Cancer Imaging最新文献

Background: To develop and validate a CT-based artificial intelligence (AI) score model integrating the R.E.N.A.L. nephrometry and contact surface area (CSA) for efficient, accurate prediction of perioperative outcomes in renal cell carcinoma (RCC) patients undergoing nephron-sparing surgery (NSS), addressing the subjectivity and inefficiency of manual score.

Methods: Retrospectively collected data from two NSS cohorts (n1 = 500, n2 = 50): 90% of cases in Cohort n1 (450 cases) were randomly assigned to the training set (315 cases), validation set (45 cases), and test set (90 cases) at a ratio of 7:1:2, which were used to develop and validate the automated kidney/tumor segmentation models, as well as to derive the AI-calculated R.E.N.A.L. score (with the "A" parameter excluded) and AI-calculated CSA score; the remaining 10% of cases in Cohort n1 (50 cases) were combined with all 50 cases in Cohort n2 to form a mixed validation set (100 cases), which was used for risk stratification prediction of NSS perioperative outcomes via AI scores. Manual image annotation/scoring was conducted by experienced radiologists and urologists. Interrater consistency was evaluated via weighted kappa coefficients; risk stratification was performed via Kruskal-Wallis tests and Mann-Whitney U tests.

Results: A total of 550 patients were included in this study (median age, 56 [IQR: 46-66] years; 341 males). The segmentation model exhibited excellent performance: Dice similarity coefficient (DSC) was 0.95 for kidneys and 0.80 for tumors; normalized surface distance (NSD) was 0.923 ± 0.082 and 0.892 ± 0.096, respectively; 95th percentile Hausdorff distance (HD95) was 9.78 ± 0.63 mm and 12.65 ± 0.84 mm, respectively. The R, E, N, L, R.E.N.A.L., and CSA score models had good consistency compared with the manual score, and the kappa coefficients were 0.82, 0.49, 0.63, 0.60, 0.65, and 0.69, respectively (all P < 0.01). Risk stratification by AI score significantly predicted warm ischemia time, surgical duration, intraoperative blood loss, serum creatinine changes, pathological T stage, and nuclear grade (all P < 0.05).

Conclusions: This study successfully developed a CT-based automated kidney/tumor segmentation model, and on this basis constructed the AI-R.E.N.A.L. and AI-CSA scoring models, providing an efficient and objective preoperative risk assessment tool for the perioperative outcomes of NSS.

Background: The current prediction of postoperative growth in synchronous nodules remaining after surgical resection of dominant lung tumors in patients with multiple subsolid lung nodules is limited. This study aims to assess the efficacy of preoperative CT-based radiomics in predicting the 5-year growth of these residual nodules (RNs), versus models constructed using commonly utilized CT morphological and quantitative features.

Methods: Data from 1392 patients who underwent resection for lung subsolid nodules confirmed as adenocarcinoma or precursor glandular lesions between 2014 and 2018 were retrospectively reviewed. Among the participants, 208 surgical patients with 603 RNs were included, with a follow-up period exceeding five years. Each RN was classified as either grown or stable based on CT imaging. All enrolled RNs were randomly allocated to training and testing sets at an approximately 4:1 ratio. Four models (radiomics, morphological, quantitative, and combined) were built separately by using Random Forest. Model performance was assessed using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analyses, and compared using DeLong test and net reclassification improvement (NRI).

Results: Patients harbored 1-26 RNs. 17.9% RNs grew in 5 years. Growth proportions varied by size: 4% for < 5 mm, 8.4% for 5-8 mm, and 48.5% for > 8 mm. Eighteen radiomics features, 5 morphological features, and 2 quantitative features were selected to build the respective models. The radiomics model showed a good ability to predict growth with an accuracy of 97.2% and 86.7% in the training and testing sets, respectively. The radiomics model showed a significantly higher area under the curve (AUC: 0.892) than the morphological model (AUC: 0.834, P < 0.05), an advantage over the quantitative model (AUC: 0.862, P = 0.251), and similarity to the combined model (AUC: 0.887) in the testing set. The radiomics model showed better reclassification than morphological (NRI = 7.4%; P = 0.017) and quantitative (NRI = 14%; P = 0.005) models in risk stratification. The calibration curves and decision curve analyses further confirmed the clinical value of radiomics.

