Search Results (688)

Objectives: Our aim was to develop an optimal model for predicting distant metastasis (DM) in rectal cancer (RC) via integrating tumor lymphocytic reaction (LR) with deep learning radiomic (DLR) features. Methods: A total of 190 patients with RC were divided into a training cohort (n = 133) and a validation cohort (n = 57) in a 7:3 ratio. Preoperative radiomics (Rad), deep transfer learning (DTL), and DLR features were extracted from T₂WI scans, and clinicopathological variables were collected. All models were constructed using the support vector machine (SVM) algorithm, and their performance was evaluated using the area under the receiver operating characteristic curve (AUC). A 3-year follow-up was conducted to analyze 3-year distant-metastasis-free survival (DMFS) outcomes and DM risk. Results: Multivariate analysis confirmed LR as an independent predictor of 3-year DMFS (p < 0.05). The fusion model integrating LR and DLR exhibited preliminary predictive performance, with AUC values of 0.911 and 0.884 in the training and validation cohorts, respectively. Subgroup observational analyses showed apparent DMFS differences associated with adjuvant chemotherapy exposure among low-risk patients, while such survival patterns were not evident in the high-risk subgroup defined by the nomogram cut-off of 0.3. A moderate negative correlation was detected between DLR signatures and LR (r = −0.44, p < 0.001), providing preliminary immune-related clues for interpreting deep learning radiomic biomarkers. Conclusions: The multimodal fusion nomogram combining pathological LR and DLR signatures shows encouraging preliminary predictive performance for 3-year DMFS risk in patients with RC. This postoperative multimodal tool may provide a preliminary reference for clinicians to implement individualized postoperative surveillance for patients with RC, and the inverse association between DLR and LR helps reveal the immune-related background of imaging signatures. Full article

Artificial intelligence (AI) is increasingly being investigated in prostate cancer (PCa) diagnosis and characterization, offering novel approaches to improve detection and risk stratification. This narrative review summarizes current evidence regarding the application of AI across the major stages of PCa management, with particular emphasis on radiomics and pathomics. Radiomics enables the extraction of high-dimensional quantitative features from medical imaging modalities, including ultrasound, computed tomography, multiparametric magnetic resonance imaging (mpMRI), and prostate-specific membrane antigen positron emission tomography (PSMA PET), providing imaging biomarkers that extend beyond conventional visual interpretation. Numerous studies have demonstrated that AI-based radiomic models improve the detection of clinically significant PCa, characterize tumor aggressiveness, predict extracapsular extension, and support individualized treatment selection. Among available imaging modalities, mpMRI remains the cornerstone for radiomics owing to its superior soft-tissue characterization, whereas PSMA PET radiomics has shown particular promise for assessing biologically aggressive disease and metastatic spread. Pathomics has further expanded the role of AI by enabling automated tumor detection, grading, quantification, and identification of adverse pathological features, with promising performance reported in selected retrospective validation studies. Despite encouraging results, widespread clinical implementation remains limited by heterogeneous imaging protocols, variability in data acquisition and annotation, lack of standardized workflows, insufficient prospective multicenter validation, and ethical and regulatory challenges. The aim of this review was to summarize current evidence on AI-based imaging analysis, radiomics, and pathomics for PCa detection, characterization, risk stratification, and pathological assessment, while highlighting the methodological challenges that currently limit clinical implementation. Full article

Ryanodine receptor 2 (RyR2) has been increasingly recognized as an important regulator in cancer biology. Previous studies demonstrate that RyR2 expression is associated with patient survival, although its precise role in tumorigenesis remains unclear. Elevated RyR2 expression has been linked to poorer survival outcomes, while gene silencing or pharmacological inhibition (e.g., with S107) has been shown to reduce cancer cell metastasis in both in vitro and in vivo models. Cytokeratin 18 (CK18), a key epithelial marker, is associated with tumor differentiation, hormonal phenotype, and clinical behavior in pituitary neuroendocrine tumors (PitNETs), reflecting underlying cytoskeletal organization. Variations in CK18 expression have been linked to distinct adenoma subtypes and may provide insight into tumor aggressiveness. This exploratory study investigated associations between RyR2 expression, CK18 status, and conventional MRI findings of pituitary tumors. Quantitative T2-weighted signal intensity and radiomic features were not assessed. This exploratory observational study used prospective patient recruitment between January 2017 and January 2026, after ethics approval was obtained in December 2016. RyR2 expression in immunohistological samples was quantified using QuPath v0.6.0. CK18 immunostaining was performed with the Dako Omnis system according to the manufacturer’s protocols. Conventional MRI findings were available for 24 patients and were retrospectively reviewed. Statistical analysis was conducted using SPSS/W 31.0. No statistically significant associations were observed between RyR2 expression and sex, hormonal activity, tumor size, invasiveness, recurrence, or the evaluated conventional MRI findings (all p > 0.05). The only statistically significant finding was higher RyR2 expression in CK18-negative than in CK18-positive tumors (median (IQR): 2.49% (18.36%) vs. 0.06% (0.39%); p = 0.006; r = 0.47). Higher RyR2 expression was observed in CK18-negative tumors. Given the exploratory design and limited sample size, this association should be considered hypothesis-generating and requires validation in larger independent cohorts. Full article

