New Concepts for Knowledge based Cataract Surgery Assistance

A First Clinical Approach

K. P. Scherer

, C. Rieder

, C. Henninger

, J. Baumeister

and J. Reutelshöfer

Karlsruhe Institute of Technology, Institute for Applied Computer Science,

76344 Eggenstein Leopoldshafen, Hermann von Helmholtz Platz 1, Germany

Denkbares GmbH, Friedrich Bergius-Ring 15, 97076 Würzburg, Germany

Keywords: Knowledge Acquisition, Computer Aided Assistance, Semantic Network, Knowledge Representations,

Knowledge Continuum, Expert System, Decision Support, Surgical Intervention.

Abstract: Concerning optimization of ophthalmosurgical interventions at human eyes, a knowledge based assistance

system is developed to support the decision process of the surgeons before the real operation at the human

eye. This is performed especially for cataract operations, when the real dark lens has to be removed and an

artificial lens system must be implanted. Especially for complicated and complex interventions the

computer-based system will be a helpful tool. The system has to guarantee a fast and efficient access to all

operation relevant information. The result of such a concept should be an enhancement of operational

quality and so an enhancement of the patient contentment. This means a very important factor for the human

life quality.

1 INTRODUCTION

In the biomechanical system of the human eye, the

intraocular lens is the most important component for

the refraction process with about 30 dioptres to

focus the rays, coming from outside to the retina.

Parallel to the human aging process there is no

possibility to prevent a fix dark cloudy lens and also

there exist no medical treatments. Cataract surgical

interventions at human eyes are the single method,

to replace the old dark cloudy human lens by a new

clear artificial lens (Augustin, 2001). This is

performed about 600.000 per year only in Germany

with increasing number. On the market, a lot of

different lens systems with different haptics, optics,

materials, and power refraction values complicate

the selection of a best system

A special configuration and selection of a patient

related lens system can only guaranteed by a

computer-aided decision support system, which

considers the patient’s need regarding all his earlier

diseases as well as the special clinical measurement

equipment

and the fabrication-related intraocular

lens (IOL) features.

Based on computer logical reasoning, the human

surgeon can plan and simulate his intervention by

the software before the action takes place at the real

patient. Figure 1 shows the way from planning the

surgical intervention through regarding the wide

spectrum of lens types to a simulated surgical

process. Dependant on the simulation results, the

operation plan can be changed and optimized.

Figure 1: Planning the special implantation.

dark cloudy

lens

lot of lens

types

clear artificial

implant

op planning

surgical process

P. Scherer K., Rieder C., Henninger C., Baumeister J. and Reutelshöfer J..

New Concepts for Knowledge based Cataract Surgery Assistance - A First Clinical Approach.

DOI: 10.5220/0004516100640068

In Proceedings of the 8th International Joint Conference on Software Technologies (ICSOFT-EA-2013), pages 64-68

ISBN: 978-989-8565-68-6

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

2 KNOWLEDGE DOMAINS

The human expert knowledge is subdivided into four

super classes of knowledge domains:

The patient situation includes all personal data

including his history with all past diseases and his

actual health state. In this situation the patient

related best IOL-implant has to be found. But a lot

of intraocular lenses are available, different in

functionality, form, material, haptics, design, and

refraction power values. To select the best option for

the patient is not trivial.

Additionally, many clinical measurement and

diagnostic methods are available and are applied in

the preoperative situation of the human being. Some

measured values can be wrong when for instance

past diseases are not known.

An additional knowledge domain is the wide

field of the different surgical techniques. The

application of these methods depends also on the

patient situation, the designed IOL, and the

determined diagnostics. This complexity yields a

computer-based structuring of the knowledge to

support the surgeons by logical reasoning including

the explanation of the reasoning process (Figure 2).

These four domains (patient’s situation,

diagnostics, IOL types, and the surgical techniques)

correlate among each other, so the dependencies are

very complex and they are no longer linear. A

formalized description of the different knowledge

domains must be performed. They should be

regarded before the final selection of the IOL and

the implantation at the real patient is done. The

elements of the knowledge domains and the

correlations between them are represented as a

semantic net with predicates, i.e., the ontology. This

structure, especially in combination with an efficient

visualisation is a very important help for the

surgeons.

Figure 2: The different knowledge domains.

3 SYSTEM ARCHITECTURE

A knowledge-based approach is well suited to make

available as well the human expertise as the

background deep knowledge to map the logical

reasoning process to computer software (Studer et

al., 1988). So, the need is to find out formal

structures to manage the different knowledge

domains. The knowledge-based support system is

conceptualized and realized in different semantic

layers:

Figure 3: Layer based architecture of the system.

