Image analysis for patient management in colorectal cancer
<p>Colorectal cancer is the second most common form of cancer in the western world and kills over 400,000 people each year worldwide. Early and accurate diagnosis is critical in order to assess the risks and determine the best course of treatment.</p><p>This thesis introduces a num...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2006
|
| Subjects: |
| _version_ | 1817932876083101696 |
|---|---|
| author | Bond, S Bond, Sarah |
| author2 | Brady, M |
| author_facet | Brady, M Bond, S Bond, Sarah |
| author_sort | Bond, S |
| collection | OXFORD |
| description | <p>Colorectal cancer is the second most common form of cancer in the western world and kills over 400,000 people each year worldwide. Early and accurate diagnosis is critical in order to assess the risks and determine the best course of treatment.</p><p>This thesis introduces a number of image analysis methods that can be used in such patient management decisions. The first stage is to remove the distortion caused by field inhomogeneities from the images. Having done this, a range of segmentation techniques are introduced, to automatically find the colorectum and mesorectum. They can then be visualised in three dimensions. A search for the lymph nodes is also implemented, and finds small regions within the mesorectum, then classifies them. A likelihood is assigned to each region that looks like a node, as to whether it is in fact likely to be cancerous.</p><p>Another major factor in patient management decisions is the changes that have occurred during chemo/radiotherapy, prior to surgery. We have developed a nonrigid registration method that can be used to compare these sets of images in a quantitative, accurate and useful way. We tested a variety of state-of-the-art, 'generic' non-rigid registration algorithms and found that they almost always fail to register the image sets, due to the large scale deformations that have occurred over the course of treatment. This thesis describes two key ways in which these algorithms can be improved by incorporating anatomical knowledge, and physiological knowledge.</p><p>Firstly, we set out a method of finding and representing the anatomical features in the images that are always present, such as the bones, the bladder and the colorectum. These representations are then used to constrain the general algorithms. This increases the robustness, such that the algorithms ran successfully on all datasets tested, that is, they provided a clinically useful registration result.</p><p>Secondly, we incorporate knowledge of the physiology, or how we expect the anatomy to change due to treatment. We can represent these changes using the Jacobian of the deformation, which describes the local size and type of change. This is used to regularise the registration, and can be incorporated simultaneously with the iterations of the registration. The final result is an accurate and robust registration result that is clinically useful for finding corresponding features on pre- and post-treatment datasets.</p> |
| first_indexed | 2024-03-06T21:15:04Z |
| format | Thesis |
| id | oxford-uuid:3f810bd0-2645-420c-9d91-169d7605b0bc |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-12-09T03:44:52Z |
| publishDate | 2006 |
| record_format | dspace |
| spelling | oxford-uuid:3f810bd0-2645-420c-9d91-169d7605b0bc2024-12-07T16:30:15ZImage analysis for patient management in colorectal cancerThesishttp://purl.org/coar/resource_type/c_db06uuid:3f810bd0-2645-420c-9d91-169d7605b0bcImage analysisTreatmentColon (Anatomy)CancerEnglishPolonsky Theses Digitisation Project2006Bond, SBond, SarahBrady, MBrady, J<p>Colorectal cancer is the second most common form of cancer in the western world and kills over 400,000 people each year worldwide. Early and accurate diagnosis is critical in order to assess the risks and determine the best course of treatment.</p><p>This thesis introduces a number of image analysis methods that can be used in such patient management decisions. The first stage is to remove the distortion caused by field inhomogeneities from the images. Having done this, a range of segmentation techniques are introduced, to automatically find the colorectum and mesorectum. They can then be visualised in three dimensions. A search for the lymph nodes is also implemented, and finds small regions within the mesorectum, then classifies them. A likelihood is assigned to each region that looks like a node, as to whether it is in fact likely to be cancerous.</p><p>Another major factor in patient management decisions is the changes that have occurred during chemo/radiotherapy, prior to surgery. We have developed a nonrigid registration method that can be used to compare these sets of images in a quantitative, accurate and useful way. We tested a variety of state-of-the-art, 'generic' non-rigid registration algorithms and found that they almost always fail to register the image sets, due to the large scale deformations that have occurred over the course of treatment. This thesis describes two key ways in which these algorithms can be improved by incorporating anatomical knowledge, and physiological knowledge.</p><p>Firstly, we set out a method of finding and representing the anatomical features in the images that are always present, such as the bones, the bladder and the colorectum. These representations are then used to constrain the general algorithms. This increases the robustness, such that the algorithms ran successfully on all datasets tested, that is, they provided a clinically useful registration result.</p><p>Secondly, we incorporate knowledge of the physiology, or how we expect the anatomy to change due to treatment. We can represent these changes using the Jacobian of the deformation, which describes the local size and type of change. This is used to regularise the registration, and can be incorporated simultaneously with the iterations of the registration. The final result is an accurate and robust registration result that is clinically useful for finding corresponding features on pre- and post-treatment datasets.</p> |
| spellingShingle | Image analysis Treatment Colon (Anatomy) Cancer Bond, S Bond, Sarah Image analysis for patient management in colorectal cancer |
| title | Image analysis for patient management in colorectal cancer |
| title_full | Image analysis for patient management in colorectal cancer |
| title_fullStr | Image analysis for patient management in colorectal cancer |
| title_full_unstemmed | Image analysis for patient management in colorectal cancer |
| title_short | Image analysis for patient management in colorectal cancer |
| title_sort | image analysis for patient management in colorectal cancer |
| topic | Image analysis Treatment Colon (Anatomy) Cancer |
| work_keys_str_mv | AT bonds imageanalysisforpatientmanagementincolorectalcancer AT bondsarah imageanalysisforpatientmanagementincolorectalcancer |