Multi-phase simultaneous segmentation of tumor in lung 4D-CT data with context information.

Lung 4D computed tomography (4D-CT) plays an important role in high-precision radiotherapy because it characterizes respiratory motion, which is crucial for accurate target definition. However, the manual segmentation of a lung tumor is a heavy workload for doctors because of the large number of lung 4D-CT data slices. Meanwhile, tumor segmentation is still a notoriously challenging problem in computer-aided diagnosis. In this paper, we propose a new method based on an improved graph cut algorithm with context information constraint to find a convenient and robust approach of lung 4D-CT tumor segmentation. We combine all phases of the lung 4D-CT into a global graph, and construct a global energy function accordingly. The sub-graph is first constructed for each phase. A context cost term is enforced to achieve segmentation results in every phase by adding a context constraint between neighboring phases. A global energy function is finally constructed by combining all cost terms. The optimization is achieved by solving a max-flow/min-cut problem, which leads to simultaneous and robust segmentation of the tumor in all the lung 4D-CT phases. The effectiveness of our approach is validated through experiments on 10 different lung 4D-CT cases. The comparison with the graph cut without context constraint, the level set method and the graph cut with star shape prior demonstrates that the proposed method obtains more accurate and robust segmentation results.

[Super-resolution reconstruction for 4-dimensional computed tomography of the lung using graph cuts].

Four-dimensional computer tomography (4D-CT) has a great value in lung cancer radiotherapy for its capability in providing lung information with respiratory motion. We employed a global graph cuts super-resolution (SR) reconstruction method to reconstruct high-resolution lung 4D-CT images. First, the high-resolution images reconstruction energy function was built based on a Maximum a posteriori Markov Random Field (MAP-MRF) formulation. The energy function was then transformed to a graph formulation, which was solved using graph cut algorithm. All the evaluation results showed that this approach outperformed the line interpolation and projection onto convex sets (POCS) approach with an improved structural clarity.

Robust low-dose dynamic cerebral perfusion CT image restoration via coupled dictionary learning scheme.

Dynamic cerebral perfusion x-ray computed tomography (PCT) imaging has been advocated to quantitatively and qualitatively assess hemodynamic parameters in the diagnosis of acute stroke or chronic cerebrovascular diseases. However, the associated radiation dose is a significant concern to patients due to its dynamic scan protocol. To address this issue, in this paper we propose an image restoration method by utilizing coupled dictionary learning (CDL) scheme to yield clinically acceptable PCT images with low-dose data acquisition. Specifically, in the present CDL scheme, the 2D background information from the average of the baseline time frames of low-dose unenhanced CT images and the 3D enhancement information from normal-dose sequential cerebral PCT images are exploited to train the dictionary atoms respectively. After getting the two trained dictionaries, we couple them to represent the desired PCT images as spatio-temporal prior in objective function construction. Finally, the low-dose dynamic cerebral PCT images are restored by using a general DL image processing. To get a robust solution, the objective function is solved by using a modified dictionary learning based image restoration algorithm. The experimental results on clinical data show that the present method can yield more accurate kinetic enhanced details and diagnostic hemodynamic parameter maps than the state-of-the-art methods.

A genetic algorithm for detecting target geometric figures.

A genetic algorithm is proposed to detect target geometric figures in an given image. Float-point encoding was adopted to process the parameters of a geometric figures to be detected. On the basis of classical Hough transform, a fitness function was obtained for each individual task, and the individuals with the highest fitness function were identified and copied into the cohort of the next generation. For the rest of the individuals, operation with multi-point crossover or uniform mutation was performed to form new individuals in the next generation. When the termination conditions for this genetic algorithm were met, the best-fitted individual was decoded and output as the parameters of the detected geometric figures. This algorithm can eliminate noise interference with good convergence and accurate results, and may save time and storage space during relatively easily programmed computation in comparison with classical Hough transform.

A simple analysis model simulating the correlation of cervical traction force with the pressure in the cervical nucleus pulposus.

OBJECTIVE: To design a simple model for analyzing the correlation between the changes in cervical traction force and that of the pressure in the cervical nucleus pulposus. METHODS: Multiple non-linear mathematical models were established by means of regression analysis to calculate the correlation coefficient and variance, and the capacity of each model to predict the changes on the test points was evaluated. RESULTS: Logarithm model was apparently not adequate for the purpose and quadratic model and power model showed poor capacity for predicting the changes of the experimental data. Exponential model and cubic model, in contrast, well illustrated the trend of the changes in the data, therefore best met the demands. CONCLUSIONS: Exponential model best describes the trends of the changes of the experimental data and can procure continuous values of the pressure reduction in the cervical nucleus pulposus in association with varied cervical traction forces, thus offering useful information for relevant theoretical analysis.

A median filter algorithm importing ISODATA dynamic clustering for medical imaging.

To improve conventional median filter algorithm employed in medical imaging, we proposed a new median filter algorithm importing ISODATA dynamic clustering for pattern recognition. The result of clustering was set as the parameters to decide if median filter was necessary and when it was, how the process was to be carried out. As shown in our test, this algorithm was capable of eliminating serious impulse noises and retain thorough image details, therefore enhanced signal to noise ratio and quality of the images in contrast with the conventional median filter algorithm.