Efficient organ segmentation is the precondition of various quantitative analysis. Segmenting the pancreas from abdominal CT images is a challenging task because of its high anatomical variability in shape, size and location. What's more, the pancreas only occupies a small portion in abdomen, and the organ border is very fuzzy. All these factors make the segmentation methods of other organs less suitable for pancreas. In this work, we propose a Markov Chain Monte Carlo (MCMC) guided convolutional neural network (CNN) approach, in order to handle such difficulties in morphological and photometric variabilities. Specifically, the proposed method mainly consists of three steps: First, registration is carried out to mitigate the body weight and location variability. Then, an MCMC scheme is designed to guide the adaptive selection of 3D patches, which are fed to the CNN for training. At the same time, the pancreas distribution is also learned for subsequent segmentation. Eventually, the same MCMC process guides the segmentation process with patch-wise predictions fused using a Bayesian voting scheme. This method is evaluated on the NIH pancreatic dataset including 82 abdominal contrast-enhanced CT volumes. We have achieved a competitive result of 78.13% Dice Similarity Coefficient value and 82.65% Recall value in testing data.

Effificient organ segmentation is the precondition of various quantitative analysis. Segmenting the pancreas from abdominal CT images is a challenging task because of its high anatomical variability in shape, size and location. What's more, the pancreas only occupies a small portion in abdomen, and the organ border is very fuzzy. All these factors make the segmentation methods of other organs less suitable for the pancreas segmentation. In this report, we propose a Markov Chain Monte Carlo (MCMC) sampling guided convolutionation neural network (CNN) approach, in order to handel such difficulties in morphologic and photometric variabilities. Specifically, the proposed method mainly contains three steps: First, registration is carried out to mitigate the body weight and location variability. Then, an MCMC sampling is employed to guide the sampling of 3D patches, which are fed to the CNN for training. At the same time, the pancreas distribution is also learned for the subsequent segmetnation. Third, sampled from the learned distribution, an MCMC process guides the segmentation process. Lastly, the patches based segmentation is fused using a Bayesian voting scheme. This method is evaluated on the NIH pancreatic datasets which contains 82 abdominal contrast-enhanced CT volumes. Finally, we achieved a competitive result of 78.13% Dice Similarity Coeffificient value and 82.65% Recall value in testing data.