules, remains widely used; its accuracy improves with radiomics and clinical variables (sensitivity 90.4%, specificity 74%). Conventional endoscopic ultrasonography (EUS) shows lower performance (sensitivity 60%, specificity 80%), but its advanced variations have improved outcomes. Contrast-enhanced EUS (CE-EUS) visualizes mural nodules with more than 90% sensitivity and involvement of the main pancreatic duct, with a sensitivity of 83.5% and a specificity of 87%. EUS–fine-needle aspiration (EUS-FNA) allows cyst fluid analysis; however, CEA, glucose, and KRAS/GNAS mutations show poor value for malignancy risk. Cytology has low sensitivity (28.7–64.8%) but high specificity (84–94%) in diagnostic malignant changes and strongly affects further management. EUS–through-the-needle biopsy (EUS-TTNB) yields high diagnostic accuracy (sensitivity 90%, specificity 95%) but carries a range of 2–23% adverse events, which limits its wide use. EUS–confocal laser endomicroscopy (EUS-nCLE) provides real-time microscopic evaluation, detecting malignant IPMN with a sensitivity of 90% and a specificity of 73%, though its availability is limited. New emerging biomarkers available in cyst fluid or blood include mucins, miRNA panels (sensitivity 66.7–89%, specificity 89.7–100%), lipidomics, and cancer metabolite profiling, with diagnostic accuracy approaching 89–91%. Pancreatoscopy (POP) enables direct main pancreatic duct (MPD) visualization and biopsy with a sensitivity of 64–100% and a specificity of 75–100%, though adverse events occur in around 12% cases. Combining advanced imaging, EUS-based tissue acquisition, and novel biomarkers holds promise for earlier and more accurate detection of malignant IPMN, potentially improving PDAC outcomes. Full article

This study aims to develop and validate a multisequence MRI-based radiomics approach for distinguishing metaplastic breast cancer (MBC) from non-metaplastic triple-negative breast cancer (TNBC) at the initial diagnosis, which could facilitate optimal treatment selection. In this retrospective study, we analyzed 105 patients (27 MBC, 78 non-metaplastic TNBC) who underwent standardized breast magnetic resonance imaging (MRI), which included T1-weighted contrast-enhanced (T1W-CE) and short-tau inversion recovery (STIR) sequences. Two radiologists performed ground truth lesion segmentation, verified by a senior radiologist. We extracted 214 radiomic features (using PyRadiomics) and used least absolute shrinkage and selection operator (LASSO) regression for feature selection. Seven machine learning classifiers were thoroughly evaluated using five-fold cross-validation, with performance assessed through ROC analysis and accuracy metrics. The combined T1W-CE and STIR analysis demonstrated superior diagnostic performance for distinguishing MBC from non-metaplastic TNBC (AUC = 0.845; accuracy = 81%) compared with either sequence alone (T1W only AUC = 0.805; accuracy = 80%; STIR only AUC:0.768; accuracy = 77%). Multisequence MRI radiomics can reliably distinguish between MBC and TNBC at the time of initial diagnosis. This could potentially facilitate the selection of more appropriate treatments and help avoid ineffective chemotherapy for MBC patients. Full article

