European Radiology最新文献

Objective: Despite advances in mammography screening, some cancers remain undetected, prompting the evaluation of artificial intelligence (AI) as an independent third reader to reduce missed cancers.

Materials and methods: In this prospective study, women eligible for the German Mammography Screening were enrolled at six sites belonging to one screening unit between August 2023 and February 2024. Each mammogram underwent double reading and was independently analyzed using Transpara, an AI-based detection software. Cases rated BI-RADS 4 or 5 by any reader or given a risk score of 10 by the software were reviewed in a consensus conference. Endpoints included: primary-cancer detection rate (CDR) and positive predictive values (PPV); secondary-analysis of cancers detected only by the software or missed by it.

Results: 15,356 female participants (mean age 58.6 ± 5.6 years) were included. Overall, 115 breast cancers were detected (CDR triple reading: 0.75%; 95% CI: 0.62%, 0.90%). CDR of double reading and standalone AI was 0.68% (95% CI: 0.56, 0.83%) and 0.66% (95% CI: 0.54, 0.81%). Using Transpara as a third reader increased the detection rate by 9.5% (95% CI: 4.7%, 16.8%) compared to double reading (p = 0.002). The PPV for consensus-conference referrals was 5.1% (95% CI: 4.2%, 6.1%), lower than double reading 7.5%(95% CI: 6.2%, 9.0%; p < 0.001). For recalled cases, the PPV was 13.7%(95% CI: 11.5%, 16.2%) versus 15.2% (95% CI: 12.6%, 18.1%; p < 0.001). All nine invasive cancers detected solely by AI were Luminal-A-like cancers. Among 13 cancers missed by the software, four were triple-negative.

Conclusion: Adding Transpara as an independent third reader improved detection rates, mainly by identifying additional Luminal-A-like cancers, and increased the workload to the consensus conference and the number of recalled cases.

Key points: Question Does the integration of AI software as an independent third reader improve cancer detection rates in mammography screening without increasing false-positive findings and recall rates? Findings AI as an independent third reader increased cancer detection by 9.5%, mainly identifying Luminal-A-like cancers, significantly decreasing the positive predictive values of cases referred to at the consensus conference and increasing the number of recalled cases. Clinical relevance Using AI as an independent third reader enhances mammographic cancer detection by offering radiologists complementary sensitivity, especially for low-risk lesions. However, maintaining human readers is essential, as AI may miss aggressive subtypes like triple-negative breast cancers.

Objectives: Early recognition of individuals at elevated risk for new ipsilateral ischemic lesions (NIILs) after carotid artery stenting (CAS) is vital for planning effective preventive interventions. The aim of this study was to develop a deep learning (DL) radiomics model to predict NIILs post-CAS from dual-energy CT (DECT) images.

Materials and methods: This study retrospectively enrolled patients from three centers. Carotid plaques were delineated on multiparametric DECT images. A combined model integrating clinical-radiological, handcrafted radiomics (HCR), and DL features was constructed using a support vector machine algorithm to predict NIILs. The model's performance was assessed through the area under the receiver operating characteristic curve (AUC). To improve the interpretability of the model, SHapley Additive exPlanations (SHAP) analysis was applied.

Results: This study involved 336 patients divided into the training (n = 135), internal validation (n = 58), and external test (n = 143) cohorts. NIILs were present in 38.5%, 37.9%, and 39.9% of the subjects, respectively. Symptomatic events and plaque ulceration were identified as independent risk factors for NIILs. The combined model incorporating 2 clinical-radiological risk factors, 9 HCR features, and 15 DL features demonstrated satisfactory performance in predicting NIILs, with AUCs of 0.908, 0.842, and 0.856 in the three cohorts, respectively. The predictions of the combined model were explained both locally and globally by SHAP analysis.

Conclusion: The combined model demonstrated high accuracy in identifying patients at elevated risk for NIILs post-CAS and can serve as an interpretable tool for optimizing treatment strategies.

Key points: Question Early prediction of new ipsilateral ischemic lesions (NIILs) after carotid artery stenting (CAS) is crucial for timely interventions, but no effective, interpretable predictive method exists. Findings The combined model incorporating deep learning radiomics features extracted from multiparametric dual-energy CT images and clinical-radiological features demonstrated high accuracy in predicting NIILs after CAS. Clinical relevance The combined model offers an interpretable tool for identifying patients at high risk for NIILs post-CAS, potentially improving personalized treatment strategies and patient outcomes by enabling targeted preventive care.

Objective: Obstructive hydrocephalus is a critical radiographic finding requiring emergent treatment. Its identification on head CT by an AI model could facilitate sooner life-saving interventions, although there are common co-occurring findings, including intracranial hemorrhage, that can confound this interpretation. This external validation assessed the accuracy of an AI model at identifying obstructive hydrocephalus, including in the presence or absence of other findings.

Materials and methods: This retrospective cohort included 177 thin (≤ 1.5 mm) series and 194 thick (> 1.5 and ≤ 5 mm) series from 200 non-contrast head CT cases. These cases were obtained from patients aged ≥ 18 years at 5 hospitals in the United States. Each case was interpreted independently by up to three neuroradiologists. Each series was then interpreted by the AI model.

