Abstract Objectives Minimally invasive local therapies such as radiofrequency ablation (RFA), microwave ablation (MWA), and cryoablation (CA) are commonly used to treat malignant lung tumors. This study aimed to identify predictors of local tumor progression (LTP), focusing on the role of minimal ablative margin. Methods The medical records of patients ≥ 20 years old with primary or metastatic lung tumors who underwent percutaneous ablation from July 2011 to December 2022 were retrospectively reviewed. Patient demographic information, tumor characteristics, and ablation details were extracted. The primary outcome was LTP, and the key variable of interest was the minimal ablative margin. Results Seventy-one patients with 88 nodules ablated were included in the analysis, and 22% of the nodules exhibited LTP. In the overall cohort, ROC analysis identified 3 mm as the optimal cut-off for minimal ablative margin to predict LTP. Nodules with margins ≥ 3 mm were associated with significantly higher LTP-free survival at 5 years (82% vs. 23% margins < 3 mm). Kaplan-Meier analysis indicated a significant difference in LTP between ablative margins ≥ 3 and < 3 mm (p < 0.001). RFA was associated with a higher LTP risk than MWA (hazard ratio [HR] = 5.43, 95% confidence interval [CI]: 1.84-16.05), as was a margin < 3 mm (HR = 24.24, 95% CI: 7.66-76.71). Conclusions In this cohort, achieving an ablative margin of at least 3 mm was associated with improved LTP-free survival after lung tumor ablation. ADVANCES IN KNOWLEDGE This threshold should be considered a pragmatic, context-dependent estimate that may require validation across ablation modalities and clinical settings.
Abstract MRI plays a pivotal role in staging gynaecological malignancies owing to its superior soft tissue contrast and assessment of loco-regional disease. For accurate diagnosis and staging of gynaecological malignancies, meticulous image acquisition and skilled interpretation play a crucial role. Hormone-dependent physiological changes and associated benign lesions can pose challenges while diagnosing and/or staging malignancies. This may result in over- or under-staging of malignancies or errors in diagnosis. This pictorial essay aims to highlight the diagnostic challenges and pitfalls encountered while imaging various gynaecological malignancies on MRI. We discuss how advanced MRI sequences, such as diffusion-weighted imaging (DWI), Dynamic contrast-enhanced 3D sequences (DCE), and a comprehensive interpretation of all acquired sequences, aid in overcoming these challenges. Meticulous image acquisition, multiparametric imaging, including DWI and dynamic contrast-enhanced sequences, and knowledge of likely pitfalls and tumour mimics aid in improving diagnostic accuracy while reporting gynaecological malignancies.
Objectives To develop a multimodal model for predicting the invasiveness of ground-glass nodule (GGNs), enabling accurate differentiation of pre-invasive lesions (PIL), minimally invasive adenocarcinoma (MIA), and invasive adenocarcinoma (IA).Methods The retrospective study included patients with pathologically confirmed GGNs who underwent surgical resection between May 2018 and May 2025. Univariate and multivariate logistic regression analysis were performed to screen clinical parameters, follow-up information, and CT morphological features, identifying independent predictive factors. The intratumoral region (ITR) was segmented semi-automatically. The peritumoral regions (PTRs) were extracted via 2 scaling strategies: fixed-size expansion (3 mm, 6 mm, 9 mm, 12 mm) and tumor diameter-ratio expansion (25%, 50%, 75%, 100%). Radiomics features were extracted from both ITR and PTRs. Following feature selection using the Least Absolute Shrinkage and Selection Operator (LASSO) regression, classification models were constructed based on random forest (RF), support vector machine (SVM), and logistic regression (LR). Model performance was evaluated via receiver operating characteristic (ROC) curve analysis. Four progressive integration models were further created: conventional, conventional + follow-up, conventional + follow-up + ITR, and conventional + follow-up + ITR + PTR.Results A total of 517 GGNs were included in the training and internal test cohorts (63 PILs, 311 MIAs, and 143 IAs), with 91 GGNs from the external test cohort (3 PILs, 45 MIAs, 43 IAs). Maximum diameter, follow-up composite parameter, vacuole sign, and air bronchus sign were