Yale

Zusammenfassung AbstractBody counting is a method for in vivo activity assessment applied to the monitoring of people with high risk of radionuclide incorporation. Energy-sensitive radiation detectors are arranged relative to the body to quantify radionuclide deposits in anatomical structures, such as lungs, liver and skeleton. This method depends on the specific detection system and is sensitive to the individual anatomy of the person. Accurate activity estimates, which are the basis for dose calculation, require extensive calibration procedures typically involving experimental measurements of anthropomorphic phantoms conforming to a reference person. Current calibration methods offer personalisation for lung and liver counting only with respect to body mass and height and do not specify uncertainties.This work revises and extends the currently applied personalisation methods using radiation transport simulation in combination with computational phantoms derived from medical imaging data. A framework was developed that allows computation of samples of calibration factors for various anatomies in standard measurement setups and anthropometric parameters quantifying anatomic properties. Those samples are applied to create statistical models to derive personalised calibration factors given specific values of anthropometric parameters measured on the person. This gives better estimates in activity assessment and, thereby, dose calculation while quantifying and reducing uncertainties.The framework was implemented in form of an abstract, modular data model, a software tool for modelling, simulation and evaluation of general body counting scenarios, and a statistical analysis method for correlating anthropometric parameters and calibration factors. This allows efficient and reproducible modelling for virtual measurement reconstruction as well as sensitivity analyses. The framework was applied to the calibration of the In Vivo Measurement Laboratory (IVM) at Karlsruhe Institute of Technology (KIT) comprising four freely arrangeable high-purity germanium detectors in lung, liver, knee and head measurement setups. Because of the interindividual anatomical variations in the applied phantoms and iv additional sensitivity analyses, it was possible to give estimates of the expected uncertainties and to reduce them through an algorithmically reproducible approach on calibration.

show abstract

“…RapidMiner (Mierswa et al 2006) is a popular open source data mining tool. It is under development since 2001 and regularly updated.…”

Section: Methodsmentioning

confidence: 99%

Personalised Body Counter Calibration Using Anthropometric Parameters

Pölz

Breustedt

2015

Radiat Prot Dosimetry

View full text Add to dashboard Cite

show abstract

“…(An example compositional system is the tool chain we used to create this document: L A T E X to compile it to a DVI file, BibTeX to add citations, dvips to convert the DVI file to postscript and finally ps2pdf to produce a PDF file.) Alternatively, the experimental platform may be integrated, such as the supervised learning systems WEKA (Witten and Frank 2005;Hall et al 2009) and RapidMiner (Mierswa et al 2006), or the numerical environment MATLAB (Moler et al 1987). (There are also integrated environments for L A T E X; both compositional and integrated approaches have their supporters.…”

Section: What Rl Software Can Do For Youmentioning

confidence: 99%

“…We have not seen this described as a design pattern though it could be taken as one. The RapidMiner documentation (Mierswa et al 2006) uses the term data transparency to refer to the property whereby all components of a system know what to do with the data they receive (such as states, actions and rewards). -The interpreter pattern can be used to implement domain-specific languages (Sect.…”

Section: Design Patternsmentioning

confidence: 99%

On the analysis and design of software for reinforcement learning, with a survey of existing systems

Kovacs

Egginton

2011

Mach Learn

View full text Add to dashboard Cite

Reinforcement Learning (RL) is a very complex domain and software for RL is correspondingly complex. We analyse the scope, requirements, and potential for RL software, discuss relevant design issues, survey existing software, and make recommendations for designers. We argue that broad and flexible libraries of reusable software components are valuable from a scientific, as well as practical, perspective, as they allow precise control over experimental conditions, encourage comparison of alternative methods, and allow a fuller exploration of the RL domain.

show abstract

“…For larger datasets the match workflow can be modeled within the GUI and then executed in batch mode. The FEVER implementation is written in Java and uses the Rapid Miner [9] 1 library of machine learners.…”

Section: Overview Of Fevermentioning

confidence: 99%

Comparative evaluation of entity resolution approaches with FEVER

2009

View full text Add to dashboard Cite

We present FEVER, a new evaluation platform for entity resolution approaches. The modular structure of the FEVER framework supports the incorporation or reconstruction of many previously proposed approaches for entity resolution. A distinctive feature of FEVER is that it not only evaluates traditional measures such as precision and recall but also the effort for configuring (e.g., parameter tuning, training) a good entity resolution approach. FE-VER thus strives for a fair comparative evaluation of different approaches by considering both the effectiveness and configuration effort.

show abstract

Yale

Cited by 735 publications

References 13 publications

Personalised Body Counter Calibration Using Anthropometric Parameters

Personalised Body Counter Calibration Using Anthropometric Parameters

On the analysis and design of software for reinforcement learning, with a survey of existing systems

Comparative evaluation of entity resolution approaches with FEVER

Contact Info

Product

Resources

About