RECONSTRUCTION AND SEPARATION OF 3D NEURON USING RANDOM FOREST ALGORITHM

Digital reconstruction (or) tracing 3-dimensional neuron structure with optical microscopy images is a primary approach for characterizing the 3-dimensional neuron, which is very important for understanding the brain functions. It is more difficult when the images are contaminated by the noise and it is very challenging tasks when the neuron images having the discontinued segments of neuron pattern. While the existing algorithm works only for the single clean neuron image and does not classify the neuron and the surrounding nerve fibres. Here, the proposed method of 3-dimensional neuron segmentation using the spatial filter, Patch extraction and classification using the random forest algorithm. This segmentation process helps to reduce and remove the noise in neuron image, so the 3-dimensional neuron tracing performance will be improved. Then the neuron and nerve fibre are separated by using the random forest algorithm. The experimental results showed that the proposed method has accurate results when compared with the other reconstruction algorithm.