Insights into Imaging最新文献

Objectives: This study aims to establish a radiological model derived from preoperative computed tomography (CT) to predict the likelihood of papillary renal cell carcinoma (PRCC) recurrence after surgical intervention.

Materials and methods: A retrospective multicenter study initially enrolled 384 patients, with 266 eligible for analysis from four centers following partial nephrectomy or radical resection for PRCC. Twelve distinct categories of CT features were evaluated. To assess reproducibility, interobserver variability in radiological assessment was evaluated. A Cox proportional hazards model was employed to identify significant radiological predictors and construct a risk score system. The model's performance was evaluated through Harrell's Concordance Index (C-index), and its effectiveness was compared with that of several histopathologic prognostic systems.

Results: A total of 266 patients were included, comprising a training dataset from one center (n = 152) and an external validation dataset from three other centers (n = 114). Inter-reader agreement was moderate to excellent for the radiological parameters (k = 0.43-0.94). Tumor margin regularity and regional lymph node size on CT scans were found to be independently associated with tumor recurrence (subdistribution hazard ratios ranging from 5.34 to 28.11; p-values ranging from < 0.001 to 0.028) and were incorporated into the predictive model. The model demonstrated superior predictive accuracy for tumor recurrence in the validation set compared to existing prognostic systems (C-index: 0.95 vs. 0.74-0.92; p-values ranging from < 0.001 to 0.08).

Conclusion: A radiological score that combines tumor margin regularity and regional lymph node size predicts PRCC recurrence, demonstrating superior performance compared to existing prognostic systems.

Critical relevance statement: This CT-based scoring system outperforms existing models in prognostic accuracy, aiding clinicians in personalized risk stratification and optimizing treatment decisions for patients.

Key points: The preoperative CT features are associated with the prognosis of papillary renal cell carcinoma (PRCC). Tumor irregularity and lymph node size on CT scans independently predict the postoperative recurrence of PRCC. A CT scoring system that incorporates these two features demonstrates superior prognostic accuracy compared to existing models.

Objectives: This study aimed to explore a multiple-instance learning (MIL) framework incorporating radiomics features and deep learning representations to predict the invasiveness of ground-glass nodules (GGNs) in lung adenocarcinoma (LUAD) using preoperative CT.

Materials and methods: We retrospectively analyzed 1247 GGNs from 1182 LUAD patients across six hospitals, and divided them into training, validation and three test sets. According to postoperative pathological findings, the data were further classified into invasive and non-invasive subgroups. Five kinds of predictive models were developed: radiomics models, 3D deep learning models, 2.5D deep learning models, deep learning-based MIL (MIL-DL) models, and deep learning and radiomics-based MIL (MIL-DL-Rad) models. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis (DCA).

Results: The MIL-DL-Rad model with the ExtraTrees classifier exhibited superior and consistent performance across all sets, achieving AUCs of 0.936, 0.881, 0.868, 0.926, and 0.918 in training, validation and external test sets. In contrast, the AUC performance of MIL-DL and radiomics models was relatively unstable. The calibration curve and DCA indicated that the integrated model achieved favorable predictive efficiency and clinical predictive benefits.

Conclusions: The MIL-DL-Rad model showed better overall performance for invasiveness prediction of GGNs in LUAD patients, providing a novel perspective on feature fusion that can contribute to more accurate preoperative predictions in clinical practice.

Critical relevance statement: Multi-instance learning integrating deep learning and radiomics enhances the prediction of ground-glass nodule (GGN) invasiveness and is expected to provide optimal preoperative clinical decision-making for lung adenocarcinoma patients.

Key points: Ground-glass nodules invasiveness directly influences surgical strategies and prognosis. Multiple-instance learning framework integrates radiomics and deep learning features. Integrated model achieves superior accuracy and consistency in invasiveness prediction.

Objective: To evaluate the diagnostic performance of combined transscrotal-transrectal ultrasonography in predicting sites of obstructive azoospermia.

Materials and methods: From June 2019 to March 2023, 166 obstructive azoospermia patients who underwent surgical exploration were enrolled in the retrospective study. The data of combined transscrotal-transrectal ultrasonography in 166 patients were collected and analyzed. The receiver operating characteristic (ROC) curve analysis was employed to evaluate the diagnostic performance of these ultrasonographic measurements for localizing different obstructive sites.

