Biomedical Engineering Reference
In-Depth Information
location of seizures focus for patients with epilepsy, and
accurate assessment of tumor sites in the brain for surgery planning.
In practice, segmentation of PET brain data is difficult to perform in an au-
tomatic fashion because of the poor contrast and high noise level in the im-
ages. A standard procedure to delineate structures from PET brain images is
to segment structures from the corresponding anatomical magnetic resonance
images and then to superimpose them on the PET images. This method re-
lies on an accurate registration between the two imaging modalities, handling
the incongruity of structures and functions. Thresholding methods can pro-
vide a direct way to segment PET images with locally uniform radioactivity
concentration and consistent structures. For more noisy data, advanced meth-
ods such as deformable models have been studied for segmentation of cortical
structure [77]. These methods must adapt to changes in individual radioactivity
concentrations.
2.4.1.2
Segmentation of Cardiac Images
A reliable noninvasive imaging modality is essential for evaluating and monitor-
ing patients with cardiac disease. Traditional screening techniques for quantita-
tive assessment of cardiac function include the following modalities:
Multigated Angiography (MUGA)
: This is a slow screening modality that
requires the injection of a radiopharmaceutical agent by a clinician. The
purpose of MUGA screening is to examine the pumping function of the
heart. After injection of a radioactive agent that labels red blood cells, a
nuclear camera creates an image of the heart's chambers by counting these
cells over numerous cardiac cycles. The information obtained can quantify
ejection fraction but not ventricular volumes.
Magnetic Resonance Imaging (MRI)
: Because of its complexity and even
though MRI machines abound in the United States, cardiac MRI has largely
been limited to university hospitals where there is a strong interest in
research. This screening modality has proven very useful in evaluating
patients' cardiac anatomy prior to surgery, in locating and characterizing
cardiac tumors and in identifying and treating cardiac abnormalities for
children with complex congenital heart disease. These clinical situations