Results: The AI model performed with an area under the curve of 0.988 (95% confidence interval (CI): 0.971-0.998) on thin series and 0.986 (95% CI: 0.969-0.997) on thick series. These results were broadly maintained in subgroups for the presence or absence of intracranial hemorrhage, parenchymal abnormality, and ventricular drain, and across demographic and scanner manufacturer subgroups.

Conclusions: The AI model accurately identified obstructive hydrocephalus in this dataset. Its performance in subgroup analyses reflected its robustness.

Key points: Question Can an artificial intelligence model accurately identify obstructive hydrocephalus on head computed tomography, including in the presence or absence of common co-occurring imaging findings? Findings This model accurately identified obstructive hydrocephalus on thin and thick series, including in the presence or absence of intracranial hemorrhage, parenchymal abnormality, and ventricular drain. Clinical relevance This model could assist with triaging abnormal cases, enabling earlier identification and management of obstructive hydrocephalus. Its maintained performance with or without co-occurring findings suggests it specifically identifies obstructive hydrocephalus rather than these findings.

Objectives: Microvascular invasion is a strong prognostic factor in hepatocellular carcinomas. The aim of our study was to assess the diagnostic value of mechanical parameters measured with compression MR elastography to detect microvascular invasion in hepatocellular carcinomas.

Materials and methods: In this prospective preoperative MR elastographic study, consecutive patients with hepatocellular carcinomas, scheduled for tumor surgical resection, were included. The tumor parameters assessed with MR elastography were the basal visco-elastic parameters (storage modulus, loss modulus, and phase angle, reflecting elasticity, viscosity and visco-elastic ratio) during expiration and inspiration, and the tumor stiffening slope during compression induced by respiration, reflecting non-linear elasticity. Microvascular invasion was determined with histopathological examination of resected tumors. Diagnostic performance of MR elastography was assessed with area under the receiver operating curve (AUC) analysis.

Results: The final study group consisted of 53 patients with complete surgical resection, MR elastography and histological data, including 31 patients with microvascular invasion. Compression stiffening slope and storage modulus difference between inspiration and expiration were significantly higher in hepatocellular carcinomas without than with microvascular invasion (p < 0.001 and p = 0.03, respectively). Among clinical, morphological and biomechanical imaging features, the MR elastography compression stiffening slope (p = 0.004) and histological WHO differentiation (p = 0.02-0.03) were the only independent determinants of hepatocellular carcinoma microvascular invasion. In contrast to basal biomechanical parameters, the compression stiffening slope had high diagnostic performance for detecting microvascular invasion (AUC_{compression stiffening} = 0.83, p < 0.001).

Conclusion: Our results suggest that the compression stiffening slope at MR elastography is useful to diagnose microvascular invasion in patients with hepatocellular carcinomas.

Key points: Question Because non-invasive imaging markers of hepatocellular microvascular invasion are lacking, the development of new MRI markers is advisable. Findings In our MR elastography study, respiration-induced tumor stiffening, in contrast to basal visco-elastic parameters, had good accuracy for diagnosing hepatocellular carcinoma microvascular invasion. Clinical relevance Our results in patients with hepatocellular carcinomas suggest that the non-invasive measurement of MR elastography tumor compression stiffening slope may assess microvascular invasion.

Objectives: This study developed an automated AI-based method for accurate image reconstruction, stenosis detection and plaque calculation in high-resolution magnetic resonance vessel wall imaging (HR-MRVWI) and compared its performance with radiologists.

Materials and methods: A deep learning algorithm trained on HR-MRVWI was collected retrospectively from three tertiary hospitals. An independent test set was collected prospectively at another hospital. Model performance was evaluated via the Dice similarity coefficient, average centerline distance and average surface distance in centerline extraction and vessel wall segmentation. Two radiologists reviewed the reconstructed images in randomized order to determine whether the quality matched the clinical diagnosis. The stenosis diagnosis and plaque calculation of the algorithm were compared with the ground truth of the consensus by two radiologists. The relationships of the calculated parameters with plaque vulnerability were also analyzed.

Results: 476 patients (mean age 61 years ± 15 [SD], 286 men) were evaluated. The accuracy of image reconstruction in the independent test set was 92.3%. The consistency between the radiologists and the deep learning-assisted algorithm for stenosis detection was 0.89 (95% CI: 85.4, 90.2) in ≥ 50% stenosis. The accuracies of algorithm in normalized wall index, eccentricity and remodeling indices were 0.94, 0.83 and 0.87. The normalized wall index was highly related to plaque vulnerability. The AI-assisted in diagnosis and vessel wall analysis, which reduced the time from 32.0 ± 11.8 to 12.9 ± 4.3 min (p < 0.001).

Conclusion: A deep learning algorithm for HR-MRVWI interpretation could achieve image reconstruction, vessel stenosis and plaque calculation, which has satisfactory diagnostic performance.

