METHODS: We propose to use 3D Axial-Attention, which requires a fraction of the computing power of a regular Non-Local network (i.e., self-attention). Unlike a regular Non-Local network, the 3D Axial-Attention network applies the attention operation to each axis separately. Additionally, we solve the invariant position problem of the Non-Local network by proposing to add 3D positional encoding to shared embeddings.
RESULTS: We validated the proposed method on 442 benign nodules and 406 malignant nodules, extracted from the public LIDC-IDRI dataset by following a rigorous experimental setup using only nodules annotated by at least three radiologists. Our results show that the 3D Axial-Attention model achieves state-of-the-art performance on all evaluation metrics, including AUC and Accuracy.
CONCLUSIONS: The proposed model provides full 3D attention, whereby every element (i.e., pixel) in the 3D volume space attends to every other element in the nodule effectively. Thus, the 3D Axial-Attention network can be used in all layers without the need for local filters. The experimental results show the importance of full 3D attention for classifying lung nodules.
METHODS: Twenty-eight patients with severe TBI (GCS ≤ 8, three patients had initial GCS = 9-10, but rapidly deteriorated to ≤8) were recruited. CSF was collected from admission to day 5 post-injury. TRP, kynurenine (KYN), kynurenic acid (KYNA), QUIN, anthranilic acid (AA) and 3-hydroxyanthranilic acid (3HAA) were measured in CSF. The Glasgow Outcome Scale Extended (GOSE) score was assessed at 6 months post-TBI. Post-mortem brains were obtained from the Australian Neurotrauma Tissue and Fluid Bank and used in qPCR for quantitating expression of KP enzymes (indoleamine 2,3-dioxygenase-1 (IDO1), kynurenase (KYNase), kynurenine amino transferase-II (KAT-II), kynurenine 3-monooxygenase (KMO), 3-hydroxyanthranilic acid oxygenase (3HAO) and quinolinic acid phosphoribosyl transferase (QPRTase) and IDO1 immunohistochemistry.
RESULTS: In CSF, KYN, KYNA and QUIN were elevated whereas TRP, AA and 3HAA remained unchanged. The ratios of QUIN:KYN, QUIN:KYNA, KYNA:KYN and 3HAA:AA revealed that QUIN levels were significantly higher than KYN and KYNA, supporting increased neurotoxicity. Amplified IDO1 and KYNase mRNA expression was demonstrated on post-mortem brains, and enhanced IDO1 protein coincided with overt tissue damage. QUIN levels in CSF were significantly higher in patients with unfavourable outcome and inversely correlated with GOSE scores.
CONCLUSION: TBI induced a striking activation of the KP pathway with sustained increase of QUIN. The exceeding production of QUIN together with increased IDO1 activation and mRNA expression in brain-injured areas suggests that TBI selectively induces a robust stimulation of the neurotoxic branch of the KP pathway. QUIN's detrimental roles are supported by its association to adverse outcome potentially becoming an early prognostic factor post-TBI.
METHODS: We used digitised mammograms for 371 monozygotic twin pairs, aged 40-70 years without a prior diagnosis of breast cancer at the time of mammography, from the Australian Mammographic Density Twins and Sisters Study. We generated normalised, age-adjusted, and standardised risk scores based on textures using the Cirrus algorithm and on three spatially independent dense areas defined by increasing brightness threshold: light areas, bright areas, and brightest areas. Causal inference was made using the Inference about Causation from Examination of FAmilial CONfounding (ICE FALCON) method.
RESULTS: The mammogram risk scores were correlated within twin pairs and with each other (r = 0.22-0.81; all P
MATERIALS AND METHODS: This study was conducted in Hospital Kepala Batas (HKB), a district hospital in Penang State, Malaysia, which was the designated regional COVID hospital during the pandemic. It was a retrospective observational study, where children who were admitted from November 2020 to March 2021, and attended follow-up clinics from Jan 2021 to May 2021, were recruited.
RESULTS: This study comprised 90 subjects, from 3 months old to 12 years old, mean of 6.5 years old. When comparing asymptomatic and symptomatic children, children with comorbidities were more likely to be symptomatic with a pvalue of 0.045 using the Pearson Chi-square test. All our patients' symptoms resolved upon discharge. During followup at 2-4 months after COVID-19 infection, all children were reported as back to their usual selves. Fifteen patients had recurrent symptoms. Most of their symptoms pointed towards an acute infection. One patient had two episodes of illness, while the rest had one. The most common symptoms were cough, fever and runny nose. The average duration of illness of these 16 episodes was 4.5 days with a standard deviation of 2.48. None of these symptoms lasted more than seven days. None of them required hospital admission. None of them had recurrent COVID-19 infections. Tweleve out of 72 children who had been going to school stopped physical school after COVID-19 infection. Our findings differed from other studies. These could be due to the limitations that we faced.
CONCLUSION: Most children who contracted COVID-19 infection recovered fully after acute infection, and most of them recovered fully without long-term sequelae.