AUTOMATIC LUNG NODULE SEGMENTATION USING AUTOSEED REGION GROWING WITH MORPHOLOGICAL MASKING (ARGMM) AND FEATURE EXTRACTION THROUGH COMPLETE LOCAL BINARY PATTERN AND MICROSCOPIC INFORMATION PATTERN

An efficient Autoseed Region Growing with Morphological Masking(ARGMM) is implemented in this paper on the Lung CT Slice to segment the 'Lung Nodules', which may be the potential indicator for the Lung Cancer. The segmentation of lung nodules carried out in this paper through Multi-Thresholding, ARGMM and Level Set Evolution. ARGMM takes twice the time compared to Level Set, but still the number of suspected segmented nodules are doubled, which make sure that no potential cancerous nodules go unnoticed at the earlier stages of diagnosis. It is very important not to panic the patient by finding the presence of nodules from Lung CT scan. Only 40 percent of nodules can be cancerous. Hence, in this paper an efficient Shape and Texture analysis computed to quantitatively describe the segmented lung nodules. The Frequency spectrum of the lung nodules is developed and its frequency domain features are computed. The Complete Local binary pattern of lung nodules is computed in this paper by constructing the combine histogram of Sign and Magnitude Local Binary Patterns. Local Configuration Pattern is also determined in this work for lung nodules to numerically model the microscopic information of nodules pattern.