Results: There were 9 sides of intratesticular obstruction, 239 sides of epididymal obstruction, 68 sides of vas deferens obstruction, and 16 sides of ejaculatory duct obstruction. The sensitivity, specificity, and the area under the curve (AUC) for combined transscrotal-transrectal ultrasonography were 44.4%, 98.5% and 0.714 for diagnosing intratesticular obstruction; 97.9%, 84.9% and 0.919 for diagnosing epididymal obstruction; 82.4%, 99.2% and 0.913 for diagnosing vas deferens obstruction; and 87.5%, 99.1% and 0.93 for diagnosing ejaculatory duct obstruction. The sensitivity, specificity, and AUC were 88.9%, 83.9% and 0.842 in diagnosing intratesticular obstruction for a rete testis thickness cut-off of 3.0 mm; 81.0%, 100% and 0.949 in diagnosing vas deferens obstruction for a 0.8 mm cutoff for the internal diameter of the scrotal section of the vas deferens; and 62.5%, 92.6% and 0.769 in diagnosing ejaculatory duct obstruction for a seminal vesicle diameter cut-off of 12.5 mm.

Conclusion: Combined transscrotal-transrectal ultrasonography, evaluating specific structures of rete testis thickness, seminal vesicle diameter, and the internal diameter of the scrotal vas deferens, could accurately localize obstruction sites in obstructive azoospermia patients.

Critical relevance statement: Combined transscrotal-transrectal ultrasonography demonstrated high diagnostic performance in predicting the sites of epididymal, vas deferens, and ejaculatory duct obstruction in patients with obstructive azoospermia.

Key points: The diagnostic performance of combined transscrotal-transrectal ultrasonography in obstructive azoospermia was evaluated. Ultrasound measurements of specific structures significantly improve the prediction of obstruction sites. Combined transscrotal-transrectal ultrasonography accurately localizes obstruction sites in obstructive azoospermia patients.

Objective: To develop a comprehensive preoperative risk stratification model using habitat imaging combined with multimodal analysis for identifying low-risk papillary thyroid carcinoma (PTC) patients suitable for active surveillance.

Materials and methods: This multicenter study analyzed 1215 patients with pathologically confirmed PTC from four Chinese institutions. Habitat imaging analysis was performed on preoperative CT and ultrasound images using K-means clustering and supervoxel segmentation. Radiomic features were extracted from ultrasound habitats using PyRadiomics, while multi-scale index (MSI) features were extracted from CT habitats. Clinical characteristics and immunological markers were identified through multivariate logistic regression. Six machine learning algorithms were evaluated with three fusion strategies to integrate imaging features with clinical data.

Results: Four ultrasound habitats and five CT habitats were identified. Ultrasound Habitat 2 achieved an AUC of 0.92 in training and 0.80-0.92 in validation. CT habitat analysis using MSI features achieved an AUC of 0.93 in training and 0.88-0.92 in validation. The optimal ensemble fusion model integrating CT-derived MSI features, ultrasound habitat characteristics, clinical parameters (chronic lymphocytic thyroiditis and tumor size) and immunological markers (platelet-to-lymphocyte ratio) achieved an AUC of 0.98 in training, 0.95 in internal validation, and 0.95-0.99 across external validation cohorts, with accuracy exceeding 0.88 in all validation sets.

Conclusion: Habitat imaging combined with multimodal analysis provides superior preoperative risk stratification for PTC, enabling personalized treatment planning and identification of low-risk patients suitable for active surveillance while potentially reducing unnecessary surgical interventions.

Critical relevance statement: Habitat imaging combined with multimodal analysis provides superior preoperative risk stratification for papillary thyroid carcinoma, enabling personalized treatment decisions and reducing unnecessary surgical interventions.

Key points: Current papillary thyroid carcinoma (PTC) risk stratification relies on postoperative pathology, limiting preoperative treatment planning. Multimodal habitat imaging achieved exceptional performance across validation cohorts. This framework enables personalized treatment planning and identifies low-risk patients for active surveillance.

Objective: This study explored the incidence of adverse events (AEs) following microwave ablation (MWA) under local anesthesia and analyzed factors related to benign uterine diseases, including uterine fibroids (UFs) and adenomyosis (AM).

Materials and methods: Overall, 366 patients who underwent percutaneous MWA were finally included in this study. Univariate and multivariate logistic regression analyses were performed to identify the main factors affecting AEs.

