Context-Aware Inductive Bias Learning for Vessel Border Detection in Multi-modal Intracoronary Imaging

A major problem in machine learning is that of inductive bias: how to choose a learner's hypothesis space so that it is large enough to contain a solution to the problem being learnt, yet small enough to ensure reliable generalization from reasonably-sized training sets. Typically such bias is supplied by hand through the skill and insights of experts. In this paper a model for automatically learning bias is investigated. The central assumption of the model is that the learner is embedded within an environment of related learning tasks. Within such an environment the learner can sample from multiple tasks, and hence it can search for a hypothesis space that contains good solutions to many of the problems in the environment. Under certain restrictions on the set of all hypothesis spaces available to the learner, we show that a hypothesis space that performs well on a sufficiently large number of training tasks will also perform well when learning novel tasks in the same environment. Explicit bounds are also derived demonstrating that learning multiple tasks within an environment of related tasks can potentially give much better generalization than learning a single task.

Download Full-text

Load Insensitivity of Preload Adjusted Maximal Power Using Echocardiographic Automated Border Detection to Assess Left Ventricular Performance

Journal of the American College of Cardiology ◽

10.1016/s0735-1097(97)83899-5 ◽

1998 ◽

Vol 31 (2) ◽

pp. 42A

Author(s):

W Mandarino

Keyword(s):

Left Ventricular ◽

Left Ventricular Performance ◽

Maximal Power ◽

Border Detection ◽

Ventricular Performance

Download Full-text

Choice of Intracoronary Imaging: When to Use Intravascular Ultrasound or Optical Coherence Tomography

Interventional Cardiology Review ◽

10.15420/icr.2016.11.01.11 ◽

2016 ◽

Vol 11 (1) ◽

pp. 11

Author(s):

Sudheer Koganti ◽

◽

Tushar Kotecha ◽

...

Keyword(s):

Optical Coherence Tomography ◽

Coronary Angiography ◽

Intravascular Ultrasound ◽

Coronary Stenosis ◽

Coronary Intervention ◽

Vessel Diameter ◽

Drug Eluting Stent ◽

Optical Coherence ◽

Disease Extent ◽

Intracoronary Imaging

Intracoronary imaging has the capability of accurately measuring vessel and stenosis dimensions, assessing vessel integrity, characterising lesion morphology and guiding optimal percutaneous coronary intervention (PCI). Coronary angiography used to detect and assess coronary stenosis severity has limitations. The 2D nature of fluoroscopic imaging provides lumen profile only and the assessment of coronary stenosis by visual estimation is subjective and prone to error. Performing PCI based on coronary angiography alone is inadequate for determining key metrics of the vessel such as dimension, extent of disease, and plaque distribution and composition. The advent of intracoronary imaging has offset the limitations of angiography and has shifted the paradigm to allow a detailed, objective appreciation of disease extent and morphology, vessel diameter, stent size and deployment and healing after PCI. It has become an essential tool in complex PCI, including rotational atherectomy, in follow-up of novel drug-eluting stent platforms and understanding the pathophysiology of stent failure after PCI (e.g. following stent thrombosis or in-stent restenosis). In this review we look at the two currently available and commonly used intracoronary imaging tools – intravascular ultrasound and optical coherence tomography – and the merits of each.

Download Full-text