Connolly, Laura; Jamzad, Amoon; Nikniazi, Arash; Poushimin, Rana; Lasso, Andras; Sunderland, Kyle R.; Ungi, Tamas; Nunzi, Jean Michel; Rudan, John; Fichtinger, Gabor; Mousavi, Parvin
An open-source testbed for developing image-guided robotic tumor-bed inspection Conference
Imaging Network of Ontario (ImNO) Symposium, 2022.
@conference{connolly2022b,
title = {An open-source testbed for developing image-guided robotic tumor-bed inspection},
author = {Laura Connolly and Amoon Jamzad and Arash Nikniazi and Rana Poushimin and Andras Lasso and Kyle R. Sunderland and Tamas Ungi and Jean Michel Nunzi and John Rudan and Gabor Fichtinger and Parvin Mousavi},
url = {https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/01/Connolly2022b.pdf},
year = {2022},
date = {2022-01-01},
urldate = {2022-01-01},
booktitle = {Imaging Network of Ontario (ImNO) Symposium},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Connolly, Laura; Degeut, Anton; Sunderland, Kyle R.; Lasso, Andras; Ungi, Tamas; Rudan, John; Taylor, Russell H.; Mousavi, Parvin; Fichtinger, Gabor
An open-source platform for cooperative semi-autonomous robotic surgery Conference
IEEE International Conference on Autonomous Systems, IEEE IEEE, Montreal, Quebec, 2021.
@conference{Connolly2021,
title = {An open-source platform for cooperative semi-autonomous robotic surgery},
author = {Laura Connolly and Anton Degeut and Kyle R. Sunderland and Andras Lasso and Tamas Ungi and John Rudan and Russell H. Taylor and Parvin Mousavi and Gabor Fichtinger},
doi = {https://doi.org/10.1109/ICAS49788.2021.9551149},
year = {2021},
date = {2021-10-01},
urldate = {2021-10-01},
booktitle = {IEEE International Conference on Autonomous Systems},
publisher = {IEEE},
address = {Montreal, Quebec},
organization = {IEEE},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Connolly, Laura; Sunderland, Kyle R.; Lasso, Andras; Degeut, Anton; Ungi, Tamas; Rudan, John; Taylor, Russell H.; Mousavi, Parvin; Fichtinger, Gabor
A platform for robot-assisted Intraoperative imaging in breast conserving surgery Conference
Imaging Network of Ontario Symposium, Imaging Network of Ontario Symposium, Online, 2021.
@conference{Connolly2021b,
title = {A platform for robot-assisted Intraoperative imaging in breast conserving surgery},
author = {Laura Connolly and Kyle R. Sunderland and Andras Lasso and Anton Degeut and Tamas Ungi and John Rudan and Russell H. Taylor and Parvin Mousavi and Gabor Fichtinger},
url = {https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/Connolly2021a_1.pdf},
year = {2021},
date = {2021-01-01},
urldate = {2021-01-01},
booktitle = {Imaging Network of Ontario Symposium},
publisher = {Imaging Network of Ontario Symposium},
address = {Online},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Ungi, Tamas; Greer, Hastings; Sunderland, Kyle R.; Wu, Victoria; Baum, Zachary M C; Schlenger, Christopher; Oetgen, Matthew; Cleary, Kevin; Aylward, Stephen; Fichtinger, Gabor
Automatic spine ultrasound segmentation for scoliosis visualization and measurement Journal Article
In: IEEE Transactions on Biomedical Engineering, vol. 67, no. 11, pp. 3234 - 3241, 2020.
