Application of geomatics techniques and genetic programming to medical image segmentation

Ultrasound imaging is a promising approach in medicine due to it being a non-intrusive technique, with the ability to obtain important data from a large number of patients. Despite its potential, images generated using this technique are still severely corrupted with high levels of noise, especially speckle, which hinders their further processing. The present study introduces an alternative, innovative method for medical image segmentation, combining a bilateral filter, a specially developed genetic programming algorithm, the CLAHE algorithm, and the Watershed segmentation technique, which is typically used for geotopographic images. The results from this methodology indicate that the proposed approach delivers results comparable to those of conventional neural networks, while requiring a smaller image dataset. The integration of the genetic algorithm offers a novel solution by enhancing local contrast, reducing image noise, and improving the Watershed segmentation process.