On the basic layer L1, a frame-based information

system is performed with all features of the different

ontologies. The descriptive layer for the interested

information categories is a hierarchical class –

subclass system with a refinement process and

special specification from class to subclass. An

ontology concept consists of the declarative part of

the semantics, the meaning in this knowledge-based

approach, the sources (publications, authors and so

on), and multimedia representations (pictures, tables,

movies) (Benjamin et al., 1998).

On the layer L2, the concept of the semantic

dependencies is subdivided into different types of

relations (predicates), based on natural language.

The knowledge elements of layer L1 can be

correlated by so called weak links, that means

semantic associations between the different items.

They are used to generate a suggestion for the

surgeon, also to consider the semantic neighboured

knowledge domain when regarding a first one. The

surgeon can be guided from one interest of point to

the next in an intelligent manner. An additional type

of relation is performed by a stronger relation like a

rule in a rule-based expert system. If the premises of

the first domain are given, then the second

knowledge items result as conclusion from the

preconditions.

Based on the static description of the concepts in

L1 including their correlations in L2 the reasoning

process in layer L3 depends on the different

OP methods

patient

diagnostics

IOL types

L1: frames of the domains:

IOL types, surgical methods,

diagnostics, patient situation

L2: correlations, dependencies

L3: knowledge processing and reasoning

L4: user interface, explanation, acquisition

NewConceptsforKnowledgebasedCataractSurgeryAssistance-AFirstClinicalApproach

correlation types. Weak links can be regarded as a

helpful navigation to neighboured interested

knowledge items in the knowledge base. Strong

links can be used under consistent inference rules in

the predicate calculus 1st order to generate necessary

conclusions in a logical way.

In layer L4 the formulized knowledge is

correlated with the surgeon’s natural language by the

man machine interface. As well the acquisition of

new cataract specific knowledge must be generated

in a user specific efficient manner. Also the

explanation of the performed reasoning process is a

very important tool for controlling the human

decision process.

4 SURGEONS INTELLIGENT

SUPPORT

Concerning an enhancement and optimization of the

decision making process of the ophthalmic surgeons,

there is a need to correlate all information getting

from the patient situation, the latest lens

characteristics, and the instantaneous powerful

methods of the diagnostic equipment.

The assistance system allows for an intelligent

user machine communication (surgeon - assistance

system). The information part of the knowledge

system is combined by an intelligent navigation

through the available knowledge domains of the

cataract surgery net. The semantics is represented by

weak and strong links to associated knowledge

domains.

So the intelligent assistance tool for the cataract

surgery has the following advantages when

compared classical software information systems:

- The knowledge base can be extended using a

comfortable graphical knowledge acquisition

component in an easy and consistent manner.

- Furthermore the assistance system supports the

reasoning process of the surgeon and it can

identify the patient-related optimized implantation.

In this sense, a very expensive surgical

intervention can be reduced and the contentment of

the patient will be improved.

- A further available feature is the possibility to use

logical forward and backward chaining

mechanisms in the associated knowledge network

to explain the logical dependencies.

Regarding the above mentioned features, the system

is predestined as a tutorial system for inexperienced

doctors and also a decision support system for

experienced surgeons. In comparison to literature

and books the knowledge-based assistance is a

dynamical system, which can be updated and

extended by the latest cognitions from the experts.

The knowledge base of the applied support

system is considered under two points of view:

Figure 4: The path from the knowledge continuum to

intelligent navigation.

The basic fundament and starting point for

modelling the layer 1 through layer 4 is the

knowledge continuum with the features in the upper

part of Figure 4. In layer 1 to layer 4 the different

knowledge representations are shown.

The frames represent the static multimedia

information (text, pictures, and audio) including the

source information (i.e. http-addresses, literature,

and links). The context and associatives represent

und describe the semantics between the different

information items by weak links, i.e., the

associations.

The final aim and the arrived advantage of this

modelled knowledge approach are demonstrated in

the lower part of Figure 4. Based on these features,

this tool is responsible for an intelligent context

sensitive navigation.

multimedia info (texte, mouvie, audio)

representation of semantics

representation of strong relations

visualisation of the semantics

(temporal, causal, free predicates)

knowledge continuum

 acquisition of of all surgical relevant

knowledge elements

 research and development of wikis

and semantic net structures

 formalisation and structuring of the

knowledge elements

intelligent navigation

 support of the surgeon by relevant

information

 hiding of non relevant information

 easy access to picture, texte audio



eas

access to the semantic correlations

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies

5 REALISATIONS

The presented approach is realized by using the

system KnowWE (Baumeister et al., 2011b).

