Description

The classification and identification of the disease in medical images were helpful in biomedical applications. The process of segmentation of the diseased portion in the images were done based on Toboggan algorithm. The lung lobes were segmented from the input images based on the gradient estimation following original Toboggan algorithm. From the gradient estimated lung lesion inside the segmented lung lobes were extracted based on the improved Toboggan algorithm. Contours were extracted over the identified lung lesion regions. Lung CT images were taken as input for the process. Lung lobes were extracted from the images based on original Toboggan algorithm. In Toboggan algorithm the input image and the gradient of the input image is taken. The neighborhood pixels of the input images were compared and based on that the required portion in the input images were selected. The selected portions were grouped inorder to identify the lung lobes in the input images. The segmentation process is then applied for the disease affected regions in the images. The lung lobes were then segmented using improved Toboggan algorithm for the identification of the lung lesion in the identified portion. The defected portions were then outlined using contour. The performance of the process is measured based on True positives, True Negatives, False Positives, False negatives, Accuracy, Sensitivity, Specificity, Area Under Curve, Dice coefficient and Hausdorff Distance. Toboggan algorithm used in the process does not requires any previous knowledge about the structure of the lung lobes and lung lesions of the input images. The contour extraction in the input images were based on seed points selected using segmented portions from the Toboggan algorithm. The performances measured indicates that proposed method is more efficient. The identification of the diseased location in all the type of images is much needed which is more applicable for the designing of CAD models.