Radiology. Imaging cancer最新文献

Purpose To evaluate the technical feasibility and safety of a nonsurgical interventional method for placement and retrieval of implantable microdevices (IMDs) in a rabbit tumor model. Materials and Methods This prospective preclinical feasibility and safety study was conducted from March 2022 to October 2024. Interventional IMD placement and retrieval were performed in 12 rabbits with single hindlimb VX2 tumors. Four or five IMDs were placed per animal. Each IMD delivered microdose quantities of three drugs (doxorubicin, topotecan, sunitinib) into 18 spatially distinct microscopic tumor regions over 24-48 hours. An over-the-wire biopsy procedure was used to extract the IMDs and surrounding tumor. Technical success was defined per animal as retrieval of at least one tumor region containing each of the three drugs. Overall region-level retrieval success rate was calculated as the overall percentage of drug-exposed regions retrieved successfully. Safety was determined by monitoring animals for adverse events. Procedure durations were assessed to inform clinical translation. Statistical analysis included calculation of binomial 95% CIs for success rates and summary statistics (mean ± SD) for region yield and procedure times. Results Twelve rabbits received 52 IMDs (mean, 4.3 ± 0.5 per animal) generating 936 drug-exposure sites. Technical success was 100% (12 of 12; 95% CI: 73.5, 100). Mean per animal retrieval was 50.7 regions ± 11.4 (range, 33-69), with an overall region-level success rate of 65% (608 of 936; 95% CI: 62, 68). No Common Terminology Criteria for Adverse Events grade 2 (moderate) or higher adverse events were observed. In two animals, IMDs dislodged from the guidewire and were retained in the tumor; no symptoms developed, and no additional intervention was required (grade 1). Implantation procedure duration was 30.3 minutes ± 5.3 and retrieval procedure duration was 49.1 minutes ± 6.3. Conclusion Interventional IMD placement and retrieval in a rabbit soft tissue tumor model was technically feasible and safe. This approach demonstrated the capacity to recover multiple drug-exposed regions for in situ assessment of local drug effects. Keywords: Implantable Microdevice, Personalized Treatment, Percutaneous Biopsy, Phase 0 Trials, Microdosing, Interventional Radiology Supplemental material is available for this article. © RSNA, 2025.

Purpose To develop and validate a deep learning model for automatic segmentation of primary central nervous system lymphoma (PCNSL) at postcontrast T1-weighted MRI. Materials and Methods Data were retrospectively collected from patients with pathologically proven immunocompetent PCNSL between September 2010 and February 2022. Postcontrast T1-weighted MRI scans were used to train and validate a deep learning model based on the nnU-Net framework. Manual segmentation by neuroradiologists served as the reference standard. The model was trained using an internal dataset from a single center and tested on both internal and external test sets from seven additional centers. Performance was assessed using Dice score, mean average surface distance, and F1 score. Statistical comparisons were performed using Mann-Whitney U test and bootstrap resampling for CIs. Results The study included 135 patients (68 female, 66 male, and one of unspecified sex; internal dataset: mean age ±SD, 67.0 years ± 12.0; external dataset: mean age, 75.5 years ± 13.6). The model achieved a mean Dice score of 0.84 (95% CI: 0.79, 0.88) on the internal test set (n = 44) and 0.88 (95% CI: 0.84, 0.91) on the external test set (n = 48), with no evidence of a difference between test sets (P = .59). Performance varied by lesion type; accuracy was highest in homogeneous discrete lesions, and performance was slightly decreased when numerous poorly defined infracentimetric lesions occurred. Strong volumetric correlation was observed between automatic and manual segmentations (internal: r = 0.99, P < .001; external: r = 0.98, P < .001). Conclusion A deep learning model achieved accurate and robust automatic segmentation of PCNSL across multiple clinical centers with different MRI acquisition parameters. Keywords: Brain Lymphoma, Brain Tumor, Automatic Segmentation, Artificial Intelligence, Deep Learning Supplemental material is available for this article. © RSNA, 2025.

Purpose To develop and test a machine learning (ML)-based model that integrates preoperative variables for prediction of advanced-stage progression (ASP) after transarterial chemoembolization (TACE). Materials and Methods This multicenter retrospective study (ResearchRegistry.com identifier no. researchregistry9425) included patients with intermediate-stage hepatocellular carcinoma (HCC) who underwent TACE at seven hospitals from June 2008 to December 2022. Thirty-four preoperative clinical and CT imaging variables were input into six ML-based models for prediction of ASP, and model performances were compared. Furthermore, the best-performing ML model was compared with the major staging systems, and its utility in performing post-TACE therapies was assessed. The performances of the models were compared by using area under the receiver operating characteristic curve (AUC) with DeLong test. Kaplan-Meier survival curves were compared using the log-rank test. Results A total of 2333 eligible patients (mean age, 54 years ± 12 [SD]; 2051 male patients) were categorized into the training set (n = 1026), the internal test set (n = 257), and the external test set (n = 1050). ASP was found in 8.4% (86 of 1026), 8.2% (21 of 257), and 6.7% (70 of 1050) of patients in the three datasets, respectively. Among all ML models, the Categorical Gradient Boosting (CatBoost) model yielded the highest AUC: 0.97 (95% CI: 0.95, >0.99) for the training set, 0.94 (95% CI: 0.92, 0.97) for the internal test set, and 0.93 (95% CI: 0.90, 0.95) for the external test set. Furthermore, it yielded better discriminatory ability with higher concordance indexes than the five staging systems (all P < .001). The time-dependent AUC of the CatBoost model was also higher than that of the clinical staging systems at various time points (all P < .001). Moreover, post-TACE systemic therapy improved progression-free survival and overall survival for patients in the high-risk group (both P < .001) but not in the low-risk group. Conclusion The CatBoost model demonstrated higher predictive performance compared with existing staging systems in predicting ASP after TACE in patients with intermediate-stage HCC. This model effectively stratified patients by risk level and identified those who benefited from post-TACE systemic therapy. Keywords: Liver, Oncology, Transarterial Chemoembolization, Hepatocellular Carcinoma, Advanced-stage Progression, Machine Learning, Risk Differentiation ResearchRegistry.com identifier no. researchregistry9425 Supplemental material is available for this article. © RSNA, 2025 See also commentary by Rouzbahani in this issue.

