Canscreen - SC IPA SA Sucursala Cluj

Download Report

Transcript Canscreen - SC IPA SA Sucursala Cluj

“Fail-safe” System for Diagnosis Quality
Control in Cancer Screening – Canscreen
Contract number 125 / 2006 financed by Romanian Government,
through CEEX 2006 - Romanian National Research & Development
Programme, VIASAN
Research domain: Health
Duration: september 2006 – september 2008
Consortium
Canscreen

CO - "Iuliu Haţieganu" University of Medicine and Pharmacy, Cluj-Napoca
www.umfcluj.ro
 project manager: prof. dr. Andrei Achimaş-Cădariu
►

P1 - "I. Chiricuţă" Oncology Institute, Cluj-Napoca
www.iocn.ro
 project responsable: dr. physician Florian Nicula
►

P2 - SC IPA SA Cluj Subsidiary
www.automation.ro
 project responsable: senior researcher eng. Ioan Stoian
►

P3 - "Babeş Bolyai" University, Cluj-Napoca
www.ubbcluj.ro
 project responsable : prof. dr. Liana Lupşa
►

P4 - “Al. Trestioreanu” Oncology Institute, Bucureşti
www.iob.ro
 project responsable: physician Milena Duma
www.automation.ro/canscreen
►
2
Project Objectives
Canscreen

Main objective - to find an appropriate cervical and breast cancer screening strategy for
Romania and thus, the reduction of incidence and mortality by cancer with the improvement of
quality of life for the patients.

Canscreen has the following scientifically and technological objectives:
 building up of a complex fail safe system which will ensure the diagnosis and treatment
quality at European Standards
 design of a database which will include the tested women and also registers with high risk
women
 design of algorithms for individualized diagnosis and a protocol for subsequent therapeutics
indications
 improving the smears interpretation
 analysis and selection of smears with cytological abnormalities
 insuring the quality of the system by random selection of a percent of samples of
interpreted smears and transmitting them to an expert cytologist
 distance transmission of mammography and ensuring their quality control
 designing the cost-effectiveness study which allows to choose the screening strategy with the
lowest ratio of cost-effectiveness using the Markov model
www.automation.ro/canscreen
3
Project Phases
Canscreen

Phase 1: Study of the actual situation in cancer pathology and of it's
monitoring in Romania ► 01.09.2006 – 30.10.2006

Phase 2: Definition of support methodology; ► 31.10.2006 – 30.04.2007

Phase 3: Designing the informatic system ► 01.05.2007 – 30.10.2007

Phase 4: Implementation and verification of the informatic system ►
31.10.2007 – 30.04.2008

Phase 5: Evaluation of preliminary results and results dissemination ►
01.05.2008 – 30.09.2008
www.automation.ro/canscreen
4
System Functions
Canscreen

Canscreen system functions:












access of “Canscreen” people monitoring – administrative personnel
defining physician dictionaries assigned to following medical facilities: family medicine, sampling
centers, cytological laboratory, image processing centers, etc
mailing screening invitations for a number of persons which respect some mandatory criteria and
generates needed documents for economical research of invitation process
establishing an individual schedule of sampling for persons that make a reservation depending on
availability/unavailability of the interpretation centers
gathering, in electronic screening format, of data for each registered patient, her cross thru
sampling center, laboratories, the result of investigation and the recommendation of the specialist
physician
recording prevailed images on the file server and on screening program’s database
images quality improvement and improper image files automatic elimination
viewing of round’s historical, episodes and invitations for a participant to screening program
automatic persons involve in screening programs planning according to the screening test result
screening persons’ monitoring in case of HPV and colposcopy testing to keep the possibility of
their reintegration in screening programs
developing specific annual statistic reports according to database’s post processing
data export to other related software applications 
www.automation.ro/canscreen
5
Medical Screening Protocols
Canscreen

Based on the medical protocols specific for the screening of cervical and
breast cancer defined by the methodology the technical specifications for the
database and for collecting and analyzing images was established.

