Segmentation of X-ray micro-computed tomography using Neural Networks trained with Statistical Information: Application to biomedical images

In the present work we describe ongoing research on the application of Artificial Neural Networks (ANNs) trained with Statistical Information in order to segment a slice of a Rhodnius Prolixus insect (vector of the Chagas's disease) μCT scan. The images were acquired at the Synchrotron Radiation for MEdical Physics (SYRMEP) beam line at the Elettra Laboratory (Trieste, Italy). Two specialized ANNs were trained with statistical information for the segmentation task. The first ANN segmented the image of interest in two regions (one of them with white pixels and the other with non-white pixels), considering the enhancement of intensity due to phase contrast effect and despite the complexity of the image. The second ANN was able to recognize, amongst the white pixels, the ones related to the insect region. Preliminary results demonstrate the viability of the method in the segmentation of X-ray μCT.