KnowWE is a semantic wiki, i.e., a web-based

knowledge system, extending textual and

multimedia information by ontological annotations

and strong-problem solving methods. Following the

idea of the knowledge formalization continuum

(Baumeister et al., 2011a), all types of knowledge

(formal knowledge but also text and multimedia) are

considered to be part of the knowledge base. In

general, a semantic wiki allows for the collaborative

authoring of the included knowledge base and its

collaborative use.

At the beginning of the project, the ontology was

defined for describing the relevant terms and actions

of the cataract domain. In the next step, the relevant

chapters of an established German textbook for

cataract surgery were transferred to the wiki system

in order to provide suitable information of the

domain.

Figure 5: The WISSASS wiki system (original German

screenshot).

Consequently, the textbook corpus was enriched by

ontological annotations populating the initially

defined ontology. In all project phases, different

domain specialists (surgeons) and knowledge

engineers were actively involved. Due to the nature

of the wiki system, the domain specialist is able to

extend the current body of knowledge by further

elements, when necessary.

From the user's view, the wiki distinguishes

restricted knowledge acquisition pages and public

system pages. The restricted knowledge acquisition

pages are only accessible by the knowledge

engineers and the surgeons and include the formal

knowledge definitions. The public pages are

accessible for everyone and they include the

textbook knowledge and the interfaces for browsing

the cataract ontology.

Figure 5 depicts a public page, where the concept

Phakoemulsifikation is described, i.e., a special

method of surgical intervention. The text describes

the requirements and consequences using this

method but also practical advices for implementing

it. From this description, the ontological

neighborhood of the concept can be browsed and the

user can navigate to interesting concepts easily.

Figure 6 shows an excerpt of the visualization for

the concept Phakoemulsifikation with neighboring

ontological concepts having a refining relation.

Figure 6: An excerpt of the ontology visualization (in

German language).

In the future, we are planning to extend the range of

visualization methods to provide more appropriate

navigation structures for the specific browsing

requirements of the users. At the current state of

development, the system includes about 400 pages

providing the textbook knowledge, annotated by

about 360 concepts with about 430 relations. At the

moment, clinical partners evaluate the system for the

last three months.

6 CONCLUSIONS

The most important advantage of the knowledge-

based assistance is the intelligent navigation of the

user (surgeon) through the knowledge continuum.

Based on the visualized network of the semantics

(knowledge elements and their correlations) the

surgeon gets exactly the information that is useful

and necessary for him in the instantaneous situation.

The semantics is of higher order because of

representation of weak and strong links to the

associated knowledge domains. In contrary to

classical information systems the expert system is

NewConceptsforKnowledgebasedCataractSurgeryAssistance-AFirstClinicalApproach

easy to extend by graphical knowledge acquisition

features. Also the reasoning process for decisions

can be explained by the visualisation of the logical

path. This assistance yields a better human decision

process and improves the results of the cataract

operation.

ACKNOWLEDGEMENTS

The research and development of the proposed

cataract surgical assistance system is funded by the

national economical government of Germany. It is

proposed to install a commercial clinical system

based on the results of the presented prototype at

clinical partners. The authors also thank the project

executing organisation AiF in Berlin, which is

responsible for the allocation of the budget to the

research centre and the commercial company.

REFERENCES

Augustin, A. J., 2001, Augenheilkunde, Springer Verlag,

Berlin Heidelberg, NewYork, ISBN 3-540-65947-1

Studer, R., Benjamins, R., Fensel, D., 1988, Knowledge

engineering in Principles and methods, Data and

Knowledge engineering Vol 25, p 161-197

Benjamins, V. R., Fensel, D., Gomez Perez, A., 1998,

Knowledge management through ontologies,

Proceedings on the. Int. conf. on Praktical aspects of

Knowledge management (PAKM 98), Basel, Schweiz,

Scherer, K. P., 2009, Hypothesis Generation in the context

of an ophthalmic application, Intern. Conf. on applied

computer science., pp 130, Genua, Italy, Oct. 2009

Baumeister, J., Reutelshoefer, J, Puppe, F., 2011a,

Engineering Intelligent Systems on the Knowledge

Formalization Continuum, International Journal of

Applied Mathematics and Computer Science (AMCS),

Vol. 21

Baumeister, J., Reutelshoefer, J., Puppe, F., 2011b,

KnowWE: A Semantic Wiki for Knowledge

Engineering, Applied Intelligence, Vol. 35, 323-344

ICSOFT2013-8thInternationalJointConferenceonSoftwareTechnologies