Medical protocols define the system technical specifications:






establishing external source of medical data to import
establishing medical data necessary to the system
analysis of the quantification possibilities and coding for medical data
designing the hardware configuration necessary for the informatics system
establishing data flow from informatics point of view
establishing requirements for the cost-effectiveness study and algorithms
www.automation.ro/canscreen
6
Screening Programs Data Collection
Canscreen

Necessary steps in data collection process:




taking the smears
processing the smears and images
smears/images interpretation
data storing
www.automation.ro/canscreen
7
System Architecture
Canscreen


Hardware architecture – n-tier architecture with the levels:
 presentation level
 logical management level
 logical data access level
 database level
Hardware architecture is composed by:
 client workstations

WEB server

database and application server
files server

www.automation.ro/canscreen
8
System Architecture
Canscreen

System’s software architecture contains the modules:







interface design module accomplishes the user access to the level of the medical centers
involved in the screening program for the cervical and breast cancer (MPI)
medical data quantification and representation module (MRCD)
medical images acquisition module (cytological and mammographic images) (MAI)
medical data generation and exploitation module offers the possibility of retrieving data
according to certain criteria defined by the screening protocol. Another module function is
the pacients tratament management which involves special investigations.(MGEBD)
images filtering module intends to optimize the interpreting procedures for the cytological
smears and mammographies, by elaborating recognition algorithms for extracting the
quantifiable features (MFI)
quality module randomly selects a percentage of the analyzed images, which are afterwards
re-read and transferred to a cytology expert (MAC)
economical analysis module decides the screening strategy in accordance with the costefficiency ratio, by using Markov models applied to existing data in the database (MAE)
www.automation.ro/canscreen
9
System Architecture
Canscreen

Software modules interaction:
www.automation.ro/canscreen
10
System Database
Canscreen


“Canscreen” database consists of the following data collections:
 clinical information store - high-order relational database - encoded medical data involved in
the screening program in accordance with the guidelines for screening protocols
 images repository - low-order relational database - only the image address is memorised in
the database, not the entire image
“Canscreen” database is designed and implemented in Erwin and Microsoft SQL Server 2005.
low-order relational database
high-order relational database
www.automation.ro/canscreen
11
System Database
Canscreen
 Global relational entities model
www.automation.ro/canscreen
12

Application Program

Implemented in Microsoft Visual Studio
2005
Manages the system database and
implements the system functions
Canscreen
www.automation.ro/canscreen
13
Cytological Images Interpretation
Canscreen

Algorithms for cytological images interpretation :




algorithms used in automatic cellular formation segmentation (found in cytological images) (Otsu method)
automatic cellular formation segmentation found in cytological images using clustering algorithms (Moving
k means clustering algorithm)
post-processing segmented image using clustering algorithms (Modified Seed Based Region Growing algorithm)
automatic cellular formation segmentation found in cytological images using clustering algorithms and
segmentation algorithms based on global segmentation limit calculation
www.automation.ro/canscreen
14
Mammographic Images Interpretation
Canscreen

Algorithms for mammographic images interpretation:




multi scale segmentation algorithms
microcalcifies detection algorithms based on crossed entropy minimization
microcalcifies detection algorithms based on bidimensional simultaneous apparition matrix
microcalcifies detection algorithms based on conditional entropy maximization
www.automation.ro/canscreen
15
Markov Model
Canscreen





“Canscreen” database is used to do an
economic
evaluation
of
healthcare
interventions to ensure the quality of the
screening programs.
Medical decision analysis using Markov model:
 modeling the progression of a chronic
disease (cervical and breast cancer)
 predict effect changes in cervical and
breast cancer practice
State
Possible transactions
W
W, L, H, D, A
L
W, L, H, D, A
H
W, L, H, UC1, D, A
UC1
UC1, C1, UC2, D, A
UC2
UC2, C2, UC3, D, A
UC3
UC3, C3, UC4, D, A
States of Markov models from which it is
impossible the leaves are known as absorbing
states (overfullfilment of including age in
screening).
UC4
UC4, C4, UC2, D, A
D1
D1, D2, S, DC, D, A
D2
D2, D3, S, DC, D, A
All possible transitions between absorbing
states in a Markov model are given by a
transition matrix.
D3
D3, D4, S, DC, D, A
D4
D4, S, DC, D, A
S
S, D, A
DC
Absorbing state
D
Absorbing state
A
Absorbing state
Transaction matrix for a 15-state Markov model
in cervical cancer screening program

