Upload
subhashis-das
View
389
Download
1
Tags:
Embed Size (px)
Citation preview
Ontology-based Support for
Brain Tumour Study
Subhashis Das
MSLIS 2011-13
DRTC, Indian Statistical Institute
Presentation Structure
Introduction and background
Problems of current informaton retrieval systems
Why I chose Ontology
Ontology building method
Use
Conclusion
Introduction
For the diagnosis and detection of brain tumour
Computer based diagnosis system proves to be helpful
Brain tumour is one of the most deadly diseases in India
It contributes significantly to morbidity
Poor prognosis
Background
Diagnosis using MRI & MRS is the main way of detection
Brain tumours remain an important cause of morbidity and mortality and afflict a large percentage of the World population.
In children over 1 year of age, brain tumours are the most common solid malignancies that cause disease-related death.
MRI: Magnetic Resonance Imaging; MRS: Magnetic Resonance Scan
Why I chose brain tumours?
Clinical importance Important cause of morbidity and mortality in adults
and children
Few improvements in outcome
New approaches to management needed via greater understanding
Problems in IR
Current information retrieval systems mostly Keyword search, Low precision. Junk retrieval
So what is the solution?
Ontology based information system
What an ontology is and is not?
Rumours about ontologies
Ontologies are overly publicised:
“Ontology” is becoming a buzz word
“Ontology” can “say” whatever one intends to say
“Ontology” means inference
“Ontology” is the ultimate solution for
interoperability
Ontology
A common language/vocabulary/terminology for various participants
Formalised in an unambiguous representation
For software agents, human experts, patients
To assist in communication between humans and computer To achieve interoperability To improve the design and quality of software systems
Ontology
A “static” conceptualisation of the world “What is?” rather than “How does?”
Allows reasoning which respects the translation of concepts as sets of (possible) individuals
Provides the underlying knowledge model for other types of reasoning, e.g. Rule Based, Case Based etc.
Ontology enhance the semantics of terms by providing richer relationships between the terms of vocabulary
Why OWL (Web Ontology Language)?
Reasoning capability Subsumption relationship (is-a)
Relies on necessary and sufficient definition of concepts Good for maintaining a consistent ontology
W3C standard Good support: existing systems and tools Good compatibility:
Many ontologies are developed in owl or will be translated into owl
Good extensibility
Benefits
The analysis and combination of the information the result will be presented in a way that makes it easier for the user to have an overview of the up-to-date knowledge about brain tumour.
The inherited organization of ontologies adds taxonomical context to search result making it easier for the research to spot conceptual relationships in data.
Any one can find relationship between different factors that are responsible for brain tumour.
Other benefits
Eliminating redundant effort
Significant head-start
Interoperability with other ontologies
Community acceptance
Methodology
Identification of the terminology Analysis Synthesis Standardization Ordering
Sources: Giunchiglia, Fausto; Dutta, Biswanath;Maltese, Vincenzo and Farazi, Feroz (2012): A facet-based methodology for the construction of a large-scale geospatial ontology
Identification of the terminology
Information sources National Brain Tumor Society (http://www.braintumor.org)-USA) American Brain Tumor Association (http://www.abta.org/) Brain Tumor Foundation of Canada (http://www.braintumour.ca/) Brain Tumor Association of Western Australia
(http://braintumourwa.com)
Resource pre-processing
Mapping the resources
Integration of the resources
Analysis and synthesis
The formal terms collected during the previous phase are analyzed per genus.
With the synthesis, formal terms are arrange into facets
Standardization
SNOMED CT®
Systematized Nomenclature of Medicine-Clinical Term (SNOMED CT) more than 311,000 active concepts with unique meanings and formal logic-based definitions organized into 19 hierarchies.
Medical Subject Headings (MeSH) The MeSH is a controlled vocabulary developed by the National Library of Medicine (NLM) for indexing and retrieval of biomedical literature, including MEDLINEMore than 1,77,000 entry
Sources: http://www.ncbi.nlm.nih.gov/mesh and http://viw2.vetmed.vt.edu/sct/menu.cfm
Principles of building brain tumour ontology
The constructed brain tumour ontology has four main branches
Types- Describing different types of brain tumour
Symptoms- Describing symptoms of brain tumour
Causes- Causes responsible for brain tumour which are mainly environmental and genetic
Treatments- Giving an overview of all treatments possible for that particular type of brain tumour
Types
Primary tumors of the brain Gliomas
Lowest grade tumors Lower grade malignancies Higher-grade malignancies Highest-grade malignancies
Meningioma Primitive neuroectodermal tumors (PNET) Pituitary tumors Pineal Tumors Choroid plexus tumors Other, more benign primary tumors Tumors of nerves and/or nerve sheaths Cyst Other primary tumors, including skull base Primary Central Nervous System Lymphoma (PCNSL)
Metastatic brain tumors and carcinomatous meningitis
Symptoms A new seizure in an adult Gradual loss of movement or sensation in an arm or leg Unsteadiness or imbalance, especially if it is associated with headache Loss of vision in one or both eyes, especially if the vision loss is more
peripheral Double vision, especially if it is associated with headache Hearing loss with or without dizziness Speech difficulty of gradual onset Other symptoms may also include nausea or vomiting that is most severe in
the morning, confusion and disorientation, and memory loss. The following symptoms are usually not caused by a brain tumor, but may
sometimes be: Headache A change in behavior
Genetic-Causes
The ontology explain that brain tumour have different types which also further divided into subtypes. Brain tumour is caused by causes which can be genetic or environmental.
Environmental causes
Treatment
There is a corresponding symptoms of observable characteristics of an ill individual and treatment possible for the disorder that can be chemotherapy, surgery, psychotherapy or medication.
Demo
Now I show you how I build ontology using Protégé 4.1 ontology editor
Use
For physician If a medical practitioner queries the system, she/he will mainly be
interested in
Symptoms
Possible treatment
Use
When a physician cannot identify disease.
Use
For researchers
Its helps on drug discovery
Its directed or may allow researcher to narrow down the region of interest on particular gene Neurofibromatosis 1 (NF1 gene), Neurofibromatosis 2 (NF2 gene), Turcots (APC gene), Gorlins (PTCH gene), Li-Fraumeni syndrome (TP53 gene).
Limitation of ontology model
Assertion errors
Relevance errors
Encoding errors
Conclusions and future work
A computer-base brain tumour ontology support the works of researcher in gathering information on brain tumour research and allows user across the world to intelligently access new scientific information much more quickly.
Shared knowledge improves research efficiency and effectiveness, as it helps to avoid unnecessary redundancy in doing the same experiments.
Reference
National Brain Tumor Society (http://www.braintumor.org)-USA American Brain Tumor Association (http://www.abta.org/) Brain Tumor foundation of Canada (http://www.braintumour.ca/) Brain Tumor Association of Western Australia
(http://braintumourwa.com) Snomed-CT ( http://www.ihtsdo.org/snomed-ct/) Medical Subject Headings
(http://www.nlm.nih.gov/pubs/factsheets/mesh.html) Hadzic, Maja and Chang, Elizabeth (2005): Ontology-based support
for human disease study. IEEE, 2005, pp.1-7.