You are on page 1of 5

2009 World Congress on Computer Science and Information Engineering

Design and Implementation of a Knowledge Engineering-based Dental


Diagnostic Expert System (DDES)

Haiguang Zhang
Rapid Manufacturing Engineering Center, Qingxi Hu, Yuan Yao, Qi Wang
Shanghai University, Shanghai 200444, Rapid Manufacturing Engineering Center,
China Shanghai University, Shanghai 200444,
College of Mechanical Engineering, China
Inner Mongolia University of Technology, huqingxi@shu.edu.cn
Hohhot, 010051, China
haiguangdd@shu.edu.cn

Abstract experiences constantly[1]. The number of experienced


dentists is relatively limited, so induce some young
In this paper, we propose an expert system which is doctors who are short of experiences to participate in
based on Knowledge Engineering named DDES - diagnosis. Obviously, it is necessary that many
Dental Diagnostic Expert System. Besides, a hybrid ordinary dentists can share abundant experiences and
knowledge acquisition strategy is proposed to knowledge which are owned by a few specialists.
efficiently elicit the knowledge from experienced If the expert system is built through digging special
dentists and construct the medical ontology. The knowledge and experiences from human experts’ brain,
system fulfills the storage of dental photos and X-ray it isn’t limited by time or space and supply
images, typical cases, patients’ information and consultation and assistance for workers who are in the
concerned data. It enables the complex query for fields [2]. In almost thirty years, the medical expert
combined conditions and the display of requisite systems have been researched and developed at home
results by interactive interface between user and and abroad [3] [4]. Up to now, it has already made a
computer. For those teeth by root canal therapy, we great evolution, but there are a few systems using for
can not only differentiate the 3D images between pre dental consultation and diagnosis [5] [6]. Most of
and post treatment, but also analyze force condition of expert systems which are related with teeth are mainly
every tooth. Theses query and analysis results supply aim to dental electronic medical record [7] ,dental
dentists with diagnostic basis. The DDES uses digital image processing [8] or information
Artificial Intelligence, Knowledge Engineering, management of dentistry. The functions of them are
Database and Computer technology and acts as a too simple to satisfy the needs of dentists.
consultant role to help dentists diagnose some dental Dental Diagnostic Expert System (DDES)
diseases. For the students of dental academy and integrates Artificial Intelligence, Knowledge
practical dentist, the DDES is a useful learning and Engineering, Database and Computer technology. It
aided diagnosis tool. can fulfill knowledge acquire automatically, complex
query for combined conditions, intelligent diagnosis of
dental diseases. It acts as a consultant role to help
1. Introduction dentists diagnose some dental diseases.

It’s very important for a better life to own healthy 2. The whole framework of DDES
teeth. The dental diseases trouble many patients. The
ratio of dentists and total population is only maintained DDES adopts modularized frame which includes
at 1:100000, so the number of dentists can’t satisfy the inference engine, dental knowledgebase and general
people’s needs in China. Besides, the stomatology is a database module, explaining module, human-machine
very strong technical and operable science, so it is interface and knowledge acquisition module. The
necessary to exercise skills and accumulate modularized structure is convenient to debugging and

978-0-7695-3507-4/08 $25.00 © 2008 IEEE 362


DOI 10.1109/CSIE.2009.589
maintenance of the program and expansion of the 2.4 Inference module
system. The structure and relationship inter-modules of
DDES are shown in Figure 1. The inference engine can simulate thinking or

