THESIS
2006
xiv, 116 leaves : ill. ; 30 cm
Abstract
The aceto whitening phenomenon, first described by Hinselmann [1], is an important diagnostic tool in colposcopy to differentiate between normal and abnormal areas of the cervical epithelium. It was noticed that the changes associated with cervical dysplasia could be seen more clearly after the application of acetic acid solutions. In general, abnormal areas appear white, while normal areas do not change in appearance. Over the past 20 years, colposcopy has become the standard method for the investigation of abnormal smears [2] now using 5% acetic acid to visualize abnormal areas as transient white patches. Images are taken during the diagnostic process and information is extracted to study the cervical epithelium. However, as the cervix would not stay completely stationary during the p...[
Read more ]
The aceto whitening phenomenon, first described by Hinselmann [1], is an important diagnostic tool in colposcopy to differentiate between normal and abnormal areas of the cervical epithelium. It was noticed that the changes associated with cervical dysplasia could be seen more clearly after the application of acetic acid solutions. In general, abnormal areas appear white, while normal areas do not change in appearance. Over the past 20 years, colposcopy has become the standard method for the investigation of abnormal smears [2] now using 5% acetic acid to visualize abnormal areas as transient white patches. Images are taken during the diagnostic process and information is extracted to study the cervical epithelium. However, as the cervix would not stay completely stationary during the process, the abnormal areas will appear in different parts of the image as the cervix moves. Much inconvenience is caused for registering the abnormal parts manually from the image. Especially when the doctor is tired or working under high pressure, false judgment may result. Therefore, an automatic tracking system can be very useful for the doctors.
more objectively and accurately.
In this work, a three-dimensional endoscopic imaging system is presented. The system consists of two channels. One of the channels, called the imaging channel, was used to obtain the image from the object of interest. The other channel, called the projection channel, was used to project a structured light pattern and generate feature points on the imaged object for the measurement of surface profile. The imaging and projection channels were calibrated and the distortions of the two channels were corrected using a modified pinhole camera model. The surface profile was extracted using the triangulation between the projected feature points and the channels of the endoscope.
The surface profile contains 3D information such as surface normal and angles between normal and is used for registrations of the motion of the surface. The imaging system was evaluated in tracking of several typical objects. The experimental results show that motion information of the objects with different shape and dimensions can be recovered with high accuracy. The motion information can then be used to establish correspondences in 2D images. Assuming small motions, the new algorithm developed is fast in speed and demonstrates a big potential in assisting medical analysis. Experiments on cervix-like objects were conducted which demonstrate the potential of the technology in clinical applications. The technology can also be useful in other similar medical analysis applications.
Post a Comment