| Summary: | 碩士 === 國立臺灣科技大學 === 材料科學與工程系 === 101 === In the medical field, there are many methods to examine the internal structures and organs of human body, such as X-ray, computed tomography (CT), magnetic resonance image (MRI), and microtomography (micro-CT), which can display different parts of the tissues or symptoms, to assist in diagnosis. This study used the image processing technology to automatically identify the inferior turbinate and maxillary sinus. After capturing the computed tomography signals, the grey level processing was performed on the image. The image was normalized, and the parametric template matching was used to capture the similarity eigenvalues specific to the inferior turbinate and maxillary sinus areas. The waveform eigenvalues of the inferior turbinate area were calculated, and entered into the back-propagation neural network (BPNN) training network database in order to determine whether there ROI exists. If yes, the level set method is used to circle the contour box of the inferior turbinate and maxillary sinus, and the volume was fit by the regression equation. According to the verification of the results of the intelligent and automatic CT system of inferior turbinate and maxillary sinus volume identification and calculation, in comparison to the areas circled by the physicians, the system reliability reached 0.86 and the overall accuracy rate was 97.92%. The proposed system is proven to be able to assist physicians in diagnosing the state of illness of inferior turbinate and maxillary sinus.
|
|---|
