Limits...
Nodule detection in a lung region that's segmented with using genetic cellular neural networks and 3D template matching with fuzzy rule based thresholding.

Ozekes S, Osman O, Ucan ON - Korean J Radiol (2008 Jan-Feb)

Bottom Line: First, to reduce the number of region of interests (ROIs) and the computation time, the lung regions of the CTs were segmented using Genetic Cellular Neural Networks (G-CNN).Finally, fuzzy rule based thresholding was applied and the ROI's were found.Our results indicate that the detection performance of our algorithm is satisfactory, and this may well improve the performance of computer-aided detection of lung nodules.

View Article: PubMed Central - PubMed

Affiliation: Istanbul Commerce University, Ragip Gumuspala Cad. No: 84 34378 Eminonu, Istanbul, Turkey. serhat@iticu.edu.tr

ABSTRACT

Objective: The purpose of this study was to develop a new method for automated lung nodule detection in serial section CT images with using the characteristics of the 3D appearance of the nodules that distinguish themselves from the vessels.

Materials and methods: Lung nodules were detected in four steps. First, to reduce the number of region of interests (ROIs) and the computation time, the lung regions of the CTs were segmented using Genetic Cellular Neural Networks (G-CNN). Then, for each lung region, ROIs were specified with using the 8 directional search; +1 or -1 values were assigned to each voxel. The 3D ROI image was obtained by combining all the 2-Dimensional (2D) ROI images. A 3D template was created to find the nodule-like structures on the 3D ROI image. Convolution of the 3D ROI image with the proposed template strengthens the shapes that are similar to those of the template and it weakens the other ones. Finally, fuzzy rule based thresholding was applied and the ROI's were found. To test the system's efficiency, we used 16 cases with a total of 425 slices, which were taken from the Lung Image Database Consortium (LIDC) dataset.

Results: The computer aided diagnosis (CAD) system achieved 100% sensitivity with 13.375 FPs per case when the nodule thickness was greater than or equal to 5.625 mm.

Conclusion: Our results indicate that the detection performance of our algorithm is satisfactory, and this may well improve the performance of computer-aided detection of lung nodules.

Show MeSH

Related in: MedlinePlus

A. A voxel which doesn't have a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", so it is not a part of the ROI.B. A voxel which doesn't have a number of adjacent neighbor voxels less than or equal to the value of "maximum distance threshold", so it is not a part of the ROI.C. A voxel which has a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", and less than or equal to the value of "maximum distance threshold", so it is a part of the ROI.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2627180&req=5

Figure 6: A. A voxel which doesn't have a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", so it is not a part of the ROI.B. A voxel which doesn't have a number of adjacent neighbor voxels less than or equal to the value of "maximum distance threshold", so it is not a part of the ROI.C. A voxel which has a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", and less than or equal to the value of "maximum distance threshold", so it is a part of the ROI.

Mentions: Voxels, which form the candidate lung nodule region, must be members of a set of adjacent neighbor voxels with densities between the "minimum density threshold" and the "maximum density threshold" values. Thus, in the first step of the ROI specification method, thresholding was performed to find the voxels with densities between the "minimum density threshold" and the "maximum density threshold" values. Figures 2D, 3D and 4D show the voxels with suitable density values. It has been observed that the diameters of lung nodules are between the upper and lower boundaries. So, to understand whether a voxel is in the center region of the shape, first, the diameter of the shape (assuming the voxel in question is the center) should be considered. At this stage, we introduce two thresholds that form the boundaries. One is the "minimum distance threshold" representing the lower boundary and the other is the "maximum distance threshold" representing the upper boundary. If a voxel has adjacent neighbors that are less than the "minimum distance threshold" or more than the "maximum distance threshold" in 8 directions, it could be concluded that this voxel couldn't be a part of the candidate lung nodule. Otherwise, it could be a part of the candidate lung nodule. Examples of determining if the voxels are a part of the ROI can be seen in Figure 6. Assume that the grey voxels in Figures 6A-C have suitable intensities. As seen in Figure 6A, if a grey voxel doesn't have a number of adjacent neighbor grey voxels that are greater than or equal to the value of the "minimum distance threshold", or as seen in Figure 6B, if a grey voxel doesn't have a number of adjacent neighbor grey voxels that are less than or equal to the value of the "maximum distance threshold" in all directions, it could be considered that the voxel under investigation is not a part of the ROI. Otherwise, as seen in Figure 6C, it could be concluded that the voxel is a part of the ROI. The values of the minimum and maximum distance thresholds deal with the resolution of the CT image. These thresholds are used to avoid very big or very small structures such as parts of the chest bones or heart and vertical vessels. In Figures 2E, 3E and 4E, the lung regions are shown with grey color and he detected ROIs are shown with black.


