Segmentation of Image Sequences of Neuron Stem Cells Based on Level-set Algorithm Combined with Local Gray Threshold
|School||Harbin Engineering University|
|Course||Signal and Information Processing|
|Keywords||Sequence of image segmentation of neural stem cells Level set Local gray threshold Adhesions, clusters cells Separate|
Image segmentation technology plays an extremely important role in medical image processing . Uneven due to the body's tissues and organs , organ peristalsis caused by medical image generally have noise, diseased tissue edge blur , so how to avoid the impact of noise and image clarity , to get accurate target area contour image segmentation problem . The neural stem cells (neural stem cell, NSC) is a kind of split potential and self-renewal capacity of neuroblastoma . Variety of ways through artificial injection of neural stem cell transplantation, stroke , spinal cord injury and senile dementia a variety of neurological disorders can be treated . Neurons in the process of generating and motion characteristics observed neuron stem cells , can greatly improve the accuracy of diagnosis and cure , and cell image segmentation process as the premise of such cell research is particularly important . Optical microscope imaging of neurons stem cells image sequences for target and background of weak contrast and cell adhesion , clusters and other issues , this paper presents a new segmentation algorithm . The algorithm is based on the need to initialize the level set algorithm , introduced to accelerate the convergence of the curvature ; reduce the complexity of the algorithm proposed measure the norm energy as the termination condition of the level set evolution ; Finally, the local gray threshold method further split the adhesion of cells . The algorithm is applied to two groups of cell image sequence of 120 image segmentation , which not only solved the the focus drift time series image imaging division problem , and be able to accurately separate the adhesions, clusters cells , and retained cells morphological characteristics and location information. The statistical results showed that the split share of the entire sequence of the percentage of successful frame waterline segmentation, the traditional level set segmentation algorithm increased by 30 % -40 % . The completion of the application of the algorithm such sequence of image segmentation has laid a good foundation for cell recognition and cell tracking .