Known knowns, known unknowns and unknown unknowns: a 2010 update on carotid artery disease

A wholistic view of continual learning with deep neural networks: Forgotten lessons and the bridge to active and open world learning

Current deep learning methods are regarded as favorable if they empirically perform well on dedicated test sets. This mentality is seamlessly reflected in the resurfacing area of continual learning, where consecutively arriving data is investigated. The core challenge is framed as protecting previously acquired representations from being catastrophically forgotten. However, comparison of individual methods is nevertheless performed in isolation from the real world by monitoring accumulated benchmark test set performance. The closed world assumption remains predominant, i.e. models are evaluated on data that is guaranteed to originate from the same distribution as used for training. This poses a massive challenge as neural networks are well known to provide overconfident false predictions on unknown and corrupted instances. In this work we critically survey the literature and argue that notable lessons from open set recognition, identifying unknown examples outside of the observed set, and the adjacent field of active learning, querying data to maximize the expected performance gain, are frequently overlooked in the deep learning era. Hence, we propose a consolidated view to bridge continual learning, active learning and open set recognition in deep neural networks. Finally, the established synergies are supported empirically, showing joint improvement in alleviating catastrophic forgetting, querying data, selecting task orders, while exhibiting robust open world application.

Recent Advances in Open Set Recognition: A Survey

In real-world recognition/classification tasks, limited by various objective factors, it is usually difficult to collect training samples to exhaust all classes when training a recognizer or classifier. A more realistic scenario is open set recognition (OSR), where incomplete knowledge of the world exists at training time, and unknown classes can be submitted to an algorithm during testing, requiring the classifiers to not only accurately classify the seen classes, but also effectively deal with unseen ones. This paper provides a comprehensive survey of existing open set recognition techniques covering various aspects ranging from related definitions, representations of models, datasets, evaluation criteria, and algorithm comparisons. Furthermore, we briefly analyze the relationships between OSR and its related tasks including zero-shot, one-shot (few-shot) recognition/learning techniques, classification with reject option, and so forth. Additionally, we also review the open world recognition which can be seen as a natural extension of OSR. Importantly, we highlight the limitations of existing approaches and point out some promising subsequent research directions in this field.

Screening for Peripheral Arterial Disease and Carotid Artery Disease in Patients With Abdominal Aortic Aneurysm

Screening for concomitant atherosclerotic disease is important in cardiovascular risk reduction. This study assessed the prevalence of carotid artery disease (CAD) and peripheral arterial disease (PAD) in patients with known abdominal aortic aneurysms (AAAs). All patients with AAA attending the vascular laboratory between the January 1, 2007, and December 31, 2009, were eligible for a carotid ultrasound and measurement of ankle brachial indices. A total of 389 (305 males) patients were identified on the AAA surveillance program with a mean (±standard deviation) age of 76 (±8) years. The mean age of the males was 75.4 (±7.8) years, and the mean age of the females was 77 (±11) years. A total of 332 patients were assessed for CAD, and 101 (30.4%) of those were found to have significant disease. A total of 289 patients were assessed for PAD of which 131 (45.3%) were found to have PAD at rest, and 289 patients were assessed for both and 59 (20.4%) patients had significant CAD + PAD. Patients with AAAs are at high risk of other atherosclerotic disorders, and, therefore, they should receive intensive medical optimization.

Comparison of Doppler ultrasound velocity parameters in the determination of internal carotid artery stenosis

This study aims to compare the evidence base and practical results of the Joint Recommendations for Reporting Carotid Ultrasound Investigations in UK, published in 2009, and existing carotid scan protocols based on the Society of Radiologists in Ultrasound 2003 Consensus. A prospective sequential evaluation of the 2009 recommendations was performed at the University Hospital of Wales, Cardiff. Additional measurements in line with the recommendations were made during carotid scans. The grading of internal carotid artery stenosis using the 2003 and 2009 UK recommendations, and recommended measures of PSV, PSV ratio and St Mary’s ratio were compared. In comparison to PSV classification, PSV ratio produced lower stenosis classification in 29% and 24% of cases in the 50–69% and 70–89% stenosis bands respectively. St Mary’s ratio produced poor classification agreement across all bands, particularly the 50%–69% stenosis band. Agreement of two measures is recommended for diagnostic confidence; however, in the 50%–69% and 70%–89% stenosis bands, agreement of two measures only occurred in 70% of scans. This evaluation suggests that the use of three measurements in the 2009 recommendations complicates rather than aids diagnosis, especially in the 50%–69% and 70%–89% stenosis bands, and does not provide significant improvement over the 2003 guidelines. No evidence was found to support the combined use of the three measures.

The current management of carotid atherosclerotic disease: who, when and how?

Ischaemic stroke represents a major health hazard in the western world, which has a severe impact on society and the health-care system. Roughly, 10% of all first ischaemic strokes can be attributed to significant atherosclerotic disease of the carotid arteries. Correct management of these lesions is essential in the prevention and treatment of carotid disease-related ischaemic events. The close relationship between diagnosis and medical and surgical management makes it necessary that all involved physicians and surgeons have profound knowledge of management strategies beyond their specific speciality. Continuous improvement in pharmacological therapy and operative techniques as well as frequently changing guidelines represent a constant challenge for the individual health-care professional. This review gives a thorough outline of the up-to-date evidence-based management of carotid artery disease and discusses its current controversies.

Current role of biomarkers in carotid disease: a systematic review

BACKGROUND AND PURPOSE

Accumulating evidence suggests that carotid plaque vulnerability can be used as a determinant of ischemic stroke risk stratification and carotid intervention. Novel markers of high-risk carotid plaque in patients are needed.

SUMMARY OF REVIEW

Advances in cellular and molecular pathophysiology, the demand for accurately predicting carotid risk, and choosing the optimal prevention strategy are stimulating great interest in the development of novel surrogate markers. Biomarkers in cardiovascular disease are expected to predict the natural history, clinical outcomes, and the efficacy of disease-modifying interventions. We aimed to review the literature regarding clinical data on novel serum biomarkers related to ischemic cerebrovascular events associated with carotid artery disease. We provide background information on the biomarkers related to all aspects of carotid disease: natural history, carotid intervention strategies for symptomatic and asymptomatic patients, perioperative risk prediction, and their therapeutic implications.

CONCLUSION

At present, heterogeneous data support evidence that biological markers can help existing practices to more accurately assess patients at risk for stroke. Randomized-controlled trials for carotid artery disease and carotid intervention, incorporating biomarkers, are needed.