Background/Objectives: This study aims to evaluate the additional value of [¹⁸F]F-fluorocholine ([¹⁸F]F-FCH) PET/CT over contrast-enhanced magnetic resonance imaging (CE-MRI) in detecting the recurrence of brain metastases (BMs) after stereotactic radiosurgery (SRS) in patients with lung cancer brain metastases (LCBMs). Methods: Thirty-one patients with suspected recurrence of BM in LCBM after SRS were enrolled in this retrospective study. They underwent both [¹⁸F]F-FCH PET/CT and CE-MRI within 2 weeks. The tumor imaging parameters and clinical features were analyzed. The results of histopathology or radiographic follow-up served as the reference standard for the final diagnosis. Results: In these 31 patients, there were 54 lesions, of which 27 lesions were proven to be BM recurrence, while 27 lesions were non-recurrence. [¹⁸F]F-FCH PET/CT showed high radiotracer uptake in recurrent lesions of BM and identified 24 positive lesions (88.89% of sensitivity), while CE-MRI indicated 23 positive lesions (85.19% of sensitivity). [¹⁸F]F-FCH PET/CT indicated higher specificity (81.48%) and accuracy (85.19%) in detecting recurrence of BM than CE-MRI (40.74% and 62.96%, both p < 0.05), particularly in frontal lobes and cerebella. For lesion sizes, the accuracy of [¹⁸F]F-FCH PET/CT in detecting recurrent lesions was higher than that of CE-MRI for lesions over 1.0 cm but below 2.0 cm (p = 0.016). The detective performance of [¹⁸F]F-FCH PET/CT combined with CE-MRI was higher than [¹⁸F]F-FCH PET/CT or CE-MRI alone (all p < 0.05). Interestingly, TLC (≥4.11) was significantly correlated with poor intracranial PFS (iPFS), meaning it was a significant prognostic factor for iPFS. Conclusions: This study identified that compared with CE-MRI, [¹⁸F]F-FCH PET/CT demonstrated higher specificity and accuracy in diagnosing recurrence of BM in LCBM after SRS. Combining [¹⁸F]F-FCH PET/CT with CE-MRI has the potential to improve diagnostic performance for recurrence of BM and management of patient treatment. TLC was an independent risk factor for iPFS. Full article

Background: Magnetic Resonance Imaging (MRI) provides rich tumor information through different imaging modalities (T1, T1ce, T2, and FLAIR). Each modality offers distinct contrast and tissue characteristics, which help in the more comprehensive identification and analysis of tumor lesions. However, in clinical practice, only a single modality of medical imaging is available due to various factors such as imaging equipment. The performance of existing methods is significantly hindered when handling incomplete modality data. Methods: A Teacher–Assistant–Student Collaborative and Competitive Net (TASCCNet) is proposed, which is based on traditional knowledge distillation techniques. First, a Multihead Mixture of Experts (MHMoE) module is developed with multiple experts and multiple gated networks to enhance information from fused modalities. Second, a competitive function is formulated to promote collaboration and competition between the student network and the teacher network. Additionally, we introduce an assistant module inspired by human visual mechanisms to provide supplementary structural knowledge, which enriches the information available to the student and facilitates a dynamic teacher–assistant collaboration. Results: The proposed model (TASCCNet) is evaluated on the BraTS 2018 and BraTS 2021 datasets and demonstrates robust performance even when only a single modality is available. Conclusions: TASCCNet successfully addresses the challenge of incomplete modality data in brain tumor segmentation by leveraging collaborative knowledge distillation and competitive learning mechanisms. Full article

Background and Objectives: Gliomas are characterized by high disability rates, frequent recurrence, and low survival rates, posing a significant threat to human health. Accurate grading of gliomas is crucial for treatment plan selection and prognostic assessment. Previous studies have primarily focused on the binary classification (i.e., high grade vs. low grade) of gliomas. In order to perform the four-grade (grades I, II, III, and IV) glioma classification preoperatively, we constructed an artificial neural network (ANN) model using magnetic resonance imaging data. Materials and Methods: We reviewed and included patients with gliomas who underwent preoperative MRI examinations. Radiomics features were derived from contrast-enhanced T1-weighted images (CE-T₁WI) using Pyradiomics and were selected based on their Spearman’s rank correlation with glioma grades. We developed an ANN model to classify the four pathological grades of glioma, assigning training and validation sets at a 3:1 ratio. A diagnostic confusion matrix was employed to demonstrate the model’s diagnostic performance intuitively. Results: Among the 362-patient cohort, the ANN model’s diagnostic performance plateaued after incorporating the first 19 of the 530 extracted radiomic features. At this point, the average overall diagnostic accuracy ratings for the training and validation sets were 91.28% and 87.04%, respectively, with corresponding coefficients of variation (CVs) of 0.0190 and 0.0272. The diagnostic accuracies for grades I, II, III, and IV in the training set were 91.9%, 89.9%, 92.1%, and 90.7%, respectively. The diagnostic accuracies for grades I, II, III, and IV in the validation set were 88.7%, 87.1%, 86.5%, and 86.9%, respectively. Conclusions: The MRI radiomics-based ANN model shows promising potential for the four-type classification of glioma grading, offering an objective and noninvasive method for more refined glioma grading. This model could aid in clinical decision making regarding the treatment of patients with various grades of gliomas. Full article