Purpose To evaluate glutamate (Glu) and glutamine (Gln) concentrations in patients with glioma using 7-T MR spectroscopic imaging, identify significant differences in metabolic ratios between tumor and peritumoral regions, and assess associations of Glu and Gln with tumor-associated epilepsy and other tumor characteristics. Materials and Methods This retrospective study included data from patients with gliomas who underwent 7-T MR spectroscopic imaging in a single university hospital between September 2018 and April 2021. Median values for nine metabolic ratios were calculated within the visible tumor and peritumoral shell, and Dice similarity coefficients were used to assess the spatial overlap of elevated metabolic regions between these compartments. Statistical significance between regions of interest and between glioma attributes (eg, isocitrate dehydrogenase status) was assessed. Results Thirty-six patients (median age, 52 years [IQR, 23 years]; 22 male, 14 female) were included in the study. The Glu to total creatine (Glu/tCr) median was significantly higher in the peritumoral volume of interest (median, 1.13) compared with the tumor (median, 0.92; P = .00015) and normal-appearing white matter (NAWM; median, 0.87; P < .00011), while the Gln/tCr median was highest in the tumor (median, 0.77; peritumoral: median, 0.44; P < .00011; NAWM: median, 0.33; P < .00011). Glu to total choline was higher in the peritumoral region as well (median, 3.44; tumoral: median, 2.23; P < .00011; NAWM: median, 2.06; P < .00011). Peritumoral Dice similarity coefficients for Glu/tCr and Gln/tCr hotspots were comparable (0.51 to 0.53). Specific metabolic ratios were significantly different between isocitrate dehydrogenase mutant and wild-type gliomas (eg, tumoral Glu/total N-acetylaspartate [tNAA], P = .0054), oligodendroglioma and astrocytoma (eg, tumoral Gln/tNAA, P = .0033), and oligodendroglioma and glioblastoma (eg, tumoral Glu/tNAA, P = .0034). Conclusion The 7-T MR spectroscopic imaging revealed increased Glu and Gln metabolic ratios within the peritumoral region compared with NAWM of patients with glioma distinct from intratumoral changes. Keywords: Glioma, 7 T, MR Spectroscopic Imaging, MRSI, Infiltration, Iisocitrate Dehydrogenase, IDH Supplemental material is available for this article. © The Author(s) 2025. Published by the Radiological Society of North America under a CC BY 4.0 license.

Purpose To compare the predictive accuracy of subjective versus objective assessment of three-dimensional functional drained liver volume (fDLV) for clinical success after percutaneous transhepatic biliary drainage (PTBD). Materials and Methods This retrospective study included patients with malignant biliary obstruction and hyperbilirubinemia who underwent de novo PTBD between January 2016 and February 2017. Clinical success was defined as achieving total bilirubin level less than 1.8 mg/dL (30.8 μmol/L) within 30 days after PTBD. Seven interventional radiologists independently categorized subjective fDLV into four groups: less than 30%, 30%-50%, 51%-75%, and greater than 75% of the total liver volume. Objective fDLV was calculated using imaging software and expressed as volume, percentage, and categorical groups. Interobserver agreement was assessed using the intraclass correlation coefficient (ICC). Predictive accuracy was analyzed via univariate and multivariate logistic regression and receiver operating characteristic curves. Results This study included 35 consecutive patients (median [IQR], 66 years [57, 71]; 23 men). Clinical success was achieved in 17 of 35 patients (49%). Interobserver agreement for subjective fDLV was low (ICC, 0.29 [95% CI: 0.12, 0.48]). Objective fDLV expressed as volume (odds ratio [OR], 1.34 [95% CI: 1.13, 1.70]; P = .004), percentage (OR, 1.06 [95% CI: 1.02, 1.10]; P = .002), and category (OR, 3.46 [95% CI: 1.64, 9.68]; P = .005) were associated with clinical success; subjective consensus category was not (OR, 2.08 [95% CI: 0.94, 4.99]; P = .081). Objective fDLV percentage demonstrated higher predictive accuracy (area under the receiver operating characteristic curve [AUC], 0.84 [95% CI: 0.69, 0.99]) than individual subjective estimates (AUC range, 0.56-0.74) and group consensus (AUC, 0.67 [95% CI: 0.49, 0.84]). A threshold of at least 76.1% objective fDLV yielded a sensitivity of 88% and specificity of 83% and remained independently associated with clinical success (OR, 1.10; P = .013). Conclusion Objective fDLV assessment outperformed subjective evaluation and more accurately helped predict short-term clinical success after PTBD. Keywords: Percutaneous Transhepatic Biliary Drainage, Functional Drained Liver Volume, Hyperbilirubinemia, Malignant Biliary Obstruction, Objective Measurement © RSNA, 2025.