Figure 1. The whole framework of DDES

2.1 Human-machine interface module reasoning process of dental experts to control and
execute problem-solving. The traditional inference
The human-machine interface module is used to methods include forward, reverse and hybrid inference.
input and output data. The users aren’t mainly versed The inference module of DDES adopts uncertainty
in computer, such as dentists or patients, so a friendly reasoning. According to symptoms known, DDES can
interface must be required. The user interface of DDES reason and diagnosis through transferring the data from
is developed by JSP. Because of web-based, it is easy database and using knowledge and rules in
to operate. Different page is used respectively to query, knowledgebase.
diagnosis, analysis, suggestion and discussion.
2.5 Explaining module
2.2 Database module
The explaining module can illuminate the process
The database module of DDES is composed of how the diagnostic results generate and what the main
dental knowledgebase and general database. reasons cause the disease. This module needs support
The general database mainly storages a large of form inference engine and dental knowledge.
data and information relating with teeth and dental
diseases, including glossary, the name and functions of 3. Key technologies and implementation
teeth, the name and features of dental diseases, and procedure
images, shape photos, cross-section photos, X ray
images and micro-CT scanning data of teeth. 3.1 Hybrid knowledge acquisition strategy
The dental knowledgebase includes knowledge and
rules using for decision-making, symptoms of normal Knowledge acquisition (KA) and knowledge
dental diseases, intermediate processing results, representation are the fundamental building blocks of
diagnosis results, typical cases, and therapeutic knowledge-based systems. The stomatology is a very
schemes. It is a critical factor to implement strong technical and operable science, so relative
knowledge-base expert system. knowledge is acquired difficultly. But it is very critical
to get knowledge from dental experts. KA has always
2.3 Knowledge acquisition module been regarded as the bottleneck in developing an
expert system. Some commonly used approaches are
This module can acquire knowledge effectively interviews, observations, taking experts through case
form dental experts, and then transform the knowledge studies, and rule induction by machines learning [9].
acquired into apprehensible form by computer and In order to help extracting knowledge from dental
input to the dental knowledgebase. experts and constructing the corresponding medical
ontology hierarchy, we propose an efficient KA
scheme to fast conceptualize medical domain

363
knowledge by using a hybrid method. Firstly, through
questionnaire surveys and interviews to patients and
dental experts, we can collect a large number of raw
data describing symptoms and construct concept
model. Meanwhile we can adopt KA tools to assist in
eliciting knowledge automatically which includes
object-oriented KA editor and web-based KA interface.
These tools are often easier and more accurate to
enumerate critical factors and construct ontology
hierarchy. Finally, dental expert or knowledge engineer
will merge these data and information to a whole
dental medical ontology. The implementation
procedure of hybrid dental knowledge acquisition is Figure 3. Sample databases
shown in Figure 2. Database is a necessary condition if users want to
complex query for combined conditions. Namely,
dentists can input combined conditions through
human-machine interface, DDES will search matched
data or information and feed results found back to
users in reports or images forms.
(1) General database is composed of four sub-
databases which are glossary, information, photos and
CT data. They separately sore relative data. For
example, if user wants to query the information of the
maxillary premolar, he just inputs name of the teeth or
click on corresponding poison in occlusion image.
Figures 4 shows information found in general database.

Shape photos Cross-section photos

Figure 2. Hybrid dental knowledge acquisition

3.2 Construction of Database


X ray images
Database of dental diseases plays an important role Figure 4. Results in general database
of the DDES. It is used for storing all the information (2) Knowledgebase is composed of facts database,
needed for diagnosis. We must classify information cases database, diagnosis rules database and
collected and add to corresponding sub-database therapeutic schemes database. The facts database stores
gradually. The databases of DDES are realized by symptom, analysis result of teeth and intermediate
using Access. Figure 3 shows some databases used in processing results. The cases database stores old
the DDES. experiences and typical cases. The diagnosis rules
database stores relative dental diseases knowledge and
therapy experiences of dental experts. The therapeutic
schemes database stores causes of disease, actions

364
needed to execute, prescriptions and prevention
schemes.

3.3 Auxiliary Diagnosis Module

The dental caries is the most common dental


disease. Now root canal therapy is its main therapeutic
method. Even the same position tooth, different patient
has obvious shape difference of tooth and different
degree and position of cavity. If dentists adopt the
same therapy scheme with that of old cases to deal
with tooth, it is very possible to make mistakes.
Aiming to solve this problem, DDES adds auxiliary
diagnosis module. The functions of the module include
two parts:
(1) Force analysis
For the tooth diagnosed to dental caries, we will
scan pre and post treatment tooth with Micro-CT and
store scanning data into CT database. Then using these
data to realize 3D reconstruction by Simpleware and
analyze force by Ansys. The analysis results can Figure 6. Extraction and analysis for middle of
display on the interface of DDES and supply dentists root canal
with important factor to modify existing methods and
decide new therapy scheme. Figure 5 shows handling
process of force analysis module.
4. Work flow of DDES
(2) Extraction and analysis for midline of root canal
The dentists must input patient’s main symptoms