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related mortality worldwide, so its detection and monitoring are critical. However, contrast-enhanced magnetic resonance imaging (CE-MRI) is particularly vulnerable to complex, unstructured noise, which compromises image quality and diagnostic accuracy. This study proposes the use of NLE-ANSNet, a deep learning-based denoizing framework that integrates multilevel noise level estimators (NLEs) and adaptive noise scaling (ANS) within residual blocks. The model performs progressive, stagewise noise suppression at multiple feature depths, dynamically adjusting normalization based on localized noise estimates. This enables context-aware denoizing, preserving fine anatomical details. To simulate clinically realistic conditions, we developed a hybrid noise simulation framework that combines Gaussian, Poisson, and Rician noise at the pixel level. This framework aims to approximate a balanced noise distribution for evaluation purposes, with both mean and median noise levels reported to enhance evaluation robustness and prevent bias from extreme cases. NLE-ANSNet achieves a PSNR of 34.01 dB and an SSIM of 0.9393, surpassing those of state-of-the-art models. The method aims to support diagnostic reliability by preserving image structure and intensity fidelity in CE-MRI interpretation. In addition to quantitative analysis, a qualitative assessment was conducted to visually compare denoizing outputs across models, further demonstrating NLE-ANSNet’s superior ability to suppress noise while preserving diagnostically critical information. Unlike previous approaches, this study introduces a denoizing framework that combines multilevel noise estimation and adaptive noise scaling specifically tailored for CE-MRI in HCC under hybrid noise conditions—a clinically relevant and underexplored area. Overall, this study supports improved clinical decision making in HCC management. Full article

(1) Background: Neoadjuvant chemotherapy (NAC) is an integral part of breast cancer management, and response to NAC is an important prognostic factor associated with improved survival outcomes. However, the current standard for response assessment relies on post-surgical histopathological analysis, which limits early therapeutic decision-making and treatment personalization. This study aimed to develop and evaluate a machine learning model that integrates pre-treatment MRI radiomics and clinical features to predict response to NAC in breast cancer patients. (2) Methods: In this study, a machine learning model was developed to predict breast cancer response to NAC using pre-treatment magnetic resonance imaging (MRI) radiomics and clinical data. Radiomic features were extracted from contrast-enhanced T1-weighted (CE-T1) and T2-weighted (T2) MRI sequences using both intratumoral and peritumoral segmentations. Furthermore, this study uniquely examined two response assessment criteria: (1) pathologic complete response (pCR) versus non-pCR, and (2) clinical response versus non-response. A total of 254 patients with biopsy-confirmed breast cancer who completed NAC were included. Radiomic features (n = 400) and clinical features (n = 7) were analyzed to build a predictive model employing the XGBoost classifier. Performance was measured in terms of accuracy, precision, sensitivity, specificity, F1-score, and AUC. (3) Results: The integration of radiomic features with clinical data significantly enhanced the predictive performance. For pCR and non-pCR prediction, the combined features model achieved an accuracy of 80% and AUC of 0.85, outperforming both the clinical features model (Accuracy = 68%, AUC = 0.81) and radiomic features model (Accuracy = 66%, AUC = 0.60). Similarly, for the clinical response and non-response prediction, the combined features model achieved an Accuracy of 74% and AUC of 0.75, outperforming both the clinical features model (Accuracy = 63%, AUC = 0.68) and radiomic features model (Accuracy = 66%, AUC = 0.57). (4) Conclusions: These findings highlight the synergistic effect of integrating clinical data and MRI-based radiomics to improve pre-treatment NAC response prediction, which has the potential to enable more precise and personalized treatment strategies. Full article