identified as independent predictive factors. Both PTR construction strategies yielded optimal models (PTR3/PTR6 for fixed-size, PTR50%/PTR75% for tumor diameter-ratio), demonstrating robust predictive efficacy. The RF model exhibited the highest stability. As feature dimensions expanded from conventional features to follow-up composite parameter, ITR, and PTR, the mean AUC of models exhibited a progressive increase. The full multimodal model (conventional + follow-up + ITR + PTR) achieved the best performance, with mean AUCs of 0.849 (training cohort), 0.795 (internal test cohort), and 0.838 (external test cohort).Conclusions The proposed multimodal predictive triple-classification model, integrating conventional, follow-up, ITR, and PTR, may act as a non-invasive tool for preoperative evaluation of GGNs invasiveness.Advances in knowledge The study first investigated PTR model via the tumor diameter-ratio method and compare it with the fixed-size method. A follow-up composite parameter (follow-up duration & times; follow-up changes status) was used to quantified GGN dynamic changes. The novel multimodal model, integrating conventional, follow-up, ITR, and PTR radiomics features, outperformed single-modal approaches in differentiating PIL, MIA, and IA.
Abstract Objectives Image-guided fluid aspiration is a key component in the diagnostic work-up of suspected prosthetic joint infection of the hip (PJI). Fluoroscopy, ultrasound (US), and computed tomography (CT) are frequently employed to guide aspiration; however, comparative evidence regarding their diagnostic performance and procedural outcomes remains limited. This study aimed to evaluate and compare the diagnostic accuracy and technical performance of these modalities. Methods A systematic review was conducted in accordance with PRISMA guidelines. PubMed, Embase, and CINAHL were searched for studies evaluating fluoroscopy-, US-, or CT-guided hip aspiration to aid PJI diagnosis. Outcomes of interest included diagnostic performance, dry tap rates, and procedural characteristics. Dry tap rates were pooled by modality using random-effects meta-analysis of proportions. Due to heterogeneity in reference standards, diagnostic accuracy outcomes were synthesised narratively. Results Ten studies comprising 1,574 hip aspirations were included (647 fluoroscopy, 831 US, 96 CT). Sensitivity ranged from 0.52–0.89 for US and 0.27–0.83 for fluoroscopy, with consistently high specificity (≥0.75) across modalities. CT-guided aspiration demonstrated sensitivity of 0.67 and specificity of 0.98 in a single study. Pooled dry tap rates were lower for US (17%, 95% CI 4–30%) compared with fluoroscopy (27%, 95% CI 9–45%), with substantial heterogeneity (I²>97%). Conclusions Reported specificity is comparable between ultrasound- and fluoroscopy-guided hip aspiration for PJI. Ultrasound-guided aspiration demonstrates higher reported sensitivity and lower dry tap rates, although interpretation is limited by heterogeneity in reference standards. Advances in knowledge This review provides a focused synthesis of comparative evidence on imaging modalities for hip aspiration in PJI. It highlights a potential diagnostic and practical advantage of ultrasound-guided aspiration.
Abstract Radiology has more regulatory-cleared artificial intelligence tools than any other medical specialty, yet they have delivered less consistent clinical impact than their benchmark performance suggested. We argue that the dominant response, scaling perceptual models and extending them toward autonomous agency, mistakes the nature of the problem. The limitation is architectural, not algorithmic: current systems perform isolated perception while remaining disconnected from the temporal, contextual, and experiential integration that interpretation requires. We trace this to a fundamental simulation gap, the absence of the counterfactual signal that lets AI reach superhuman performance in other domains, and argue that scaling alone cannot close it. In response, we outline a hybrid orchestration layer, grounded in established clinical frameworks, that assembles a comprehensive evidence base for the radiologist while keeping generative components constrained and interpretive judgment human. The shift is one of design paradigm: from isolated tools deployed alongside the radiologist to interpretive infrastructure that composes validated outputs into coherent evidence without anchoring the reader’s own assessment.