Key points: Question Can a deep learning system achieve image reconstruction, stenosis diagnosis and plaque calculation in high-resolution MR vessel wall imaging (HR-MRVWI)? Findings The overall time reduced from 32.0 ± 11.8 to 12.9 ± 4.3 min (p < 0.001) with the aid of the system. Clinical relevance This effective deep learning system has great potential for processing head and neck HR-MRVWI images; it assists radiologists' workloads and saves considerable time in hospitals. Additionally, it provides plaque-related parameters automatically for the evaluation of atherosclerosis patients.

Objectives: To assess the diagnostic accuracy of chest CT for rib fractures in live children investigated for suspected physical abuse (SPA), using initial and follow-up chest radiographs (CXRs) as the reference standard.

Materials and methods: A retrospective 10-year (September 2011-2021) multicentre search was performed for children less than two years of age who received CXRs and chest CT for SPA. Nineteen consultant radiologists independently read the images: Round 1 (initial CXRs only), Round 2 (CTs only) and Round 3 (initial and follow-up CXRs). No reporter performed Round 3 before Round 1 or 2. Radiologists reported the presence of rib fractures, fracture age, fracture location and confidence level. CT diagnostic accuracy (sensitivity, specificity, and accuracy) was calculated per patient, per rib and per specific location along the rib arc.

Results: A total of 64 patients (36 boys) with a median age of 2 months were included and assessed by 19 independent consultant radiologists. Patient level analysis: CT sensitivity = 90.6% (95% confidence interval [CI]: 88.2-92.6), specificity = 74.2% (95% CI: 70.2-78.0). Rib level analysis: CT sensitivity = 85.6% (95% CI: 84.1-87.0), specificity = 94.16% (95% CI: 93.8-94.4). Location level analysis: CT sensitivity = 75.7% (95% CI: 74.0-77.4), specificity = 97.09% (95% CI: 96.9-97.2).

Conclusion: Chest CT confers accurate rib fracture detection in live children with SPA, with the potential to replace the current standard of performing six CXRs as part of initial and follow-up imaging for SPA.

Key points: Question What is the diagnostic performance of chest CT in detecting rib fractures in live children with SPA, using CXR as a reference standard? Findings Chest CT showed 90.6% sensitivity and 74.2% specificity for detecting rib fractures on patient-based analysis, with 79.7% sensitivity for posterior rib fractures. Clinical relevance Chest CT accurately detects rib fractures in children investigated for SPA and may serve as an alternative to initial and follow-up CXR, supporting timely clinical assessment and management.

Objective: To evaluate the safety and efficacy of intravascular lithotripsy (IVL)-assisted endovascular revascularisation in patients with chronic mesenteric ischaemia (CMI) and heavily calcified mesenteric artery stenoses.

Materials and methods: In this single-centre retrospective study (May 2020-June 2025), consecutive patients with symptomatic CMI, ≥ 50% mesenteric artery stenosis, and moderate-to-severe calcification on CT angiography underwent IVL-assisted endovascular revascularisation. Outcomes included technical success (successful IVL with ≤ 30% residual stenosis after any adjunctive therapy), moderate-to-severe adverse events (AEs), symptom recurrence, clinically driven target vessel revascularisation (CD-TVR), patency, and survival. Kaplan-Meier analysis assessed patency and survival at 6 and 12 months.

Results: Fifty-one patients (median age, 71.5 years; 51% women) underwent treatment of 57 arteries (median stenosis, 72.0%; 96.5% moderate-to-severe calcification). IVL was followed by stenting in 53 de-novo lesions (47 bare-metal, 6 covered), and balloon angioplasty in 4 lesions (3 de-novo, 1 in-stent restenosis). Technical success was 93.0%, with predilatation required in 45.6% of vessels. Median residual stenosis was 16.7% (IQR 11.7), and median lumen gain was 3.5 mm (IQR 2.1). Moderate-to-severe AEs occurred in 27.5% of patients. Two patients were lost to follow-up. During a median follow-up of 578.0 days (IQR 529.5), symptom recurrence occurred in 18.4% of patients, and CD-TVR was required in 16.3%. Primary clinical patency was 93.4% at 6 months and 91.0% at 12 months. Survival rates were 91.7% and 89.4% at 6 and 12 months, respectively; mesenteric ischaemia-related mortality was 2.0%.

Conclusion: IVL is a safe and effective vessel preparation strategy for heavily calcified mesenteric arteries, facilitating endovascular revascularisation in CMI.

Key points: Question Can vessel preparation with intravascular lithotripsy reduce the rate of endovascular treatment failure associated with moderate-to-severe calcification in mesenteric artery stenosis without amplifying procedural risks? Findings Calcium modification with intravascular lithotripsy prior to stenting yielded high technical and clinical success with favourable lumen gain, safety profile, and durable patency. Clinical relevance Adjunctive intravascular lithotripsy is a valuable strategy to mitigate the challenges of calcification in mesenteric artery stenosis, achieving high technical and clinical success while preserving procedural safety, thereby broadening treatment feasibility and improving outcomes in complex disease.