Objective: To verify the ability of the deep learning model in identifying five subtypes and normal images in noncontrast enhancement CT of intracranial hemorrhage. Method: A total of 351 patients (39 patients in the normal group, 312 patients in the intracranial hemorrhage group) performed with intracranial hemorrhage noncontrast enhanced CT were selected, with 2768 images in total (514 images for the normal group, 398 images for the epidural hemorrhage group, 501 images for the subdural hemorrhage group, 497 images for the intraventricular hemorrhage group, 415 images for the cerebral parenchymal hemorrhage group, and 443 images for the subarachnoid hemorrhage group). Based on the diagnostic reports of two radiologists with more than 10 years of experience, the ResNet-18 and DenseNet-121 deep learning models were selected. Transfer learning was used. 80% of the data was used for training models, 10% was used for validating model performance against overfitting, and the last 10% was used for the final evaluation of the model. Assessment indicators included accuracy, sensitivity, specificity, and AUC values. Results: The overall accuracy of ResNet-18 and DenseNet-121 models were 89.64% and 82.5%, respectively. The sensitivity and specificity of identifying five subtypes and normal images were above 0.80. The sensitivity of DenseNet-121 model to recognize intraventricular hemorrhage and cerebral parenchymal hemorrhage was lower than 0.80, 0.73, and 0.76 respectively. The AUC values of the two deep learning models were above 0.9. Conclusion: The deep learning model can accurately identify the five subtypes of intracranial hemorrhage and normal images, and it can be used as a new tool for clinical diagnosis in the future.

miRNAs are small non-coding RNAs modulating gene expression, and variants in miRNA genes are involved in the pathogenesis of ischemic stroke (IS). However, the effect of miR-34a polymorphisms on IS susceptibility has rarely been reported. In the present study, we investigated the association between rs12128240, rs2666433, and rs6577555 of the miR-34a gene and IS susceptibility. Snapshot assay was used to detect miR-34a polymorphisms in 548 IS patients and 560 controls. Relative expression of miR-34a was measured by quantitative real-time PCR. We found that rs2666433 was associated with a significantly increased risk of IS (AA vs. GG: OR = 1.61, 95% CI = 1.05–2.52, P = 0.031; AA vs. GG+GA: OR = 1.58, 95% CI = 1.05–2.45, P = 0.026). For the IS subtypes, rs2666433 was associated with large artery atherosclerosis (AA vs. GG: OR = 2.09, 95% CI = 1.16–3.51, P = 0.007; AA vs. GG+GA: OR = 2.02, 95% CI = 1.15–3.33, P = 0.007; A vs. G: OR = 1.36, 95% CI = 1.07–1.81, P = 0.021). Additionally, the level of miR-34a was significantly up-regulated in IS patients compared to the controls ( P < 0.001), and patients with rs2666433 AA genotype had a higher level of miR-34a than those with GG+GA genotypes ( P < 0.001). Furthermore, increased level of homocysteine was observed in IS patients compared to the controls ( P < 0.001), especially in patients carrying the rs2666433AA genotype compared to those carrying the rs2666433 GG+GA genotypes ( P < 0.001). However, no significant association between rs12128240 or rs6577555 and IS was found. Collectively, our study found the association between miR-34a polymorphisms and the risk of IS among the Chinese population. The results may provide an explanation for etiology of IS and a potential biomarker or therapeutic target for IS. HIGHLIGHTS MiR-34a rs2666433 polymorphism was associated with an increased risk of ischemic stroke. The level of miR-34a was significantly up-regulated in ischemic stroke patients compared with controls, and patients with rs2666433 AA genotype had a higher level miR-34a than those with GG+GA genotypes. Furthermore, increased level of homocysteine was showed in IS patients compared to controls, and in patients carrying the rs2666433AA compared to those carrying the rs2666433 GG+GA.

No abstract

No abstract