| Oral lesions are important findings on computed tomography images. They are difficult to detect on CT images because of low contrast, arbitrary orientation of objects, complicated topology and lack of clear lines indicating lesions. In this thesis, a fully automatic method to detect oral lesions from dental CT images is proposed to identify (1) Closed boundary lesions and (2) Bone deformation lesions. Two algorithms were developed to recognize these two types of lesions, which cover most of the lesion types that can be found on CT images. The results were validated using a dataset of 52 patients. Using non training dataset, closed boundary lesion detection algorithm yielded 71% sensitivity with 0.31 false positives per patient. Moreover, bone deformation lesion detection algorithm achieved 100% sensitivity with 0.13 false positives per patient. Results suggest that, the proposed framework has the potential to be used in clinical context, and assist radiologists for better diagnosis. |