Results: The overall AEs rate for benign uterine disease was 77.32% (283/366), and was significantly higher in patients with AM than in those with UFs (95.38% vs. 73.42%, p < 0.001). AM (odds ratio (OR) = 3.77, p = 0.039) and higher transformed symptom severity score (higher tSSS) (25-40: OR = 2.98, p = 0.007; > 40: OR = 2.36, p = 0.022) were independent risk factors for AEs. In the subgroup analysis of patients with UFs, moderate-to-severe pain during MWA was significantly associated with AE occurrence (OR = 2.35, p = 0.048) and abdominal pain (OR = 3.63, p < 0.001). Although multivariate regression analysis showed that higher tSSS (25-40: OR = 3.22, p = 0.003; > 40: OR = 3.32, p = 0.001) was an independent influencing factor for vaginal discharge, univariate analysis suggested that vaginal discharge risk also increased with FIGO 0-3 (OR = 2.53, p = 0.010).

Conclusion: Our results demonstrated that AM and higher tSSS were identified as significant independent risk factors, facilitating better patient selection and improved patient counseling. Moderate-to-severe pain during MWA was strongly associated with AE occurrence, highlighting the need for further investigation of anesthesia optimization. Further, patients with FIGO 0-3 fibroids exhibited a higher risk of postoperative vaginal discharge, necessitating procedural refinement to preserve endometrial integrity.

Critical relevance statement: Our study makes a significant contribution to the literature because it provides a comprehensive analysis of microwave ablation-related adverse events and their associated risk factors, facilitating better patient selection, procedural refinements, and improved patient counseling.

Key points: This study addresses a critical gap in the literature by investigating the safety of ultrasound-guided microwave ablation (MWA) for uterine fibroids (UFs) and adenomyosis (AM) under local anesthesia. Our results demonstrated the overall AE rate for UFs and AM following MWA was 77.32%, with AM and higher transformed symptom severity scores identified as significant independent risk factors. Given the differences in AE risk between UFs and AM, as well as related risk factors, tailored treatment protocols should be considered to optimize outcomes.

Introduction: Deep-learning methodologies for predicting early response in locally advanced nasopharyngeal carcinoma (LA-NPC) remain unvalidated, with techniques like 2.5-D imaging and Transformers underexplored.

Materials and methods: MRI images from LA-NPC patients diagnosed between January 2020 and March 2024 at two centers were analyzed. Patients (n = 184) were split into training (n = 89), validation (n = 39), and test (n = 56) sets. Three segmentation models-SegResNet, Unet, and UnetR-automatically delineated regions of interest (ROIs). A 2.5D approach integrated adjacent tumor sections into a transfer learning framework, leading to three predictive models: Clinical, Transformer, and Combined. Performance was assessed using ROC curves, calibration curves, and decision curve analysis (DCA).

Results: The Transformer model outperformed others, achieving AUCs of 0.968, 0.957, and 0.830 for the training, validation, and test sets, respectively. The Clinical model had lower AUCs (0.898, 0.759, 0.658). The Combined model, integrating clinical data, matched or exceeded Transformer performance, particularly in the test set (AUC = 0.874).

Conclusion: The Combined model, leveraging Transformer architecture and clinical factors, demonstrates strong efficacy in predicting early response in LA-NPC patients undergoing chemoradiotherapy, suggesting its potential for improved personalized treatment.

Critical relevance statement: This study critically validates a novel 2.5-D radiomic-Transformer fusion model that improves early response prediction for locally advanced nasopharyngeal carcinoma, directly advancing personalized chemoradiotherapy planning in clinical radiology.

Key points: Early treatment response prediction in locally advanced nasopharyngeal carcinoma lacks validated deep learning models using 2.5D imaging and Transformers. Transformer-based model achieved superior predictive performance compared to clinical or combined models. Integrating clinical data with Transformer imaging analysis improves personalized chemoradiotherapy outcome prediction.

Objective: Thyroid capsule invasion (TCI) predicts early progression in papillary thyroid microcarcinoma (PTMC). This study aimed to develop an integrated model that combines handcrafted peri-tumoral radiomics features with deep learning (DL)-derived intra-tumoral features for accurate early prediction of TCI, to support clinical decision-making.

Materials and methods: Retrospective data from 964 patients with 964 pathologically confirmed PTMC lesions across three centers were collected. Radiomics features were extracted from multiple peri-tumoral regions, and the optimal peri-tumor region with the best radiomics features was selected using a support vector machine (SVM). The selected radiomics features were then combined with intra-tumoral DL features extracted from the tumors before being fed into four different DL models for training and validation. Performance was validated on the internal (n = 177) and external (n = 84) test sets. Six radiologists (senior/attending/junior) assessed TCI with/without DL assistance.

