Limits...
Interactive Volumetry Of Liver Ablation Zones.

Egger J, Busse H, Brandmaier P, Seider D, Gawlitza M, Strocka S, Voglreiter P, Dokter M, Hofmann M, Kainz B, Hann A, Chen X, Alhonnoro T, Pollari M, Schmalstieg D, Moche M - Sci Rep (2015)

Bottom Line: For the quantitative and qualitative analysis of the algorithm's results, manual slice-by-slice segmentations produced by clinical experts have been used as the gold standard (which have also been compared among each other).The results show that the proposed tool provides lesion segmentation with sufficient accuracy much faster than manual segmentation.The visual feedback and interactivity make the proposed tool well suitable for the clinical workflow.

View Article: PubMed Central - PubMed

Affiliation: Department of Neuroscience and Biomedical Engineering, Aalto University, Rakentajanaukio 2 C, 02150 Espoo, Finland.

ABSTRACT
Percutaneous radiofrequency ablation (RFA) is a minimally invasive technique that destroys cancer cells by heat. The heat results from focusing energy in the radiofrequency spectrum through a needle. Amongst others, this can enable the treatment of patients who are not eligible for an open surgery. However, the possibility of recurrent liver cancer due to incomplete ablation of the tumor makes post-interventional monitoring via regular follow-up scans mandatory. These scans have to be carefully inspected for any conspicuousness. Within this study, the RF ablation zones from twelve post-interventional CT acquisitions have been segmented semi-automatically to support the visual inspection. An interactive, graph-based contouring approach, which prefers spherically shaped regions, has been applied. For the quantitative and qualitative analysis of the algorithm's results, manual slice-by-slice segmentations produced by clinical experts have been used as the gold standard (which have also been compared among each other). As evaluation metric for the statistical validation, the Dice Similarity Coefficient (DSC) has been calculated. The results show that the proposed tool provides lesion segmentation with sufficient accuracy much faster than manual segmentation. The visual feedback and interactivity make the proposed tool well suitable for the clinical workflow.

No MeSH data available.


Related in: MedlinePlus

This figure shall illustrate the course of action of the mincut for different Δr values.On the left side, the inter-edges for a Δr value of zero have been constructed. Thus the mincut will separate all nodes on the same “node level” to avoid costs for cutting inter-edges. Note that the location of the cut (green) depends here on other factors, like the underlying gray values. Similar to the second image of Fig. 12, the inter-edges for a Δr value have been constructed for the following three rightmost images. As you can see, the mincut has two options for cutting inter-edges (red scissors) producing the same costs: (1.) on the same “node level” (second image from the right) and (2.) cutting on different “node levels” with a node distance on one (third image from the right). However, cutting on different “node levels” with a node distance on two (or greater) will produce higher cost and therefore will automatically be avoided by the mincut algorithm as seen in the rightmost figure. The same principle applies also for larger values of Δr.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4612735&req=5

f13: This figure shall illustrate the course of action of the mincut for different Δr values.On the left side, the inter-edges for a Δr value of zero have been constructed. Thus the mincut will separate all nodes on the same “node level” to avoid costs for cutting inter-edges. Note that the location of the cut (green) depends here on other factors, like the underlying gray values. Similar to the second image of Fig. 12, the inter-edges for a Δr value have been constructed for the following three rightmost images. As you can see, the mincut has two options for cutting inter-edges (red scissors) producing the same costs: (1.) on the same “node level” (second image from the right) and (2.) cutting on different “node levels” with a node distance on one (third image from the right). However, cutting on different “node levels” with a node distance on two (or greater) will produce higher cost and therefore will automatically be avoided by the mincut algorithm as seen in the rightmost figure. The same principle applies also for larger values of Δr.

