Lejla Alic, MSc

1
Monitoring of cancer treatment by spatiotemporal
pattern analysis of MRI
Lejla Alic, MSc www.bigr.nl/alic
2
Presentation outline

• Background
• Cancer facts
• Data example CE-MRA
• Research focus
• Data example DCE-MRI
• Temporal data behaviour
• Spatial data behaviour
• Approach
• Preliminary results

3
Research background
Goal improved tumor diagnosis and therapeutic
assessment Field biomedical image
analysis Trend shift from static to dynamic
imaging Problem data can no longer be analyzed
visually Situation lack of analysis techniques
and tools
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
4
Cancer facts macro level
Cancer is defined as a group of diseases
characterized by uncontrollable growth and
spread of abnormal cells.
Approximately 70.000 new cancer cases are
estimated for 2001in the Netherlands ?13
Cancer is the 2nd leading cause of death in the
Netherlands
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
5
Cancer facts macro level
NKR
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
6
Cancer facts micro level
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
7
Cancer facts micro level
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
8
Data example CE-MRA
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
9
Research focus
Permeability Spatiotemporal behaviour of the
CA leakage in the cancer EES
Vascularization Quantification of the vessel
morphology
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
10
Data example DCE-MRI

Time point N
Time point 1
Time point 2
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
11
Temporal data behaviour
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
12
Spatial data behaviour
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
13
Features extraction


Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
14
Features extraction
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
15
Data labeling
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
16
Data processing
Feature map
DCE-MRI data set
labeled ROIs
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
17
Preliminary results
Time to pick
Area under the curve
Slope in
Original data ROI
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
18
Approach
Temporal behaviour
Spatial behaviour
Classification
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
19
Heterogeneity
Heterogeneity in enhancement patterns important
descriptor of malignancy Quantification of
heterogeneity issue of research
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
20

Study design
Heterogeneity quantification methods Histogram
based analysis Principal Component Analysis
Texture based analysis methods (fractals)
Correlation with Histo-pathology Fluorescence
images Global outcome measures, tumor response
Available data Animal Model Patient Study
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
21
Preliminary results
Evaluation of isolated limb perfusion treatment
using pixel-wise data analysis of DCE-MRI M. van
Vliet, L. Alic, J.F.Veenland, A.M. Eggermont,
G.P. Krestin, C.F. van Dijke To be submitted to
Skeletal Radiology
Quantification of DCE-MRI spatiotemporal
heterogeneity Assessment of tumor blood
perfusion. L. Alic, M. van Vliet, CF,
J.F.Veenland To be submitted to J Magn Reson
Imaging
Background
Cancer facts
CE-MRA
DCE-MRI
Focus
Data behaviour
Approach
Results
22
(No Transcript)