Magnetic resonance imaging (MRI)-based prediction of multiple sclerosis (MS) progression depends on how imaging data are prepared, represented, modelled, and evaluated. This systematic review synthesized artificial intelligence (AI) methods across the MRI-to-prediction pipeline. PubMed, Scopus, Web of Science, and Google Scholar were searched for studies published from 2010 to March 2026. Eligible studies included patients with MS or clinically isolated syndrome, used brain MRI directly or as the source of predictors, evaluated future progression-related outcomes, and reported quantitative predictive performance. Thirty-seven studies were included and synthesized narratively because of methodological and outcome heterogeneity. Prediction targets included CIS-to-MS/CDMS conversion, future MRI disease activity, PIRA/PIRMA, disability worsening or confirmed progression, future EDSS or disability status, and RRMS-to-SPMS conversion. Lesion morphology and location, radiomics, tissue atrophy, longitudinal imaging changes, and multimodal variables provided useful prognostic information. Classical machine learning (ML) remained effective for structured biomarkers, while deep learning (DL) enabled direct image-based modelling; neither approach was uniformly superior. PROBAST + AI assessment rated model Development as High concern in 35 of 37 studies and Unclear in two, Evaluation as High risk of bias in all 37 studies, and Applicability as Low concern in 32, Unclear in two, and High in three. Overall, MRI-based AI shows promising prognostic potential, while stronger data separation, calibration, transparent reporting, and independent validation are needed to support reproducible clinical translation. Full article

Purpose: To develop an exploratory prognostic model integrating multiparametric MRI radiomics and clinical features for recurrence risk stratification in young-age breast cancer (YABC). Materials and Methods: In this prospective single-institution cohort, women under 40 years with invasive breast cancer were enrolled between March 2017 and August 2019 and followed for at least 5 years. Tumor and contralateral fibroglandular tissue were segmented on multiparametric breast MRI using a deep learning-assisted model. Radiomic features were extracted and reduced via principal component analysis, and then integrated with clinical variables. Recurrence-free survival (RFS) was compared between risk groups using Kaplan–Meier analysis and the log-rank test. Internal assessment used 1000 bootstrap resamples to evaluate selection stability and optimism-corrected performance. Results: Fifty women (mean age, 34.8 ± 3.6 years) were included, and 11 patients (22%) had recurrence during a median follow-up of 64.5 months. Triple-negative subtype and lower T1_cancer3 showed strong exploratory associations, whereas lower ADC_FGT4 and larger pathologic tumor size showed weaker associations. Risk groups differed in RFS (log-rank p = 0.0022). Bootstrap analysis showed moderate selection stability, with preserved discrimination but limited calibration stability after optimism correction. Conclusions: MRI radiomics combined with clinical factors may provide complementary prognostic information for recurrence risk stratification in YABC, warranting validation in larger independent cohorts. Full article

