Biomedical Engineering Reference
In-Depth Information
observations in the literature. However, we have for the first time a combined and
quantitative model of all the changes together. Of course, the number of parameters
that could influence the shape of the heart is so large that we would need much
more subjects in order to obtain an accurate and reliable quantitative model of the
heart remodeling. Nevertheless, the shape analysis pipeline we proposed is more
than a black box that learns the statistical relationship between sets of variable: it
gives a generative shape model that can be explored. For instance, the visualization
of the remodeling (or the deformation correlated to other clinical variables) could
allow to draw some intuition about the potential underlying links. Ideally this would
lead to a simplified model that could be tested against new data, thus gradually
improving the knowledge that we have about the disease. As the anatomy is the
support of the physiology, one of the goals is now to couple the anatomical changes
with the physiological evolution in order to better understand how one influences the
other.
Other interesting clinical questions related to rToF could be investigated using
our approach. For instance, there is nowadays a clinical consensus that a longer
QRS complex duration measured on the electrocardiogram (see for instance Fig. 3.3
in Chap. 3) correlates with RV dilation. Correlating this parameter with the RV shape
could reveal how the abnormal conduction impacts the RV anatomy and function.
Similarly, studying body-mass index (BMI) jointly with BSA could provide a more
comprehensive representation of the patient growth. One could apply the PLS
method between two sets of multivariate variables (BMI, BSA and QRS on the
one hand, the shape vectors on the other hand) and get a growth model that explains
both features. Unfortunately, QRS durations and BMI were not available for all
the patients as the population was retrospective, keeping these questions for future
work. The effect of genetic factors that regulate myocardium stiffness on the long-
term RV remodeling could also be analyzed. Patients with stiffer myocardium are
known to be more protected against RV dilation. These patients may be the outliers
of the model with “abnormally” smaller RV than the average. The decision for
valve implant may be delayed and based on different features for these patients.
Finally, tools are now been developed and integrated in several hospitals to keep
track of the patient records during a long time-period, which opens the possibility
to now correlate the shape features directly with the clinical outcome. Such a shape-
based model of the outcome, when validated, would be immediately usable by the
clinicians as a computer aided diagnosis tool.
5.5
Online Resources
Integrated Models of Paediatric Heart Diseases
http://www-sop.inria.fr/asclepios/projects/Health-e-Child/DiseaseModels/content/cardiac/
This site summarizes some modeling issues in pediatric cardiac diseases. It explains
the approach that was adopted in the European FP6 Health-e-Child project.
Search WWH ::
Custom Search