Mentions: In more detail, the whole set of edges E (of a constructed graph G) consists of edges that connect the sampled nodes to the source and the sink, and edges that establish connections between two sampled nodes. However, in a first step rays (leftmost image of Fig. 11, blue lines) are send out from the user-defined seed point through the surface points of an polyhedron (leftmost image of Fig. 11, red dots). Afterwards, the graphs nodes are sampled along these rays between the user-defined seed point and the surface points of the polyhedron (image in the middle of Fig. 11, red dots). Then, the so called intra-edges between the nodes (red) along one ray are constructed (rightmost image of Fig. 11, blue arrows). These edges ensure that all nodes below a segmented surface in the graph are included to form a closed set, or in other words, the interior of the object (the ablation zone) is separated from the exterior (the surrounding background) in the data. Next, the inter-edges of the graph are constructed. These edges establish (a) connections between nodes that have been sampled along different rays and (b) connections from the sampled nodes to the source and the sink. Thereby, the intra-edges (a) constrain the set of possible segmentations and enforce the smoothness of the segmentation result over an integer parameter Δr. The larger this parameter is, the larger is also the number of possible segmentations and a value of zero enforces the segmentation result to be a sphere. Figure 12 exemplarily shows the inter-edges (between the nodes sampled along three rays) for three different Δr values: zero (leftmost image), one (image in the middle) and two (rightmost image). Supplementary, Fig. 13 demonstrates how the different Δr values influence the segmentation outcome for two adjacent rays and their sampled nodes. For the leftmost image of Fig. 13, a Δr value of zero was chosen. Thus, only inter-edges between nodes on the same “node level” are established, which will also lead the mincut to be on the same “node level” of different rays (green); elsewise “costs” would arise by cutting the inter-edges (which will the mincut automatically avoid). Furthermore, the position (or the “node level”) of the cut (in this example between the second and third nodes when counting from the bottom), depends on other factors like the gray values in the image. However, in any case the segmentation outcome will be a sphere and the “node level” of the cut determines the size of this sphere. The next three images of Fig. 13 will illustrate what happens for a Δr value of one. As you can see, the inter-edges have been constructed between different “node levels” equivalent to the image in the middle of Fig. 12. For a Δr value of one, the mincut has definitely to cut two inter-edges (red scissors), regardless if the cut is on the same “node level” (second image from the right of Fig. 13) or if the cut is between different “node levels” (second image from the right of Fig. 13). However, for a Δr value of one, this only applies if the cut also appears within a maximum “node level” distance of one. For a cut with a “node level” distance of two (or larger) and a Δr value of one, “costs” for cutting four (or more) inter-edges would arise as shown in the rightmost image of Fig. 13 (note: the mincut algorithm will automatically avoid this cut). Accordingly, this principle applies to larger Δr values, like 2, 3, etc. Once the connections between the nodes (sampled nodes and virtual nodes) have been established, a weight is assigned to every edge. These weights are the costs that arise when the mincut algorithm cuts through the edges. The intra-edges along one ray and the inter-edges resulting from the Δr value are assigned a maximum value ∞, e.g. the maximum float value when implemented. The weights of the edges connecting the sampled nodes with the two virtual nodes source and sink depend on the gray values of the image, namely the positions of the sampled nodes. Technically speaking, the weights depend on cost-values describing the absolute value of the difference of an average ablation zone value and the gray value at . Figure 14 provides an example for nodes that have been sampled along one ray and then bound via the absolute value of the difference of an average (ablation zone) value to the source (red) and the sink (blue). Additionally, the intra-edges (1.-8.) are drafted in the Figure. As you can see the lowest cost of ninety arise when the mincut algorithm cuts through the third intra-edge (green). The following Fig. 15 emphasizes the value of the intra-edges: if removed from Fig. 14, the mincut could avoid cutting any edges (green) resulting a total cost of zero. However, the average ablation zone (gray) value has a huge influence on the segmentation result, but we can assume that the user places the seed point inside the ablation zone. Thus we can use this information to determine the average ablation zone value during the interactive segmentation. In addition, to avoid outliers, e.g. produced by the RFA needle (if the needle is still in place and therefore visible within the image), we integrate over a small area of about one cm3 around the current position of the user-defined seed point. In doing so, we handle situations, where the user places the seed point at a position of the RFA needle within the image and the corresponding gray value is much too bright/large to be used as an average ablation zone value. This strategy also enables to automatically handle different image acquisitions and makes the algorithm kind of disease invariant and scan independent (for example against possible inhomogeneities within ablation zones). But the greatest advantage is still the automatic handling of scans with and without RFA needles visible within the medical image.


