Academic
Publications
Fluid Vector Flow and Applications in Brain Tumor Segmentation

Fluid Vector Flow and Applications in Brain Tumor Segmentation,10.1109/TBME.2009.2012423,IEEE Transactions on Biomedical Engineering,Tao Wang,Irene Ch

Fluid Vector Flow and Applications in Brain Tumor Segmentation   (Citations: 8)
BibTex | RIS | RefWorks Download
In this paper, we propose a new approach that we call the ldquofluid vector flowrdquo (FVF) active contour model to address problems of insufficient capture range and poor convergence for concavities. With the ability to capture a large range and extract concave shapes, FVF demonstrates improvements over techniques like gradient vector flow, boundary vector flow, and magnetostatic active contour on three sets of experiments: synthetic images, pediatric head MRI images, and brain tumor MRI images from the Internet brain segmentation repository.
Journal: IEEE Transactions on Biomedical Engineering - IEEE TRANS BIOMED ENG , vol. 56, no. 3, pp. 781-789, 2009
Cumulative Annual
View Publication
The following links allow you to view full publications. These links are maintained by other sources not affiliated with Microsoft Academic Search.
    • ...Active contour (AC) models (or snakes) [11]‐[13] are commonly used as segmentation techniques for medical images, though prior knowledge of the region of interest (ROI) is often needed in order to guide the execution and successful convergence of these algorithms...
    • ...However, as discussed in earlier research [13], traditional parametricACmodelshavelimitedcapturerange.Itcannotconverge accurately unless an initial contour is specified close to the...
    • ...Although AC models based on energy minimization are effective segmentation methods, the traditional AC models have severaldrawbacks:theyaresensitivetotheinitialcontourplacement, can have a capture range too small to detect cavities, may need to apply different parameter values to guide convergence, and do not perform well on weak edges [13]...
    • ...This AC modelperformswellforcontrastchangesandweakedges.Since multiple cavities can exist [see Fig. 2(b)], we cannot apply fluid vector flow [13], which is designed to segment only a single region...
    • ...Like in [13], we define the Tanimoto measure as TM = � R...

    Rui Shenet al. A Hybrid Knowledge-Guided Detection Technique for Screening of Infecti...

    • ...Although these methods have an ability to choose the edge’s type in object segmentation, their external force is in a form of vector filed that is derived from the image gradient, which is known to have some weaknesses such as the sensitivity to noise, poor convergence into concave parts, and the inevitable problems with saddle and stationary points [9]...

    Sopon Phumeechanyaet al. Edge type-selectable active contour using local regional information o...

    • ...Recent development of active contours [5-6] have been made by adding certain dynamic force term into the external force model to solve this problem, such as magneto-static active contour (MAC) and FVF...
    • ...Among edge-based snakes, FVF [5] generates the external force dynamically and has the largest capture range...
    • ...It outperforms other methods in medical image segmentation (see comparison results presented in [5]), however, edge leakage and control point selection are major limitations of this method...
    • ...Various efforts have been made to improve the active contours’ performance by designing different external forces [2-6]...
    • ...In FVF, the external force design is divided into two steps: vector flow initialization and FVF computation [5]...
    • ...Although the original FVF has the largest capture range and the ability to handle acute concave shapes compared to the previous edge-based snakes [5], edge leakage and inefficient control point selection problems still exist...
    • ...If the boundary detected by the boundary tracing method mentioned in [5] encounters significant edge gaps, the FVF will fail to achieve accurate result...

    Tao Xuet al. An Improved Fluid Vector Flow for Cavity Segmentation in Chest Radiogr...

    • ...GVF and balloons [5] and the topic has been previously addressed by other researchers [6]-[10], all these improvements involve an auxiliary or different mathematical machinery than that specific to the GVF, leading to solutions located outside of the original conceptual framework of the GVF...

    Tiberiu Radulescuet al. A hidden property of the gradient vector flow diffusion process

    • ...A lot of energy was spent to derive different other models able to address such cases [6]-[10] or to overcome this intrinsic limitation of the VFC by quite a sophisticated method for the snake initialization near to boundaries [11]...

    Tiberiu Radulescuet al. Extended vector field convolution snake for highly non-convex shapes s...

Sort by: