Objective Todeterminetherelationshipbetweenthetumorvolumeoftheperipherallungadenocarcinomawith maximum diameter≤3cmandlymphnodemetastasis(LNM).Methods TheMSCTmanifestationsof235subjectswhowerediagnosedasperipheral lungadenocarcinomawithmaximumdiameter≤3cm wereretrospectivelyanalyzed.Thesepatientsweregroupedaccordingtodifferent parametersincludingsmokinghistory,differentiation,tumorconsistencyandavailabilityoftumornecrosis.Tumorvolumeandratesof LNMamongthesegroupswerecompared.ROCanalysiswasusedtocalculatethecut-offvalueanddiagnosticaccuracy.Results (1) ThetumorvolumeofLNMgroupwaslargerthanthatofnoLNMgroup,cut-offvaluewas5.5cm3,andAUCwas0.76;(2)Therates ofLNMofthewell,moderate-well,moderate,moderate-poorandpoordifferentiationgroupswere0%,8.7%,17.7%,45.6%and46.7%respectively.Theratesofpuregroundglassopacity(p-GGO),mixedandsolidtumorwere0%,8.3%and29.3%respectively.The ratesofthetumorpresentandabsentofnecrosiswere47.8%,22.0%respectively(P＜0.05).Conclusion Usingthevolumeoftumor on MSCTtopredictLNMisanewnon-invasivewayofassessingLNM,withhighsensitivityandspecificity,whichcouldsupplymore imaginginformationforsurgeonstochoosethewayoflymphnodedissection.

Objective: To clinically validate the feasibility and accuracy of cine images acquired through the multitasking method, with no electrocardiogram gating and free-breathing, in measuring left ventricular (LV) function indices by comparing them with those acquired through the balanced steady-state free precession (bSSFP) method, with multiple breath-holds and electrocardiogram gating. Materials and Methods: Forty-three healthy volunteers (female:male, 30:13; mean age, 23.1 ± 2.3 years) and 36 patients requiring an assessment of LV function for various clinical indications (female:male, 22:14; 57.8 ± 11.3 years) were enrolled in this prospective study. Each participant underwent cardiac magnetic resonance imaging (MRI) using the multiple breath-hold bSSFP method and free-breathing multitasking method. LV function parameters were measured for both MRI methods. Image quality was assessed through subjective image quality scores (1 to 5) and calculation of the contrast-to-noise ratio (CNR) between the myocardium and blood pool. Differences between the two MRI methods were analyzed using the Bland–Altman plot, paired t-test, or Wilcoxon signed-rank test, as appropriate. Results: LV ejection fraction (LVEF) was not significantly different between the two MRI methods (P = 0.222 in healthy volunteers and P = 0.343 in patients). LV end-diastolic mass was slightly overestimated with multitasking in both healthy volunteers (multitasking vs. bSSFP, 60.5 ± 10.7 g vs. 58.0 ± 10.4 g, respectively; P < 0.001) and patients (69.4 ± 18.1 g vs. 66.8 ± 18.0 g, respectively; P = 0.003). Acceptable and comparable image quality was achieved for both MRI methods (multitasking vs. bSSFP, 4.5 ± 0.7 vs. 4.6 ± 0.6, respectively; P = 0.203). The CNR between the myocardium and blood pool showed no significant differences between the two MRI methods (18.89 ± 6.65 vs. 18.19 ± 5.83, respectively; P = 0.480). Conclusion Multitasking-derived cine images obtained without electrocardiogram gating and breath-holding achieved similar image quality and accurate quantification of LVEF in healthy volunteers and patients.

Objective To propose an innovative self-supervised learning method for vascular segmentation in computed tomography angiography (CTA) images by integrating feature reconstruction with masked autoencoding. Methods A 3D masked autoencoder-based framework was developed, where in 3D histogram of oriented gradients (HOG) was utilized for multi-scale vascular feature extraction. During pre-training, random masking was applied to local patches of CTA images, and the model was trained to jointly reconstruct original voxels and HOG features of masked regions. The pre-trained model was further fine-tuned on two annotated datasets for clinical-level vessel segmentation. Results Evaluated on two independent datasets (30 labeled CTA images each), our method achieved superior segmentation accuracy to the supervised neural network U-Net (nnU-Net) baseline, with Dice similarity coefficients of 91.2% vs. 89.7% (aorta) and 84.8% vs. 83.2% (coronary arteries). Conclusion The proposed self-supervised model significantly reduces manual annotation costs without compromising segmentation precision, showing substantial potential for enhancing clinical workflows in vascular disease management.