Introduction: MRI is the gold-standard imaging modality for rectal cancer (RC) local staging, but the ability to determine tumor invasion (pT) and nodal status (pN) remains limited in clinical practice. The main objective of this study is to evaluate the performance of different machine learning (ML) models based on clinical–radiological and radiomic variables for predicting these categories from preoperative MRI. Methods: A retrospective observational study was conducted involving 152 patients with RC (70 without neoadjuvant therapy and 82 with neoadjuvant therapy). Radiomic features were extracted from high-resolution T2 sequences using two independent segmentations: tumor and tumor + mesorectum. Twenty-three ML algorithms were evaluated using cross-validation to predict pT and pN. For each combination of outcome, cohort, data source and segmentation, an optimal model was selected based on the area under the curve (AUC). Results: Models based on clinical and radiological variables showed the most consistent performance, particularly in the overall cohort, with AUCs of 0.767 for pT and 0.764 for pN. The radiomic and combined models achieved a moderate and heterogeneous performance, with maximum AUCs of 0.770 for pT and 0.732 for pN. Conclusions: The clinico-radiological variables analyzed using ML showed a predictive performance similar to that of a radiologist. Radiomics did not show significant improvement in this setting. Full article

Background/Objectives: Small enhancing lesions (often termed foci if lesion ≤ 5 mm) observed on breast magnetic resonance imaging (MRI) pose a persistent challenge. Their clinical relevance and the most effective management strategy remain unclear, often leading to unnecessary biopsies. We sought to uncover whether quantitative radiomic features of these lesions could reliably discriminate malignant ones from benign ones. Methods: In this single-center retrospective study (2015–2024), we analyzed 31 contrast-enhancing breast lesions with a maximum diameter ≤ 10 mm measured on a single axial MRI slice from 30 patients who underwent 1.5-T or 3.0-T breast MRI followed by histologic confirmation (10 malignant, 21 benign). Lesions were manually delineated, and 105 radiomic variables were extracted from early-phase dynamic contrast-enhanced (DCE) subtraction images. Feature importance was quantified with a Random Forest model; iterative top-k pruning found the highest-performing variables. A multilayer perceptron (MLP) classifier was trained and validated using a leave-one-subject-out cross-validation (LOSO-CV) scheme, with the sensitivity, specificity, accuracy, and AUC as the primary performance metric. Results: Of the 105 extracted features, 44 carried predictive information (feature importance score ≥ 0.20). Progressive feature reduction yielded an optimal subset of nine radiomic features (six texture-based and three shape-based). Texture-derived features predominated among the most informative variables. The MLP achieved a sensitivity of 0.80, specificity of 0.89, accuracy of 0.86, and AUC of 0.82. Conclusions: In this exploratory study, a preliminary nine-feature radiomic signature extracted from early-phase DCE-subtraction images has the potential to distinguish benign from malignant small breast lesions when used with an MLP model. Prospective validation is needed before the model can influence clinical decision-making. Full article