Interactive Volumetry Of Liver Ablation Zones.

Egger J, Busse H, Brandmaier P, Seider D, Gawlitza M, Strocka S, Voglreiter P, Dokter M, Hofmann M, Kainz B, Hann A, Chen X, Alhonnoro T, Pollari M, Schmalstieg D, Moche M - Sci Rep (2015)

This figure shall illustrate the course of action of the mincut for different Δr values.On the left side, the inter-edges for a Δr value of zero have been constructed. Thus the mincut will separate all nodes on the same “node level” to avoid costs for cutting inter-edges. Note that the location of the cut (green) depends here on other factors, like the underlying gray values. Similar to the second image of Fig. 12, the inter-edges for a Δr value have been constructed for the following three rightmost images. As you can see, the mincut has two options for cutting inter-edges (red scissors) producing the same costs: (1.) on the same “node level” (second image from the right) and (2.) cutting on different “node levels” with a node distance on one (third image from the right). However, cutting on different “node levels” with a node distance on two (or greater) will produce higher cost and therefore will automatically be avoided by the mincut algorithm as seen in the rightmost figure. The same principle applies also for larger values of Δr.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4612735&req=5

f13: This figure shall illustrate the course of action of the mincut for different Δr values.On the left side, the inter-edges for a Δr value of zero have been constructed. Thus the mincut will separate all nodes on the same “node level” to avoid costs for cutting inter-edges. Note that the location of the cut (green) depends here on other factors, like the underlying gray values. Similar to the second image of Fig. 12, the inter-edges for a Δr value have been constructed for the following three rightmost images. As you can see, the mincut has two options for cutting inter-edges (red scissors) producing the same costs: (1.) on the same “node level” (second image from the right) and (2.) cutting on different “node levels” with a node distance on one (third image from the right). However, cutting on different “node levels” with a node distance on two (or greater) will produce higher cost and therefore will automatically be avoided by the mincut algorithm as seen in the rightmost figure. The same principle applies also for larger values of Δr.
Mentions: In more detail, the whole set of edges E (of a constructed graph G) consists of edges that connect the sampled nodes to the source and the sink, and edges that establish connections between two sampled nodes. However, in a first step rays (leftmost image of Fig. 11, blue lines) are send out from the user-defined seed point through the surface points of an polyhedron (leftmost image of Fig. 11, red dots). Afterwards, the graphs nodes are sampled along these rays between the user-defined seed point and the surface points of the polyhedron (image in the middle of Fig. 11, red dots). Then, the so called intra-edges between the nodes (red) along one ray are constructed (rightmost image of Fig. 11, blue arrows). These edges ensure that all nodes below a segmented surface in the graph are included to form a closed set, or in other words, the interior of the object (the ablation zone) is separated from the exterior (the surrounding background) in the data. Next, the inter-edges of the graph are constructed. These edges establish (a) connections between nodes that have been sampled along different rays and (b) connections from the sampled nodes to the source and the sink. Thereby, the intra-edges (a) constrain the set of possible segmentations and enforce the smoothness of the segmentation result over an integer parameter Δr. The larger this parameter is, the larger is also the number of possible segmentations and a value of zero enforces the segmentation result to be a sphere. Figure 12 exemplarily shows the inter-edges (between the nodes sampled along three rays) for three different Δr values: zero (leftmost image), one (image in the middle) and two (rightmost image). Supplementary, Fig. 13 demonstrates how the different Δr values influence the segmentation outcome for two adjacent rays and their sampled nodes. For the leftmost image of Fig. 13, a Δr value of zero was chosen. Thus, only inter-edges between nodes on the same “node level” are established, which will also lead the mincut to be on the same “node level” of different rays (green); elsewise “costs” would arise by cutting the inter-edges (which will the mincut automatically avoid). Furthermore, the position (or the “node level”) of the cut (in this example between the second and third nodes when counting from the bottom), depends on other factors like the gray values in the image. However, in any case the segmentation outcome will be a sphere and the “node level” of the cut determines the size of this sphere. The next three images of Fig. 13 will illustrate what happens for a Δr value of one. As you can see, the inter-edges have been constructed between different “node levels” equivalent to the image in the middle of Fig. 12. For a Δr value of one, the mincut has definitely to cut two inter-edges (red scissors), regardless if the cut is on the same “node level” (second image from the right of Fig. 13) or if the cut is between different “node levels” (second image from the right of Fig. 13). However, for a Δr value of one, this only applies if the cut also appears within a maximum “node level” distance of one. For a cut with a “node level” distance of two (or larger) and a Δr value of one, “costs” for cutting four (or more) inter-edges would arise as shown in the rightmost image of Fig. 13 (note: the mincut algorithm will automatically avoid this cut). Accordingly, this principle applies to larger Δr values, like 2, 3, etc. Once the connections between the nodes (sampled nodes and virtual nodes) have been established, a weight is assigned to every edge. These weights are the costs that arise when the mincut algorithm cuts through the edges. The intra-edges along one ray and the inter-edges resulting from the Δr value are assigned a maximum value ∞, e.g. the maximum float value when implemented. The weights of the edges connecting the sampled nodes with the two virtual nodes source and sink depend on the gray values of the image, namely the positions of the sampled nodes. Technically speaking, the weights depend on cost-values describing the absolute value of the difference of an average ablation zone value and the gray value at . Figure 14 provides an example for nodes that have been sampled along one ray and then bound via the absolute value of the difference of an average (ablation zone) value to the source (red) and the sink (blue). Additionally, the intra-edges (1.-8.) are drafted in the Figure. As you can see the lowest cost of ninety arise when the mincut algorithm cuts through the third intra-edge (green). The following Fig. 15 emphasizes the value of the intra-edges: if removed from Fig. 14, the mincut could avoid cutting any edges (green) resulting a total cost of zero. However, the average ablation zone (gray) value has a huge influence on the segmentation result, but we can assume that the user places the seed point inside the ablation zone. Thus we can use this information to determine the average ablation zone value during the interactive segmentation. In addition, to avoid outliers, e.g. produced by the RFA needle (if the needle is still in place and therefore visible within the image), we integrate over a small area of about one cm3 around the current position of the user-defined seed point. In doing so, we handle situations, where the user places the seed point at a position of the RFA needle within the image and the corresponding gray value is much too bright/large to be used as an average ablation zone value. This strategy also enables to automatically handle different image acquisitions and makes the algorithm kind of disease invariant and scan independent (for example against possible inhomogeneities within ablation zones). But the greatest advantage is still the automatic handling of scans with and without RFA needles visible within the medical image.