Figure 5. Handling process of force analysis module

During the process of root canal therapy, curing through human-machine interface at first, and then
machines may injure root canal. For the dentists who DDES will check the description. If the main
are lack of experiences, the probability of misoperation symptoms are not matching dental diseases
is higher. If the dentists can clearly know difference characteristics, system will reply the “Wrong Input”
between pre and post treatment, they can select better message to dentists. Otherwise, system will move on to
therapy scheme. DDES can extract the midline of root diagnostic processes. DDES will automatically reason
canal and analyze deviation, as if figure 6. by using diagnosis knowledge and rules stored in
knowledgebase. After successful matching completed,
DDES will identify the class and name of dental
disease and the system will generate conclusions of the
diagnosis, treatment methods and prescriptions. At the
same time, the explaining module will get all records
and information about the inference processes. The
diagnosis processing and result will display in HTML
file format on the interface, so the dentists may clearly
know diagnosis of DDES.

365
The working process of the DDES is shown in This paper has been supported by Graduate
Figure 7. Innovative Foundation of Shanghai University.
Authors are also grateful to reviewers for the
thoughtful suggestions that helped improve the
manuscript.

References
[1] Duan Weihong, Rong Li, and Chen Xiangtao,
“Innovation and practice for clinical teaching content and
method of prosthodotics”, Beijing Journal of Stomatology,
1999,7(03), pp.151-152.
[2] RuePpel R A. Stream eiphers, “Contemporary, the
Science of Information”, IEEE Press, 1992, pp. 65-134.
[3]Ali Keles, Ayturk Keles, “Expert system for thyroid
diseases diagnosis”, Expert Systems with Applications, 2008,
pp. 242–246.
[4] Razzouk, J.J. Mari, I. Shirakawa, J. Wainer and D.
Sigulem, “Decision support system for the diagnosis of
schizophrenia disorders”, Schizophrenia Research, Volume
63, Issue 3, 1 October 2003, Pages 295-296.
[5] F. John, Firriolo DDS, and Terry Wang BSc, “Diagnosis
of selected pulpal pathoses using an expert computer
system”, Oral Surgery Oral Medicine Oral Pathology Oral
Radiology and Endodontic, Volume 76, Issue: 3, pp. 390-
396.
[6] Firestone, AR, Sema, D; Heaven, TJ, et al. “The effect of
a knowledge-based, image analysis and clinical decision
support system on observer performance in the diagnosis of
approximal caries from radiographic images”, CARIES
RESEARCH, 1998, Volume: 32 , Issue: 2 , pp. 127-134.
[7]Gu Xiaodong, “Intelligent Generation Technique of
Dental Medical Records Based on CBR”, Hefei University of
Figure 7. Working flowchart of DDES Technology master dissertation, 2007,5.
[8] Li Yani, “Oral Medicine Digital Image Processing and
5. Conclusion Assistant Treatment System”, Guangdong University of
Technology master dissertation, 2007, 5.
[9] Turban and J.E. Aronson, “Decision support systems and
In this paper, we have presented a knowledge-
intelligent systems “(6th ed.), Prentice Hall, NJ (2000).
based dental diagnostic expert system named DDES.
The DDES integrates Artificial Intelligence,
Knowledge Engineering, Database, Computer and
stomatology technologies. DDES is a high-efficiency
and practical computer aided medical expert system.
Our experience has shown the effectiveness of this
system. Specially, hybrid knowledge acquisition
strategy, complex query for combined conditions,
intelligent diagnosis of dental diseases and auxiliary
diagnosis module are characteristics in the DDES.
DDES can serve as not only an educational tool to the
students of stomatological colleges and practical
dentist but also an assistant to dentists diagnosing
dental diseases and inquire relative material. In a word,
DDES can improve quality and level of serves. It has a
wide application prospect and commercial value.

Acknowledgments

366

You might also like