Background/Objectives: Tetralogy of Fallot (TOF) is the most common cyanotic congenital heart defect, and cardiac computed tomography (CT) is increasingly central to its anatomical and pre-procedural assessment. Artificial-intelligence research in TOF is dominated by MRI; deep learning on cardiac CT in congenital heart disease exists but addresses multi-class diagnosis and segmentation, and the one binary TOF-versus-control CT study used slice-level validation without confounder control, and, to our knowledge, no CT study reports controlling the confounding intrinsic to a TOF-versus-control comparison. This confounding is structural: TOF is imaged predominantly in infancy, so a naive classifier can learn age, body size, and acquisition protocol rather than pathology. We develop and internally evaluate a confounder-matched, anatomy-guided deep-learning pipeline for TOF on cardiac CT. Methods: Contrast-enhanced cardiac CT from a single scanner was de-identified and restricted to one reconstruction (FC15 kernel, 0.5 mm), then matched 1:1 on age and sex, yielding 42 TOF and 42 controls (n = 84); controls were children imaged for suspected but excluded cardiovascular disease, so scan indication, unlike age and sex, was not matched. Standardized volumes were decomposed into four fixed sub-volumes positioned to approximate the components of the diagnostic tetrad: malalignment ventricular septal defect (VSD), overriding aorta, right-ventricular outflow tract (RVOT), and right-ventricular hypertrophy (RVH). Whether each sub-volume contains its named target was audited against independent physician region-of-interest annotations. Per region, a 2.5D transfer-learning classifier (ImageNet ResNet18) and a 3D CNN (DenseNet121) were trained with leak-free patient-level five-fold cross-validation and the branches fused. Optimism was assessed by repeated cross-validation and, for model selection, by nested cross-validation with the component subset and operating point chosen inside an inner loop. Discrimination was reported with bootstrap 95% confidence intervals (CIs); AUROCs were compared by DeLong test, with Benjamini–Hochberg correction applied to a seven-member family (the four within-component comparisons, two hybrid-versus-VSD contrasts, and hybrid versus whole-heart) and other comparisons reported uncorrected. Results: Matching removed the age difference (median 0.33 years, IQR 0.17–0.92 vs. 0.33, IQR 0.27–0.73; p = 0.86) with balanced sex (p = 1.00). The pre-specified four-component hybrid reached AUROC 0.829 (95% CI 0.74–0.91); the VSD region alone reached 0.828 (0.74–0.91), so the tetrad decomposition did not improve accuracy, and the containment audit shows it does not deliver the intended anatomical interpretability either. The 2.5D model exceeded the 3D CNN for every component (0.769–0.828 vs. 0.573–0.656; raw DeLong p = 0.007–0.037, Benjamini–Hochberg q up to 0.065 under a seven-member family, the weakest comparison (RVH) not surviving correction). Repeated cross-validation gave 0.811 ± 0.026 and nested cross-validation 0.787 ± 0.029; a stronger backbone with multi-phase data, handcrafted radiomics, and a large CT foundation model did not significantly improve on the matched pipeline. Grad-CAM maps were sensitive to both model weights and labels and superior to a centred-blob null in all eight comparisons and significantly so in seven, but not consistently superior to a resolution-matched random attribution, so no localization claim is made. Calibration was imperfect (slope 0.67) and recalibration gave no net gain; at an in-sample Youden threshold sensitivity was 0.93 and specificity 0.64. Occlusion sensitivity on the whole-heart baseline model showed it relies on the physician-marked septal, aortic and right-ventricular sites 1.8–4.1 times more than distance-matched surrounding tissue, while gross morphometry alone reached 0.651–0.663. Two of the four sub-volumes did not contain their target: the RVOT prior contained the physician annotation in 48.1% of cases and, because the model samples only the central band, excluded it in 99.4%; the RVH box was offset toward the midline, containing the marked target in 43.4% of annotations. Repositioning the priors, leak-free and derived from controls only, did not change discrimination (all p ≥ 0.10), and boxes placed at random positions inside the standardized heart reached 0.765 on average against 0.796 for the published priors, a difference this cohort cannot resolve. Conclusions: As a proof of concept, confounder-matched deep learning can recognize TOF on cardiac CT. Increasing model capacity did not significantly improve on the matched pipeline; the separate contribution of matching itself was not isolated against an unmatched comparator. Given the small, single-centre sample and the absence of external validation, these findings are hypothesis-generating and require external, multi-centre confirmation before any clinical use. Full article

Background/Objectives: Patients with suspicious prostate mpMRI but negative baseline biopsy remain challenging to manage, as some are later diagnosed with clinically significant prostate cancer (csPCa). This study evaluated whether baseline whole-gland radiomic features from T2-weighted imaging and apparent diffusion coefficient (ADC) maps could predict subsequent csPCa detection. Methods: This retrospective radiomics sub-analysis of the MULTIPROS trial included men with suspicious baseline mpMRI findings (PI-RADS v2.0 categories 3–5) and no csPCa detected at baseline biopsy. The primary outcome was subsequent csPCa detection during follow-up, defined as ISUP Grade Group ≥2. Whole-gland prostate segmentation was performed on baseline T2-weighted images using a semi-automated approach with manual refinement, and the resulting T2-derived masks were applied to the spatially corresponding ADC maps. Radiomic features were extracted using PyRadiomics version 3.1.0a2 after standardised preprocessing. Feature robustness was assessed in a 20-patient reproducibility subset using intra- and inter-reader intraclass correlation coefficients, followed by correlation filtering and LASSO-based feature prioritisation. Radiomics-only, clinicoradiological-only, and clinicoradiological–radiomic logistic regression models were evaluated using stratified five-fold cross-validation. Results: T2-based modelling included 151 patients, of whom 36/151 (23.8%) had subsequent csPCa detected; the median time to csPCa detection or last follow-up was 64 months. ADC analyses included 144 patients, of whom 34/144 (23.6%) had subsequent csPCa detected. The T2 clinicoradiological–radiomic model achieved a mean AUC of 0.696 ± 0.060, compared with 0.713 ± 0.072 for the T2 clinicoradiological-only model and 0.456 ± 0.091 for the T2 radiomics-only model. In the ADC-available cohort, ADC radiomics-only, ADC clinicoradiological–radiomic, and combined T2 + ADC clinicoradiological–radiomic models achieved mean AUCs of 0.684 ± 0.052, 0.635 ± 0.157, and 0.663 ± 0.106, respectively. Calibration was imperfect, with deviations at higher predicted probabilities and overprediction in the highest predicted-risk bin. Conclusions: Baseline whole-gland MRI radiomics showed limited and inconsistent predictive value for subsequent csPCa detection after negative biopsy and did not demonstrate consistent incremental improvement over clinicoradiological predictors. These findings are exploratory and hypothesis-generating, and external validation is required before clinical use. Full article