Results: The radiomics features, which achieved the best diagnostic performance with an AUC of 0.795 using SVM, were extracted from the peri-tumor region with 30% expansion from the original tumor. By further combining these radiomics features with intra-tumoral DL features, four different DL models were established to identify TCI in PTMC. Swin-Transformer achieved superior performance (internal AUC: 0.923; external AUC: 0.892). With DL model assistance, the AUCs of six radiologists significantly improved, for example, from 0.720 to 0.796 and from 0.725 to 0.790 for senior radiologists, and similar gains were observed for attending and junior radiologists.

Conclusions: As an effective clinical assistive tool,  this integrated model can provide TCI identification with high level of accuracy. With its ability to enhance radiologists' diagnostic performance, it supports early PTMC risk stratification and personalized intervention.

Critical relevance statement: This retrospective multicenter study establishes an integrated model for identifying TCI in PTMC. The model significantly enhances radiologists' diagnostic precision across multiple experience levels, supporting early clinical decision-making for optimized intervention strategies.

Key points: Accurate prediction of TCI facilitates early assessment of PTMC progression and guides subsequent individualized clinical management. DL significantly improves the predictive performance for TCI. DL effectively assists radiologists in TCI diagnosis.

The ESR equity, diversity and inclusivity (EDI) subcommittee is a part of the Young ESR committee created in 2024. This statement paper is the first in our series regarding EDI and radiology. In this paper, we examine and discuss issues which have been studied and reported regarding the inequity of imaging services. Inequity is prevalent in radiology and imaging circles in Europe. The variations observed in women, ethnic, age, disabled, non-binary, and gender groups are examined, as well as the variations in radiology research and in artificial intelligence-related imaging. Radiology departments need to be aware of the existing variations in radiology services. They need to educate their personnel on the etiquette and interaction with diverse populations. There should be versatile equipment to serve patients with disabilities. Radiologists should be aware of the lack of evidence-based knowledge with regard to female and non-white populations. Regarding clinical AI, departments need to actively audit and check for possible biases in AI in clinical use. CRITICAL RELEVANCE STATEMENT: Understanding how EDI affects patient care is vital to providing equitable service to all patients. Radiologists should be aware of the lack of evidence-based knowledge regarding female and non-white populations, and be sensibly critical of guidelines which lack proper evidence. KEY POINTS: The workflow of the department should be organised so that all patients are served equitably. Radiologists need to be aware of the lack of evidence-based knowledge about female and non-white populations, and be critical of guidelines which lack proper evidence. Regarding AI, radiologists must actively audit and check for possible biases in AI in clinical use.

Contrast-enhanced mammography (CEM)-guided biopsy enables the tissue sampling of enhancing breast lesions that are not visible on conventional imaging. The technique combines dual-energy stereotactic acquisition with intravenous contrast administration, allowing accurate targeting of recombined-only lesions. It represents a practical alternative to MRI-guided biopsy, particularly in settings where MRI access is limited or contraindicated. This review examines current evidence, procedural experience, and challenges associated with CEM guidance. Early studies support its technical feasibility, although data remain scarce and heterogeneous regarding lesion selection, procedural experience, and outcome definitions. Broader implementation is challenged by equipment specifications, contrast administration practices, logistics, and reimbursement issues. As clinical adoption increases, structured patient triage pathways, standardised protocols, and prospective validation are essential. CEM-guided biopsy is a promising technique in breast imaging and has the potential to reduce reliance on MRI guidance. However, further research is required to define its role and ensure consistent performance across clinical settings. CRITICAL RELEVANCE STATEMENT: This review critically examines current evidence, technical feasibility, and implementation challenges of contrast-enhanced mammography-guided biopsy. It highlights potential advantages for clinical settings where MRI guidance is limited, while addressing existing limitations and areas that require further research. KEY POINTS: Contrast-enhanced mammography-guided biopsy is a dual-energy stereotactic procedure that enables the targeting of enhancing lesions that lack conventional imaging correlates. The modality is accurate and feasible, though its implementation is challenged by technical heterogeneity and the absence of standardised protocols. Broader clinical adoption requires structured diagnostic workflows, validated contrast administration strategies, and prospective multicentre evaluation.