To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
Computer Aided Diagnosis of Thyroid Cancer Using Image Processing Techniques |
|
Author |
Massoud Sokouti, Mohsen Sokouti, and Babak Sokouti |
| Citation |
Vol. 18 No. 4 pp. 1-8 |
|
Abstract |
One of the mathematical based techniques used for processing the medical images in order to identify the irregularity properties of these images is employing different types of Image processing algorithms. Detecting these abnormalities in terms of their risk of malignancy such as tumors in thyroid gland is of a major importance in nuclear medicine which is known as cold nodules. In this study, the radioisotope image of thyroid gland is used for extracting the cold nodules based on hot nodule extraction. For this purpose, two routes are employed which will be unified later for nodule extraction through a simple intelligent system. At First, the gray level of blue channel of the target image is mapped from range of [0,1] to [1,0] that a portion of the resulted image will be selected according to a predefined threshold. Second, the color image will be transformed into gray scale which will be then enhanced by circular average filtering and highlighting the intensity methods. Moreover, thresholding stage will also be applied to the obtained image in the second approach. After all both resulted images will be added and mapped will be then from range of [0,1] to [1,0]which will be ready for feature extraction using the simple hill climbing methodology. The results showed that after executing the process for hundred times the best area for cold nodule will be identified with high accuracy. In conclusion, it has been shown that by using image processing technique along with a simple intelligent system such as hill climbing algorithm, one can achieve the highest performance (i.e., accuracy of 0.9896) in detecting the malignancy in several medical related images. |
|
Keywords |
Image processing Thyroid Cancer Cold Thyroid Nodules Hill climbing Feature Extraction. |
|
URL |