Background/Objectives: Bone metastasis is one of the most common manifestations of advanced malignancy and a major cause of morbidity, particularly when involving the spine. Magnetic resonance imaging (MRI) plays a central role in its detection and characterisation due to its high sensitivity for bone marrow infiltration. However, bone metastases may be missed on MRI, whilst interpretation can be time-consuming and challenging. The purpose of this study is to review and summarise the present evidence for artificial intelligence (AI) applications in the detection and classification of bone metastasis on MRI. Methods: A systematic, detailed search of the main electronic medical databases (PubMed, MEDLINE, Web of Science, and clinicaltrials.gov, last accessed on 1 January 2026) was undertaken in concordance with the PRISMA guidelines. Results: A total of 34 studies were included. AI applications were identified across several domains, including lesion detection, segmentation, disease classification, and predictive modelling. Deep learning approaches demonstrated strong performance for automated detection and segmentation, while radiomics-based models were frequently used for lesion differentiation and prediction tasks. Reported performance metrics were generally high, with area under the curve values commonly ranging from approximately 0.72–0.94, with most studies reporting AUCs exceeding 0.80 in internal validation, although substantial heterogeneity in study design, datasets, and validation strategies was observed. External validation and prospective evaluation were limited across most studies. Conclusions: Within the domain of bone metastasis, AI-based approaches have demonstrated encouraging performance and hold substantial potential to support clinical decision-making, including prognostication and prediction of treatment response. Nevertheless, further research is required to validate their clinical utility and to facilitate successful integration into routine clinical practice. Full article

Laryngeal pathology requires individualized management because the larynx integrates airway protection, phonation, swallowing, and respiratory function within a compact and highly variable anatomical framework. This narrative review examines how laryngeal anatomy and morphometry support interdisciplinary collaboration and personalized care in oncologic, stenotic, functional, and reconstructive laryngeal disease. Evidence from morphometric studies, CT, MRI, endoscopy, ultrasonography, three-dimensional reconstruction, artificial intelligence, and multidisciplinary clinical workflows was synthesized qualitatively. Key parameters—including vocal fold length, glottic width, thyroid cartilage angle, cricoid diameter, subglottic diameter, anterior commissure thickness, and the status of paraglottic and pre-epiglottic spaces—provide actionable information for diagnosis, T-staging, airway assessment, surgical planning, reconstruction, and functional rehabilitation. Morphometry concretely informs clinical decisions: thyroid cartilage angle guides thyroplasty and phonosurgical planning; subglottic diameter supports stenosis surgery and airway instrumentation; and anterior commissure, conus elasticus, cartilage, and deep-space measurements refine oncologic staging and margin strategy. Technological accelerators, including AI segmentation, radiomics, 3D printing, photogrammetry, and ultrasonography, extend morphometry from static measurement toward predictive modeling and patient-specific simulation. However, implementation remains limited by heterogeneous CT protocols, inconsistent measurement planes, uneven access to advanced technologies, lack of global normative databases, and unresolved ethical issues surrounding AI validation and data governance. This review supports standardizing CT morphometry using parallel vocal fold planes, routinely measuring anterior commissure thickness in T1 glottic cancer, incorporating ultrasonography as a first-line morphometric tool in voice clinics, and validating AI segmentation against population-specific morphometric norms. Laryngeal morphometry should therefore become a routine decision-making framework for precision laryngology. Full article

