Hari Krishnan P* Pages 1 - 7 ( 7 )
This paper puts forth an innovative methodology for measuring fetal volume using Ant Colony based Gabor filter for better segmentation of Ultrasound (US) images. Mostly the US images were prone to affected by speckle pattern of noise signals along with small quantity artifacts. The measurements of various fetal parameters are affected because of this noise pattern. Existing methods use conventional noise removal techniques which in one other way could lead to image data loss. In this proposed method Gabor filtering process is hybridized with Ant Colony Algorithm for better de-speckling to yield better output images for improved fetal parameter measurement.
Methods: The methodology involves innovative application of ACA, which are of heuristic type. They are highly successful particularly for combinatorial type of optimization. As ants follow a particular phenomenon of path finding by using their pheromone deposition, correspondingly the pixels of image is mapped through this technique for a desired output for a better result. The above process is further enhanced by pre-filtering stage with Gabor mechanism.
Results: After carrying out a detailed study of earlier methods such as Perona P. and J.Malik , Isointensity contour and Euclidean shortening flow Iso-intensity contour, it is inferred that the proposed method is well ahead these earlier methods. The biggest error in P&M enhancement technique followed by Iso-intensity contour can be caused by the choice of the thresholding value. The salient feature of the proposed method is its optimizing capability of the various pixel values from those randomly varying noise signals. In such case, the filtering methodology would be able to distinguish the exact ROI (i.e. head and body portions of the baby of the fetus) from the rest of the portions. The average variations in terms of error are about 4.3% differing with absolute values of volume.
Conclusion: The extensive analysis carried out for volume calculation of fetus, has yielded a better results than earlier image processing techniques. After closely examining with the earlier techniques, it is evident that the earlier methodologies have serious lapses while dealing with noise pixels which drastically change the calculation results. Hence the proposed method solves the purpose with utmost accuracy.
Ant Colony Algorithm, Fetal, image processing, Speckle Noise, segmentation, Ultrasound, Gabor filtering.
Department of EEE, Anna University, Regional Campus-Coimbatore, Coimbatore