Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery
Real time tracking and high accuracy and repeatability localization of the patients’ kidneys and its corresponding motions are considered challenges for modem non- invasive surgeries of the kidneys. A suitable tracking algorithm is implored that is capable of tracking an object within an image by ma...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/53044 |
| _version_ | 1811692828969926656 |
|---|---|
| author | Yao Derek Castro |
| author2 | Sunita Chauhan |
| author_facet | Sunita Chauhan Yao Derek Castro |
| author_sort | Yao Derek Castro |
| collection | NTU |
| description | Real time tracking and high accuracy and repeatability localization of the patients’ kidneys and its corresponding motions are considered challenges for modem non- invasive surgeries of the kidneys. A suitable tracking algorithm is implored that is capable of tracking an object within an image by masking it with a kernel and defining the feature characteristics of the object like color information. It would then represent the target or the object being tracked as a set of the feature histograms based on its Epanechnikov profiles. This would in turn induce spatially-smooth similarity functions based on a metric measurement using Bhattacharyya coefficient and apply weights to the kernels to create a basin of attraction of the local maxima. The weights would be sufficient to implement localization using a gradient optimization method or the mean shift algorithm and guide the algorithm to converge into a new centre of the kernel which is actually the new position of the object in the current image frame. The tracking system was proven to be robust and effective in tracking two or more distinguishable feature of the kidney at the same time. A correlation was made between the tracking results to the known kidney moments. The tracking mechanism in ultrasound video analysis system is imperative for the successful development of an assisting technology for detection and analysis of kidney motion in the future. |
| first_indexed | 2024-10-01T06:42:00Z |
| format | Thesis |
| id | ntu-10356/53044 |
| institution | Nanyang Technological University |
| language | English |
| last_indexed | 2024-10-01T06:42:00Z |
| publishDate | 2013 |
| record_format | dspace |
| spelling | ntu-10356/530442023-03-11T17:05:01Z Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery Yao Derek Castro Sunita Chauhan School of Mechanical and Aerospace Engineering DRNTU::Engineering::Mechanical engineering Real time tracking and high accuracy and repeatability localization of the patients’ kidneys and its corresponding motions are considered challenges for modem non- invasive surgeries of the kidneys. A suitable tracking algorithm is implored that is capable of tracking an object within an image by masking it with a kernel and defining the feature characteristics of the object like color information. It would then represent the target or the object being tracked as a set of the feature histograms based on its Epanechnikov profiles. This would in turn induce spatially-smooth similarity functions based on a metric measurement using Bhattacharyya coefficient and apply weights to the kernels to create a basin of attraction of the local maxima. The weights would be sufficient to implement localization using a gradient optimization method or the mean shift algorithm and guide the algorithm to converge into a new centre of the kernel which is actually the new position of the object in the current image frame. The tracking system was proven to be robust and effective in tracking two or more distinguishable feature of the kidney at the same time. A correlation was made between the tracking results to the known kidney moments. The tracking mechanism in ultrasound video analysis system is imperative for the successful development of an assisting technology for detection and analysis of kidney motion in the future. Master of Science (Biomedical Engineering) 2013-05-29T08:20:31Z 2013-05-29T08:20:31Z 2011 2011 Thesis http://hdl.handle.net/10356/53044 en 106 p. application/pdf |
| spellingShingle | DRNTU::Engineering::Mechanical engineering Yao Derek Castro Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery |
| title | Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery |
| title_full | Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery |
| title_fullStr | Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery |
| title_full_unstemmed | Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery |
| title_short | Real-time monitoring and control algorithm for kidney dislocation during non-invasive surgery |
| title_sort | real time monitoring and control algorithm for kidney dislocation during non invasive surgery |
| topic | DRNTU::Engineering::Mechanical engineering |
| url | http://hdl.handle.net/10356/53044 |
| work_keys_str_mv | AT yaoderekcastro realtimemonitoringandcontrolalgorithmforkidneydislocationduringnoninvasivesurgery |