Background: Breast cancer brain metastases (BCBMs) exhibit notable receptor discordance between primary tumors and metastases, yet obtaining intracranial biopsies is rarely practical. We created an interpretable deep learning radiogenomic model to predict estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status from MRI scans in BCBM cases. Methods: A total of 241 post-contrast T1-weighted MRIs from 142 patients were analyzed. We developed a mask-free 3D Residual Neural Network (ResNet) ensemble trained directly on cropped, bias-corrected brain images, with comprehensive 3D geometric and intensity augmentations. The model was optimized in a multi-label setting using Asymmetric Focal Loss. Hyperparameters were fine-tuned via Bayesian optimization, and class probabilities were calibrated. Additionally, Integrated Gradients (IG) offered voxel-level saliency maps. Results: Our ResNet ensemble model achieved a micro-averaged AUROC of 0.74 and a macro-averaged AUROC of 0.67. Receptor-specific AUROCs were 0.57 for ER, 0.79 for PR, and 0.64 for HER2. The F1-scores were 0.56, 0.62, and 0.88 for ER, PR, and HER2, respectively. The held-out cohort contained only four HER2-negative patients, so threshold-dependent HER2 metrics are strongly prevalence-driven and AUROC is the more appropriate summary. Qualitatively inspected saliency maps were concentrated on enhancing metastases with limited background attribution. Conclusions: This proof-of-concept study shows that a segmentation-free, interpretable 3D Convolutional Neural Network (CNN) can be trained to capture receptor-associated patterns in post-contrast T1-weighted MRI of BCBMs. Accuracy was modest relative to published radiomics models, and the mask-free, single-sequence design is offered as a methodological contribution rather than a performance gain. These findings are preliminary, do not establish clinical utility, and require validation in larger, multi-center cohorts. Full article

Osteonecrosis of the femoral head (ONFH) is a progressive and potentially disabling condition caused by compromised blood supply to the femoral head, leading to bone necrosis and collapse. Early diagnosis is essential to enable joint-preserving interventions and improve patient outcomes. Magnetic resonance imaging (MRI) is currently considered the most sensitive modality for early detection, whereas computed tomography (CT) provides superior assessment of subchondral bone integrity and structural collapse. In recent years, artificial intelligence (AI), particularly machine learning (ML) and deep learning (DL) techniques, has emerged as a promising tool to enhance diagnostic accuracy, automate lesion segmentation, and predict disease progression. This review aims to provide an overview of current AI applications in MRI for ONFH, focusing on early diagnostic, disease staging and classification, volumetric assessment, differential diagnosis and prognostic prediction. A total of 61 articles were initially identified, of which 13 studies (2021–2025) met the inclusion criteria. Results indicate that DL models, particularly convolutional neural networks (CNNs), achieve excellent diagnostic performance, with reported accuracies up to 98.4% and area under the curve (AUC) values reaching 0.98 for early-stage detection. Several models demonstrated performance comparable to or exceeding that of experienced clinicians, particularly in differentiating ONFH from other hip pathologies and in early disease recognition. AI algorithms also showed high accuracy in staging and classification (AUC up to 99.7% in internal validation), as well as in automated segmentation and volumetric assessment (Dice coefficients up to 0.89), enabling objective quantification of necrotic lesions. Furthermore, prognostic models integrating radiomics and ML techniques demonstrated promising results in predicting femoral head collapse (AUC up to 0.85). From a clinical perspective, AI appears to function primarily as a supportive tool, improving diagnostic consistency, efficiency, and reproducibility, and acting as a “second reader” capable of reducing variability among less experienced clinicians. However, significant limitations remain, including dataset heterogeneity, predominance of retrospective and monocentric studies, and limited integration of clinical data. In conclusion, AI-based MRI analysis shows strong potential to enhance the diagnosis, staging, and management of ONFH. Future research should focus on multicenter prospective validation, integration of multimodal clinical data, and development of explainable and generalizable models to facilitate widespread clinical adoption. Full article

