12-05-2011, 09:58 AM
[attachment=13605]
CHAPTER-1
IMAGE SEGMENTATION
1.1 INTRODUCTION TO THE PROJECT:
An image may be defined as two dimensional fumction as f(x,y), where x and y are spatial (plane) coordinates and the amplitude of f at any pair of coordinates (x,y) is called the intensity or gray level of the image at that point.
When x,y and the amplitude values of f are all finite discrete quantities , we call the image a digital image. The field of digital image processing refers to processing digital images by a digital computer.Elements are referred to as picture elements,image elements,pels and pixels.
Segmentation refer to the process of partitioning a digital imageinto multiple regions. The goal of segmentation is to simplify and / or change the representation of an image into something that is more meaningful and easier to analyze. Image segmentation is typically used to locate objects and boundaries (lines, curves) in images. The result of Image Segmentation is a set of regions that cover the entire image or a set of contours extracted from the image.
1.2 STATEMENT OF THE PROJECT:
A Voxel is a volume element representing a value on a regular grid in 3-D space. T his is analogous to a pixel , which represents 2-D image data voxels are frequently used in the visualization and analysis for medical and scientific data.
In 3-D space each of the co-ordinates is defined interms of its position,color and density. Think of a cube where any point on an outer side is expressed with an x,y and the third Z co-ordinate defines a location into the cube from that side , its density and its color wih this information and 3-D rendering software , a 2-D view from various angles of an image canbe obtained and viewed at our computer.
Figure 1-1: Voxel representation and data set of voxels for macromolecule
Medical Practitioners and Researchers are now using images defined by voxels and 3-D software to view X-rays cathode tube scans and Magnetic Resonance Imaging MRI) scans from different angles effectively to see the inside of the body from outside.
1.3 SCOPE OF THE PROJECT:
The main objective of Image Segmentation is to divide an image into regions that can be considered homogenious with respect to a given criterion such as color or texture. Segmentation is an essential part of any Image analysis system and especimedical Medical environments where segmented images provide valuable information for Diagnosis.
CHAPTER – 2
MEDICAL IMAGE SEGMENTATION
2.1 Medical Image Segmentation:
Medical image segmentation is the process of labeling each voxel in a medical image dataset to indicate its tissue type or anatomical structure. The labels that result from this process have a wide variety of applications in medical research and visualization. Segmentation is so prevalent that it is difficult to list the most oft-segmented areas, but a general list would include at least the following: the brain, the heart, the knee, the jaw, the spine, the pelvis, the liver, the prostate, and the blood vessels [20, 39, 35, 18, 25].
The input to a segmentation procedure is grayscale digital medical imagery, for example the result of a CT or MRI scan. The desired output, or “segmentation,” contains the labels that classify the input grayscale voxels. Figure 2-1 is an example of a very detailed segmentation of the brain, along with the original grayscale imagery used to create the segmentation.
The purpose of segmentation is to provide richer information than that which exists in the original medical images alone. The collection of labels that is produced through segmentation is also called a “labelmap,” which succinctly describes its function as a voxelby- voxel guide to the original imagery. Frequently used to improve visualization of medical imagery and allow quantitative measurements of image structures, segmentations are also valuable in building anatomical atlases, researching shapes of anatomical structures, and tracking anatomical changes over time.
Figure 2-1: Segmentation example. A grayscale MR image of the brain (left) and a detailed matching segmentation, also known as a label map (right).
The procedure followed to create the segmentation was partially automated, but a large amount of human effort was also required. The segmentation was initialized using an automatic gray matter/white matter/cerebrospinal fluid segmenter, and then individual neural structures were manually identified. This grayscale data set and segmentation were provided by Dr. Martha Shenton’s Schizophrenia Research Group at the Surgical Planning Lab at Brigham and Women’s Hospital.
2.2 Applications of Segmentation:
The classic method of medical image analysis, the inspection of two-dimensional grayscale images on a light box, is not sufficient for many applications. When detailed or quantitative information about the appearance, size, or shape of patient anatomy is desired, image segmentation is often the crucial first step. Applications of interest that depend on image segmentation include three-dimensional visualization, volumetric measurement, research into shape representation of anatomy, image-guided surgery, and detection of anatomical changes over time.
2.2.1 Visualization:
Segmentation of medical imagery allows the creation of three dimensional surface models, such as those in Figure 2-2, for visualization of patient anatomy. The advantage of a surface model representation of anatomy is that it gives a three-dimensional view from any angle, which is an improvement over two-dimensional cross sections through the original grayscale data [10]. Surface models can be created from segmented data using an algorithm such as Marching Cubes [26]. (Though three-dimensional models could be created directly from grayscale data using Marching Cubes, the segmentation step is used to provide the desired user-defined isosurfaces to the algorithm.)
2.2.2 Volumetric Measurement:
Measurement of the volumes of anatomical structures is necessary in medical studies, both of normal anatomy and of various pathological conditions or disorders. This is an obvious application of segmentation, since it is not possible to accurately measure anatomical volumes visually.