Nodule detection in a lung region that's segmented with using genetic cellular neural networks and 3D template matching with fuzzy rule based thresholding.

Ozekes S, Osman O, Ucan ON - Korean J Radiol (2008 Jan-Feb)

A. A voxel which doesn't have a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", so it is not a part of the ROI.B. A voxel which doesn't have a number of adjacent neighbor voxels less than or equal to the value of "maximum distance threshold", so it is not a part of the ROI.C. A voxel which has a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", and less than or equal to the value of "maximum distance threshold", so it is a part of the ROI.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2627180&req=5

Figure 6: A. A voxel which doesn't have a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", so it is not a part of the ROI.B. A voxel which doesn't have a number of adjacent neighbor voxels less than or equal to the value of "maximum distance threshold", so it is not a part of the ROI.C. A voxel which has a number of adjacent neighbor voxels greater than or equal to the value of "minimum distance threshold", and less than or equal to the value of "maximum distance threshold", so it is a part of the ROI.
Mentions: Voxels, which form the candidate lung nodule region, must be members of a set of adjacent neighbor voxels with densities between the "minimum density threshold" and the "maximum density threshold" values. Thus, in the first step of the ROI specification method, thresholding was performed to find the voxels with densities between the "minimum density threshold" and the "maximum density threshold" values. Figures 2D, 3D and 4D show the voxels with suitable density values. It has been observed that the diameters of lung nodules are between the upper and lower boundaries. So, to understand whether a voxel is in the center region of the shape, first, the diameter of the shape (assuming the voxel in question is the center) should be considered. At this stage, we introduce two thresholds that form the boundaries. One is the "minimum distance threshold" representing the lower boundary and the other is the "maximum distance threshold" representing the upper boundary. If a voxel has adjacent neighbors that are less than the "minimum distance threshold" or more than the "maximum distance threshold" in 8 directions, it could be concluded that this voxel couldn't be a part of the candidate lung nodule. Otherwise, it could be a part of the candidate lung nodule. Examples of determining if the voxels are a part of the ROI can be seen in Figure 6. Assume that the grey voxels in Figures 6A-C have suitable intensities. As seen in Figure 6A, if a grey voxel doesn't have a number of adjacent neighbor grey voxels that are greater than or equal to the value of the "minimum distance threshold", or as seen in Figure 6B, if a grey voxel doesn't have a number of adjacent neighbor grey voxels that are less than or equal to the value of the "maximum distance threshold" in all directions, it could be considered that the voxel under investigation is not a part of the ROI. Otherwise, as seen in Figure 6C, it could be concluded that the voxel is a part of the ROI. The values of the minimum and maximum distance thresholds deal with the resolution of the CT image. These thresholds are used to avoid very big or very small structures such as parts of the chest bones or heart and vertical vessels. In Figures 2E, 3E and 4E, the lung regions are shown with grey color and he detected ROIs are shown with black.

Bottom Line: First, to reduce the number of region of interests (ROIs) and the computation time, the lung regions of the CTs were segmented using Genetic Cellular Neural Networks (G-CNN).Finally, fuzzy rule based thresholding was applied and the ROI's were found.Our results indicate that the detection performance of our algorithm is satisfactory, and this may well improve the performance of computer-aided detection of lung nodules.

View Article: PubMed Central - PubMed

Affiliation: Istanbul Commerce University, Ragip Gumuspala Cad. No: 84 34378 Eminonu, Istanbul, Turkey. serhat@iticu.edu.tr

ABSTRACT

Objective: The purpose of this study was to develop a new method for automated lung nodule detection in serial section CT images with using the characteristics of the 3D appearance of the nodules that distinguish themselves from the vessels.

Materials and methods: Lung nodules were detected in four steps. First, to reduce the number of region of interests (ROIs) and the computation time, the lung regions of the CTs were segmented using Genetic Cellular Neural Networks (G-CNN). Then, for each lung region, ROIs were specified with using the 8 directional search; +1 or -1 values were assigned to each voxel. The 3D ROI image was obtained by combining all the 2-Dimensional (2D) ROI images. A 3D template was created to find the nodule-like structures on the 3D ROI image. Convolution of the 3D ROI image with the proposed template strengthens the shapes that are similar to those of the template and it weakens the other ones. Finally, fuzzy rule based thresholding was applied and the ROI's were found. To test the system's efficiency, we used 16 cases with a total of 425 slices, which were taken from the Lung Image Database Consortium (LIDC) dataset.

Results: The computer aided diagnosis (CAD) system achieved 100% sensitivity with 13.375 FPs per case when the nodule thickness was greater than or equal to 5.625 mm.

Conclusion: Our results indicate that the detection performance of our algorithm is satisfactory, and this may well improve the performance of computer-aided detection of lung nodules.

Show MeSH
Related in: MedlinePlus