www.automation.ro/canscreen
16
Quantifiable Results
Canscreen

Direct quantifiable results:





developing a model of a fail-safe system of diagnosis control in screening program
developing a database of data and images that will contain information about identification
of tested women, medical parameters, and also registers of women with high risk
developing a study of economical analysis, a study of cost-effectiveness to decide the
screening strategy
impact on working people, specialization in informatics field of the medical personnel, about
10 persons
Indirect quantifiable results :


increasing the life expectancy of the patients, by elaborating individualized diagnosis
algorithms and prognosis with consecutive therapeutic indications
increasing the management capability of the project coordinator in leading interdisciplinary
projects
www.automation.ro/canscreen
17
Expected Beneficiaries
Canscreen

Oncology units from the public health network

The project can be extended, as a network at the level of rural localities, geographically
isolated or in the consulting rooms of family doctors and in health centers. Thus it will
increase efficiency of public health system and it will bring the Romanian health
system on European coordinates.
The project offers the following benefits:





access of high risk feminine population for developing breast or ovarian cancer to be
monitored according to European Community standards
access of feminine population to organized screening for precocious identification of
cervical cancer, using an optimal strategy
developing of a complex system to control for the diagnosis quality in the cancer screening
program
developing a study of economical analysis – a cost-effectiveness study to decide the screening
strategy
www.automation.ro/canscreen
18
Project Dissemination
Canscreen

List of published research papers in national and international conferences proceedings and jounals:








The Role of Cancer Registries in Suveillance and Cancer Care, Improvement of Data Quality in
Institutional Cancer Registry by Data Linkage, Luciana Neamţiu, Daniela Coza, Florian Nicula, Ofelia
Suteu, pp. 129, Poster, Octomber 2006, Amsterdam (Netherlands)
XII-th International Conference on Applied Stochastic Models and Data Analysis (ASMDA), Stochastic
Fractal Interpolation Function and its Applications to Fractal Analysis of Normal and Pathological
Body Temperature Graphs by Children, Anna Soos, May – June 2007, Chania, Creta (Grece)
Meditech 2007, Data Management System for Cervical Cancer Screening – CanScreen, Teodora
Sanislav, Ioan Stoian, Dorina Căpăţînă, Mihaela L. Drăgan, Luciana Neamţiu, Florian Nicula, Ofelia Şuteu,
Liana Lupşa, September 2007, Cluj-Napoca (Romania)
ENUMATH 2007, Multiple Dynamic Programming Applied in Cervical Screening, Luciana Neamţiu,
Ioana Chiorean, Final Program and Abstracts, September 2007, Graz (Austria)
ENUMATH 2007, Special Efficient Solutions for Multicriterial Transport Problems, Liana Lupşa,
Ioana Chiorean, Dorel I. Duca, Luciana Neamţiu, Final Program and Abstracts, September 2007, Graz
(Austria)
IACR 29TH ANNUAL MEETING, The Role of Cancer Control Department in Our Comprehensive
Cancer Center – Oncological Institute „Ion Chiricuţă”, Daniela Coza, Florian Nicula, Ofelia Şuteu,
Luciana Neamţiu, Ioana Rădulescu, Mihaela Sârbu, Lidia Lovasz, September 2007, Ljubljana (Slovenia)
Seminar of Functional Equation, Approximation and Convexity, Lex-Min Assignment Problem, Luciana
Neamţiu, Radu-Lucian Lupşa, Volume 5, Septembrie 2007, Cluj-Napoca (Romania)
Tiberiu Popoviciu Seminar of Functional Equation, Approximation and Convexity, Posibilităţi de aplicare
în medicină a măsurilor de neconvexitate, Liana Lupşa, September 2007, Cluj-Napoca (Romania)
www.automation.ro/canscreen
19