For example, in studies of schizophrenia, volume measurement is used to quantify the variation in neural anatomy between schizophrenic and control patients. Areas of interest in such studies include the lateral ventricles, structures in the temporal lobe such as the hippocampus, amygdala, and parahippocampal gyrus, the planum temporale, and the corpus callosum [27]. It is a time-intensive process to obtain accurate measurements of such regions, as the current method employs manual segmentation.
Volume measurement is also used to diagnose patients; one example is in measurement of the ejection fraction. This is the fraction of blood that is pumped out of the left ventricle of the heart at each beat, which is an indicator of the health of the heart and its pumping strength. To measure the ejection fraction, the blood in the left ventricle is segmented at different times in the cardiac cycle.
Figure 2-2: Example of three-dimensional surface models, created from segmented data
using the Marching Cubes algorithm.
These models were used in surgical planning and guidance. Each image is composed of five models: skin (light pink), neural cortex (light white), vessels (dark pink), tumor (green), and fMRI of the visual cortex (yellow). The fMRI, or functional MRI, shows areas of the brain that were activated during visual activities (areas which should be avoided during surgery).
2.2.3 Shape Representation and Analysis:
Various quantitative representations of shape are studied in order to mathematically describe salient anatomical characteristics. The first step in creating a representation of anatomical shape is segmentation: intuitively, one needs to know the structure’s position and the location of its boundaries before its shape can be studied.
One example of a shape representation is a skeleton, a construct which is similar to the centerline of a segmented structure. One way to imagine a skeleton is the “brush fire” approach: one thinks of simultaneously lighting fires at all points on the boundary of the structure. The fires burn in ward, traveling perpendicular to the boundary where they started, and then extinguish when they hit another fire. The connected “ash” lines left where the fires extinguish is the skeleton of the structure.
A richer shape representation is the distance transform, a function that measures the distance from each point in a structure to the nearest point on that structure’s boundary. The distance transform can also be imagined with the pyrotechnic approach: it is the time that the fire first reaches each point in the structure. Consequently it is considered richer than the skeleton, since it contains more information.
Presumably, shape representations will become increasingly useful in making quantitative anatomical comparisons. Distance transform shape representations have already been applied to the classification of anatomical structures in a study that aims to differentiate between the hippocampus-amygdala complexes of schizophrenics and normals . An example of grayscale MR image data and the shape representation derived from it for this study can be seen in Figure 2-3.
Shape representations can also be used to aid the segmentation process itself by providing anatomical knowledge. A generative shape model, once trained from a population of shape representations, can then be used to visualize new shapes according to the learned modes of variance in the shape population (allowing visualization of “average” anatomy and of the main anatomical variations that may occur). Then, at each step of the segmentation of new data, fitting the model to the current most likely segmentation can provide anatomical information to the algorithm.
2.2.4 Image-Guided Surgery:
Image-guided surgery is another medical application where segmentation is beneficial. In order to remove brain tumors or to perform difficult biopsies, surgeons must follow complex trajectories to avoid anatomical hazards such as blood vessels or functional brain areas. Before surgery, path planning and visualization is done using preoperative MR and/or CT scans along with three-dimensional surface models of the patient’s anatomy such as those in Figure 2-2.
Figure 2-3: Shape representation example.
A segmentation of the hippocampus-amygdala complex (left), a 3D surface model of the hippocampus-amygdala complex (center), and a distance map used to represent the shape of the hippocampus-amygdala complex (right).
During the procedure, the results of the preoperative segmentation may still be used: the surgeon has access to the pre-operative planning information, as three-dimensional models and grayscale data are displayed in the operating room. In addition, “on-the-fly” segmentation of real time imagery generated during surgery has been used for quantitative monitoring of the progression of surgery in tumor resection and cryotherapy . Figure 1-4 shows the use of preoperative surface models during a surgery.
2.2.5 Change Detection:
When studying medical imagery acquired over time, segmenting regions of interest is crucial for quantitative comparisons. The Multiple Sclerosis Project at Brigham and Women’s Hospital measures white matter abnormalities, or lesions, in the brains of patients suffering from MS. Because MS is a disorder that progresses over time, accurate temporal measurements of neural changes may lead to a better understanding of the disease. The stated goals of the MS project are analysis of lesion morphology and distribution in MS, quantitative evaluation of clinical drug trials, and monitoring of disease progression in individuals.
Figure 1-4: Surgical planning and navigation using surface models.
To this end, automatic segmentation is used to identify MS lesions, which appear as bright regions in T1- and T2-weighted MR scans of the brain, as shown in Figure 1-5. The volume of such lesions, as measured from segmented data, has been shown to correlate with clinical changes in ability and recognition. The “before” picture (top) shows several 3D surface models used in surgical planning, along with one grayscale slice from the MR scan that was segmented to create the models.
MS Lesions Automatic Segmentation
Figure 1-5: Multiple Sclerosis Lesions (seen as bright spots).
The green model is the tumor that was removed during the surgery. The “after” picture (bottom) shows an image that was scanned during surgery, at the same location in the brain as the top image. The yellow probe is a graphical representation of the tracked probe held by the surgeon. Automatic segmentation is used to track disease progression over time. Images were provided by Mark Anderson of the Multiple Sclerosis Group at the Surgical Planning Lab at Brigham and Women’s Hospital.
