Background

Carotid and coronary atherosclerosis are critical determinants of cardiovascular risk, yet their interrelationship in middle-aged populations is incompletely understood. This study assessed carotid plaque composition, risk-factor associations, coronary disease, and sex differences in a subclinical cohort.

Methods

Within the Swedish CArdioPulmonary bioImage Study (SCAPIS), 533 asymptomatic individuals aged 50–64 years with carotid plaque ≥2.7 mm on ultrasound underwent 3 T multi-contrast carotid cardiovascular magnetic resonance (CMR) and coronary computed tomography angiography. Carotid plaque characteristics were determined manually using established criteria on multi-contrast weighted carotid CMR. Bayesian regression models evaluated associations between cardiovascular risk factors and coronary atherosclerosis.

Results

Lipid rich necrotic core (LRNC) was present in 60% and intraplaque hemorrhage (IPH) in 5.4%; calcification occurred in 48.6%. Maximum carotid wall thickness was 1.8 (1.6-2.0) mm, and mean lumen area 31.3 (26.7-36.1) mm². Coronary atherosclerosis was present in 63.6% of participants, with ≥50% stenosis in 12.9%, and coronary artery calcium score >400 in 12.8%. Men (N=367) had larger carotid lumen area, mean wall area, and maximum wall thickness (all p < 0.001) than women (N=166), differences that persisted after body-surface-area adjustment (all p < 0.01). LRNC was present in 66% of men compared to 47% of women (p < 0.001). LRNC presence was not associated with coronary atherosclerosis, whereas IPH was associated with coronary involvement.

Conclusion

In middle-aged individuals, distinct cardiovascular risk factors were positively linked to presence and volume of LRNC and calcified plaques. The substantial prevalence of high-risk plaque features, particularly LRNC and especially in men, highlights a significant subclinical carotid disease burden.

Lay summary

This study used state-of-the-art magnetic resonance imaging to characterize atherosclerotic plaques in the carotid arteries in middle-aged individuals without clinical cardiovascular disease, offering the following insight into early, subclinical atherosclerosis:

Lipid rich necrotic core (LRNC), a marker of high-risk plaques, was present in 60% of participants with carotid plaques, suggesting a substantial burden of potentially vulnerable atherosclerosis even in asymptomatic individuals.Carotid plaque features such as increased wall thickness, calcification, and presence of LRNC were variably associated with cardiovascular risk factors and plaques with increased wall area, wall thickness, and calcium showed correlations with coronary artery calcium and plaque burden on CT, indicating systemic atherosclerotic involvement.

Purpose: To implement and evaluate deep learning-based methods for the classification of pediatric brain tumors in MR data.

Materials and methods: A subset of the “Children’s Brain Tumor Network” dataset was retrospectively used (n=178 subjects, female=72, male=102, NA=4, age-range [0.01, 36.49] years) with tumor types being low-grade astrocytoma (n=84), ependymoma (n=32), and medulloblastoma (n=62). T1w post-contrast (n=94 subjects), T2w (n=160 subjects), and ADC (n=66 subjects) MR sequences were used separately. Two deep-learning models were trained on transversal slices showing tumor. Joint fusion was implemented to combine image and age data, and two pre-training paradigms were utilized. Model explainability was investigated using gradient-weighted class activation mapping (Grad-CAM), and the learned feature space was visualized using principal component analysis (PCA).

Results: The highest tumor-type classification performance was achieved when using a vision transformer model pre-trained on ImageNet and fine-tuned on ADC images with age fusion (MCC: 0.77 ± 0.14 Accuracy: 0.87 ± 0.08), followed by models trained on T2w (MCC: 0.58 ± 0.11, Accuracy: 0.73 ± 0.08) and T1w post-contrast (MCC: 0.41 ± 0.11, Accuracy: 0.62 ± 0.08) data. Age fusion marginally improved the model’s performance. Both model architectures performed similarly across the experiments, with no differences between the pre-training strategies. Grad-CAMs showed that the models’ attention focused on the brain region. PCA of the feature space showed greater separation of the tumor-type clusters when using contrastive pre-training.

Conclusion: Classification of pediatric brain tumors on MR-images could be accomplished using deep learning, with the top-performing model being trained on ADC data, which is used by radiologists for the clinical classification of these tumors.