Conclusions: CT-based radiomics demonstrated superior predictive performance for the 5-year growth of RNs, and can be used independently as a promising tool for future clinical guidance.

Objectives: To develop and validate an automated diagnostic framework that combines deep learning and radiomics models for the segmentation and classification of benign and malignant parotid gland tumors on magnetic resonance imaging (MRI).

Methods: In total, 493 patients with pathologically confirmed parotid tumors (396 benign and 97 malignant) were included. Patients were stratified by MRI scanner type into a training cohort (n = 288), an internal validation cohort (n = 123), and an external testing cohort (n = 82). An automated tumor segmentation model based on the nnU-NetV2 architecture was developed and evaluated using the Dice similarity coefficient (DSC) and Intersection over Union (IoU). Based on the automatically segmented regions, a radiomics-based classifier and a ResNet18-based deep learning model were independently constructed to differentiate malignant from benign tumors. A combined diagnostic model was further developed by integrating deep learning outputs, radiomics features, and clinical-radiological features. Model performance was assessed using the area under the receiver operating characteristic curve (AUC).

Results: The automated segmentation model achieved a Dice similarity coefficient (DSC) of 0.93 and an Intersection over Union (IoU) of 0.88 in the training cohort, 0.91 and 0.84 in the validation cohort, and 0.84 and 0.76 in the testing cohort, respectively. The ResNet18-based DL model achieved AUCs of 0.90, 0.84, and 0.77, respectively, compared to the radiomics model's AUCs of 0.79, 0.72, and 0.71. The combined model demonstrated superior performance, with AUCs of 0.92 in the validation cohort and 0.90 in the testing cohort, outperforming the clinical-radiological model, which achieved AUCs of 0.69 and 0.82 (p < 0.001 in validation, p = 0.042 in testing).

Conclusions: This automated MRI-based framework, combining deep learning and radiomics approaches, enables accurate segmentation and reliable classification of parotid gland tumors. It offers a promising noninvasive tool to assist in clinical decision-making.

Clinical trial number: Not applicable.

Objective: This study aims to systematically evaluate the global distribution, academic impact, and technological trends of ultrasound technology in lung cancer research from 2000 to 2024 through bibliometric analysis, providing references for future research directions.

Materials and methods: Based on the Web of Science Core Collection database, 2,617 publications from 2000 to 2024 were included. Bibliometric analysis was conducted using VOSviewer and CiteSpace, covering publication trends, countries, institutions, authors, journals, highly cited literature, and keyword co-occurrence networks. Metrics such as publication volume, citation count, and total link strength (TLS) were quantitatively assessed.

Results: From 2000 to 2024, the annual number of publications on ultrasound in lung cancer diagnosis and treatment surged from 19 to 218, with the last five years accounting for 38% of the total, indicating a continuous increase in research activity. The United States (554 publications) and China (387 publications) contributed 48% of the global output. The U.S. led in total citations (24,413), while Germany demonstrated superior research quality with an average of 57 citations per publication. Chiba University (Japan, 48 publications), the National Cancer Center Japan, and Shanghai Jiao Tong University School of Medicine (each with 46 publications) were the most productive institutions. The international collaboration network formed a multi-center cluster with the U.S. as the core. Japanese scholar Kazuhiro Yasufuku (56 publications/3,320 citations) had the highest influence, and highly cited authors generally relied on strong international collaborations (TLS ≥ 76). Chest (86 publications/10,664 citations) was the most influential core journal in the field. Citation analysis revealed that endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) technology and lung cancer staging guidelines (highest cited: 225 times) were key research focuses. Keyword clustering identified four major research directions: bronchoscopic diagnosis, minimally invasive biopsy, comprehensive treatment, and imaging staging. Burst detection highlighted emerging hotspots such as "safety" (strength = 15.13), "yield" (17.91), "peripheral pulmonary lesions" (12.98), and "radial endobronchial ultrasound" (12.72), indicating a shift from traditional diagnosis toward efficiency quantification, precision navigation (e.g., radial ultrasound), and intelligent analysis.

Conclusion: From 2000 to 2024, the academic influence of ultrasound in lung cancer research significantly increased, with research hotspots evolving from technical standardization to precise staging and multimodal applications. The integration of AI and multimodal technology represents a core future direction.