Bottom Line: For the quantitative and qualitative analysis of the algorithm's results, manual slice-by-slice segmentations produced by clinical experts have been used as the gold standard (which have also been compared among each other).The results show that the proposed tool provides lesion segmentation with sufficient accuracy much faster than manual segmentation.The visual feedback and interactivity make the proposed tool well suitable for the clinical workflow.

View Article: PubMed Central - PubMed

Affiliation: Department of Neuroscience and Biomedical Engineering, Aalto University, Rakentajanaukio 2 C, 02150 Espoo, Finland.

ABSTRACT
Percutaneous radiofrequency ablation (RFA) is a minimally invasive technique that destroys cancer cells by heat. The heat results from focusing energy in the radiofrequency spectrum through a needle. Amongst others, this can enable the treatment of patients who are not eligible for an open surgery. However, the possibility of recurrent liver cancer due to incomplete ablation of the tumor makes post-interventional monitoring via regular follow-up scans mandatory. These scans have to be carefully inspected for any conspicuousness. Within this study, the RF ablation zones from twelve post-interventional CT acquisitions have been segmented semi-automatically to support the visual inspection. An interactive, graph-based contouring approach, which prefers spherically shaped regions, has been applied. For the quantitative and qualitative analysis of the algorithm's results, manual slice-by-slice segmentations produced by clinical experts have been used as the gold standard (which have also been compared among each other). As evaluation metric for the statistical validation, the Dice Similarity Coefficient (DSC) has been calculated. The results show that the proposed tool provides lesion segmentation with sufficient accuracy much faster than manual segmentation. The visual feedback and interactivity make the proposed tool well suitable for the clinical workflow.

No MeSH data available.


Related in: MedlinePlus