| Medical image segmentation as an earlier application field in image segmentation is the key technology of medical image analysis,and is also a key point and difficulty in clinical application.From the medical image,we can extract the regions of interest through the segmentation algorithm and display them separately,which provides more intuitive information about the lesion or normal tissue structure.Although existing traditional models have achieved good results,they still have some problems.For example,the multiplicative intrinsic component optimization model cannot segment noise images well and is not robust to noise.Therefore,based on the disadvantages of this model and inspired by it,this paper proposes two different methods to improve the multiplicative intrinsic component optimization model.For medical MR images,this paper proposes an accurate and robust active contour model based on the two-phase level set,and successfully extended to a four-phase model for human brain MR images.We define a new energy functional by combining the data term and the length term,where the data term is defined by transforming the energy functional of the Multiplicative intrinsic component optimization model into the level set framework after adding an edge detector function.Besides,the split Bregman method is applied to efficiently minimize the energy functional.We use our model to segment lots of images,verifying that our model can correct and segment images with bias field well.Experimental results also validate that our model is robust to initials and noises.At the same time,we compared the new model with the multiplicative intrinsic component optimization model through experimental results and numerical results,and the results show that our model is superior to the multiplicative intrinsic component optimization model in both segmentation accuracy and correction effect.For CBCT images of teeth,especially those surrounded by gums,an accurate antinoise model with prior condition information is proposed in this paper.Through the analysis of the tooth image,we default a single tooth to an oval shape,and obtain the prior condition information through a series of analysis.The energy function of the model is mainly composed of the target image data term,the final contour curve length term and the prior condition information term.Then the gradient descent method is used to minimize the energy function.We used this model to segment a large number of toothimages.The experimental results show that this model can accurately segment tooth images with uneven intensity and similar target boundary to the background.And through experimental comparison,we find that the newly introduced prior condition information plays a crucial role in the segmentation ability of the new model. |
