StRegA: Unsupervised anomaly detection in brain MRIs using a compact context-encoding variational autoencoder.

Expert interpretation of anatomical images of the human brain is the central part of neuroradiology. Several machine learning-based techniques have been proposed to assist in the analysis process. However, the ML models typically need to be trained to perform a specific task, e.g., brain tumour segmentation or classification. Not only do the corresponding training data require laborious manual annotations, but a wide variety of abnormalities can be present in a human brain MRI - even more than one simultaneously, which renders a representation of all possible anomalies very challenging. Hence, a possible solution is an unsupervised anomaly detection (UAD) system that can learn a data distribution from an unlabelled dataset of healthy subjects and then be applied to detect out-of-distribution samples. Such a technique can then be used to detect anomalies - lesions or abnormalities, for example, brain tumours, without explicitly training the model for that specific pathology. Several Variational Autoencoder (VAE) based techniques have been proposed in the past for this task. Even though they perform very well on controlled artificially simulated anomalies, many of them perform poorly while detecting anomalies in clinical data. This research proposes a compact version of the "context-encoding" VAE (ceVAE) model, combined with pre and post-processing steps, creating a UAD pipeline (StRegA), which is more robust on clinical data and shows its applicability in detecting anomalies such as tumours in brain MRIs. The proposed pipeline achieved a Dice score of 0.642 ± 0.101 while detecting tumours in T2w images of the BraTS dataset and 0.859 ± 0.112 while detecting artificially induced anomalies, while the best performing baseline achieved 0.522 ± 0.135 and 0.783 ± 0.111, respectively.

Antifibrinolytic agents for reducing blood loss in scoliosis surgery in children.

BACKGROUND: This is an updated version of the original Cochrane review first published in 2008. Scoliosis surgery is often associated with substantial blood loss and potentially detrimental effects in children. Antifibrinolytic agents are often used to reduce perioperative blood loss. Clinical trials have evaluated their efficacy in children undergoing surgical correction of scoliosis, but no systematic review has been published. This review was first published in 2008 and was updated in 2016. OBJECTIVES: To assess the efficacy and safety of aprotinin, tranexamic acid and aminocaproic acid in reducing blood loss and transfusion requirements in children undergoing surgery for correction of idiopathic or secondary scoliosis. SEARCH METHODS: We ran the search for the previous review in June 2007. For this updated version, we searched the Cochrane Central Register of Controlled Trials (CENTRAL; 2015, Issue 7), MEDLINE (1946 to August week 1 2015), Embase (1947 to 2015 week 38), Latin American Caribbean Health Sciences Literature (LILACS) (1982 to 14 August 2015), Database of Abstracts of Reviews of Effects (DARE; 2015, Issue 2) and reference lists of reviews and retrieved articles for randomized controlled trials in any language. We also checked the clinical trial registry at http://www.clinicaltrials.gov on 8 October 2015. SELECTION CRITERIA: We included blinded and unblinded randomized controlled trials (RCTs) that evaluated the effects of antifibrinolytics on perioperative blood loss in children 18 years of age or younger and undergoing scoliosis surgery. DATA COLLECTION AND ANALYSIS: Two review authors independently extracted data. The primary outcome was total blood loss (intraoperative and postoperative combined). Secondary efficacy outcomes were the number of participants receiving blood transfusion (both autologous and allogeneic) or receiving allogeneic blood transfusion alone, and the total amount of blood transfused. Safety outcomes included the number of deaths, the number of participants reporting any adverse event or a serious adverse event, withdrawals due to adverse events and the number of participants experiencing a specific adverse event (i.e. renal insufficiency, hypersensitivity or thrombosis). We assessed methodological risk of bias for each included study and employed the GRADE approach to assess the overall quality of the evidence. MAIN RESULTS: We included three new studies (201 participants) in this updated review, for a total of nine studies (455 participants). All but one study employed placebo as the control group intervention. For the primary outcome, antifibrinolytic drugs decreased the amount of perioperative blood loss by 427 mL (95% confidence interval (CI) 251 to 603 mL), for a reduction of over 20% versus placebo. We rated the quality of evidence for our primary outcome as low on the basis of unclear risk of bias for several domains in most studies and the small total number of participants.For secondary outcomes, fewer participants receiving antifibrinolytic drugs received transfusion (allogeneic or autologous) versus those receiving placebo (risk ratio (RR) 0.65, 95% CI 0.50 to 0.85, number needed to treat to prevent one additional harmful outcome (NNTp) 5; very low-quality evidence). Only two studies specifically evaluated the number of participants transfused with only allogeneic blood (risk difference (RD) -0.15, 95% CI -0.26 to -0.03, NNTp 7; very low-quality evidence). Antifibrinolytic drugs decreased the volume of blood transfused by 327 mL (95% CI -186 to -469 mL; low-quality evidence).No study reported deaths in active or control groups. Data were insufficient to allow performance of meta-analysis for any safety outcome. No studies adequately described their methods in assessing safety. The only adverse event of note occurred in one study, when three participants in the placebo group developed postoperative deep vein thrombosis. AUTHORS' CONCLUSIONS: Since the last published version of this review (2008), we have found three new studies. Additional evidence shows that antifibrinolytics reduce the requirement for both autologous and allogeneic blood transfusion. Limited evidence of low to very low quality supports the use of antifibrinolytic drugs for reducing blood loss and decreasing the risk, and volume, of transfusion in children undergoing scoliosis surgery. Evidence is insufficient to support the use of a particular agent, although tranexamic acid may be preferred, given its widespread availability. The optimal dose regimen for any of these three agents has not been established. Although adverse events appear to occur infrequently, evidence is insufficient to confirm the safety of these agents, particularly for rare but potentially catastrophic events. No long-term safety data are available.