Diffusion-weighted imaging (DWI) has evolved from a niche research sequence into an increasingly valuable adjunct to conventional magnetic resonance imaging (MRI) in musculoskeletal (MSK) radiology. By providing qualitative and quantitative information on tissue microstructure through assessment of water diffusion and apparent diffusion coefficient (ADC) mapping, DWI offers functional insights beyond conventional morphological imaging. We aim to present the current evidence for DWI in MSK imaging organised around established applications and emerging applications, with particular emphasis on composition-related interpretive pitfalls relevant to differentiating tumours and other pathologies, and to review the technique’s evolving role in routine practice. This narrative review synthesises the current literature on the clinical utility of DWI in MSK imaging. It is structured in four parts: foundations and the tissue composition signal framework, including the basis of qualitative and quantitative assessment; established applications; emerging applications; and assessment of tissue composition-related interpretive as well as technical pitfalls, including those arising due to myxoid matrix, chondroid matrix, blood degradation products, organising thrombus, crystalline or mineralised material, keratinaceous debris, purulent content, cellular haematopoietic marrow, by using original cases from the authors’ institution, which have been confirmed either histologically or surgically. Applications are stratified by strength of evidence. Established applications of DWI include soft tissue abscess detection, differentiation of malignant from benign soft tissue tumours, differentiation of malignant from benign vertebral compression fractures, and myeloma staging and response assessment, as well as treatment response in soft tissue and bone sarcomas. Whole-body MRI with DWI for staging and response assessment in multiple myeloma is guideline-endorsed and supported by prospective multicentre data. Soft tissue abscess detection, soft tissue and bone tumour characterisation, and characterisation of vertebral compression fractures are supported by consistent evidence from multiple independent cohorts, although no universally transferable ADC threshold exists. The emerging applications, which are promising adjuncts supported by small, single-centre or heterogeneous studies with thresholds that have not been externally validated, include ADC ghost sign in osteomyelitis (high specificity but sensitivity of only 20%), peripheral nerve sheath tumour characterisation and surveillance in NF1 patients, peripheral neuropathy and plexopathy, predisposing conditions such as Li Fraumeni syndrome in paediatric cancers, inflammatory myopathy, and postsurgical assessment of residual disease, as well as opportunistic detection of venous thrombosis. Radiomics and machine learning approaches remain experimental. Recent technical advances, including reduced field-of-view imaging, multi-shot acquisition and improved fat suppression, have mitigated but not eliminated historical limitations of susceptibility artefacts and limited spatial resolution. DWI has become an important functional imaging technique that complements conventional MRI across a broad range of musculoskeletal disorders. Understanding the relationship between tissue composition and the diffusion signal is central to both interpreting DWI correctly and avoiding its characteristic pitfalls. DWI is best regarded not as a stand-alone technique but as one component of a multiparametric assessment, in which its functional information is integrated with conventional morphological imaging. Ongoing technical improvement and expanding clinical evidence are expected to further support its integration into routine MSK imaging and its development as a quantitative biomarker for diagnosis, prognostication, and treatment monitoring. Full article

Background/Objectives: In skull-base proton therapy, T2-weighted brain magnetic resonance (MR) imaging may be acquired with or without fat saturation (FS). The present study investigated the impact of this protocol variation on radiomics feature stability and brain radionecrosis (BRN) prediction. Methods: Paired T2-weighted FS and non-FS follow-up MR scans of proton-treated skull-base chordoma patients (n = 52) were used to assess feature stability. For BRN prediction (CTCAEv5 grade ≥ 1), baseline planning scans of chordoma and chondrosarcoma patients (n = 80) with mixed FS protocols were used. Following automated brain tissue segmentation, 1911 radiomics features were extracted from cerebrospinal fluid, gray, and/or white matter using PyRadiomics (v3.1.0). Feature stability was quantified using Lin’s concordance correlation coefficient (CCC). Several MR image- and feature-level processing configurations were explored, and the stability of combined gray and white matter features served as the reference for selecting the optimal configuration. Features were stratified based on CCC thresholds and were subjected to univariable feature selection using a Mann–Whitney U-test. Logistic regression was used for predictive modeling and its performance was measured using micro-averaged AUC within a repeated stratified cross-validation framework. Results: Radiomics features were highly sensitive to FS variations (median CCC range: 0.21–0.58). Combat harmonization without image-level processing was selected as the optimal configuration, under which 30% of features achieved CCC ≥ 0.70, and 10.7% demonstrated a CCC ≥ 0.85. Restricting modeling to a highly stable subset eliminated nearly 90% of the baseline features while significantly improving predictive performance (ΔAUC = 0.05, p < 0.001). Conclusions: A subset of radiomics features robust to FS variations was identified. Use of these stable features may mitigate the impact of protocol-induced variability in mixed-FS T2-weighted MR data while simultaneously improving BRN prediction performance. These findings are preliminary and should be interpreted cautiously, given the pilot nature of the study. Full article