@article{Ungi2020,
title = {Automatic spine ultrasound segmentation for scoliosis visualization and measurement},
author = {Tamas Ungi and Hastings Greer and Kyle R. Sunderland and Victoria Wu and Zachary M C Baum and Christopher Schlenger and Matthew Oetgen and Kevin Cleary and Stephen Aylward and Gabor Fichtinger},
url = {https://ieeexplore.ieee.org/document/9034149
https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/Ungi2020.pdf},
doi = {10.1109/TBME.2020.2980540},
year = {2020},
date = {2020-03-01},
urldate = {2020-03-01},
journal = {IEEE Transactions on Biomedical Engineering},
volume = {67},
number = {11},
pages = {3234 - 3241},
abstract = {<p>\emph{Objective:} Integrate tracked ultrasound and AI methods to provide a safer and more accessible alternative to X-ray for scoliosis measurement. We propose automatic ultrasound segmentation for 3-dimensional spine visualization and scoliosis measurement to address difficulties in using ultrasound for spine imaging. \emph{Methods:} We trained a convolutional neural network for spine segmentation on ultrasound scans using data from eight healthy adult volunteers. We tested the trained network on eight pediatric patients. We evaluated image segmentation and 3-dimensional volume reconstruction for scoliosis measurement. \emph{Results:} As expected, fuzzy segmentation metrics reduced when trained networks were translated from healthy volunteers to patients. Recall decreased from 0.72 to 0.64 (8.2% decrease), and precision from 0.31 to 0.27 (3.7% decrease). However, after finding optimal thresholds for prediction maps, binary segmentation metrics performed better on patient data. Recall decreased from 0.98 to 0.97 (1.6% decrease), and precision from 0.10 to 0.06 (4.5% decrease). Segmentation prediction maps were reconstructed to 3-dimensional volumes and scoliosis was measured in all patients. Measurement in these reconstructions took less than 1 minute and had a maximum error of 2.2° compared to X-ray. \emph{Conclusion:} automatic spine segmentation makes scoliosis measurement both efficient and accurate in tracked ultrasound scans. \emph{Significance:} Automatic segmentation may overcome the limitations of tracked ultrasound that so far prevented its use as an alternative of X-ray in scoliosis measurement.</p>},
keywords = {},
pubstate = {published},
tppubtype = {article}
}
<p>Objective: Integrate tracked ultrasound and AI methods to provide a safer and more accessible alternative to X-ray for scoliosis measurement. We propose automatic ultrasound segmentation for 3-dimensional spine visualization and scoliosis measurement to address difficulties in using ultrasound for spine imaging. Methods: We trained a convolutional neural network for spine segmentation on ultrasound scans using data from eight healthy adult volunteers. We tested the trained network on eight pediatric patients. We evaluated image segmentation and 3-dimensional volume reconstruction for scoliosis measurement. Results: As expected, fuzzy segmentation metrics reduced when trained networks were translated from healthy volunteers to patients. Recall decreased from 0.72 to 0.64 (8.2% decrease), and precision from 0.31 to 0.27 (3.7% decrease). However, after finding optimal thresholds for prediction maps, binary segmentation metrics performed better on patient data. Recall decreased from 0.98 to 0.97 (1.6% decrease), and precision from 0.10 to 0.06 (4.5% decrease). Segmentation prediction maps were reconstructed to 3-dimensional volumes and scoliosis was measured in all patients. Measurement in these reconstructions took less than 1 minute and had a maximum error of 2.2° compared to X-ray. Conclusion: automatic spine segmentation makes scoliosis measurement both efficient and accurate in tracked ultrasound scans. Significance: Automatic segmentation may overcome the limitations of tracked ultrasound that so far prevented its use as an alternative of X-ray in scoliosis measurement.</p>
Wu, Catherine O.; Sunderland, Kyle R.; Filippov, Mihail; Sainsbury, Ben; Fichtinger, Gabor; Ungi, Tamas
Workflow for creation and evaluation of virtual nephrolithotomy training models Conference
SPIE Medical Imaging Conference 2020, vol. 11315, 2020.
@conference{CWu2020,
title = {Workflow for creation and evaluation of virtual nephrolithotomy training models},
author = {Catherine O. Wu and Kyle R. Sunderland and Mihail Filippov and Ben Sainsbury and Gabor Fichtinger and Tamas Ungi},
url = {https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/CWu2020a-manuscript.pdf},
doi = {10.1117/12.2549354},
year = {2020},
date = {2020-03-01},
urldate = {2020-03-01},
booktitle = {SPIE Medical Imaging Conference 2020},
volume = {11315},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Wu, Victoria; Ungi, Tamas; Sunderland, Kyle R.; Pigeau, Grace; Schonewille, Abigael; Fichtinger, Gabor
SPIE Medical Imaging, 2020.
@conference{Wu2020a,
title = {Automatic segmentation of spinal ultrasound landmarks with U-net using multiple consecutive images for input},
author = {Victoria Wu and Tamas Ungi and Kyle R. Sunderland and Grace Pigeau and Abigael Schonewille and Gabor Fichtinger},
url = {https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/CWu2020a-manuscript.pdf},
doi = {10.1117/12.2549584},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {SPIE Medical Imaging},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Laframboise, Jacob; Ungi, Tamas; Sunderland, Kyle R.; Zevin, Boris; Fichtinger, Gabor
Open source platform for automated collection of training data to support video-based feedback in surgical simulators Conference
SPIE Medical Imaging, SPIE, Houston, United States, 2020.
@conference{Laframboise2020a,
title = {Open source platform for automated collection of training data to support video-based feedback in surgical simulators},
author = {Jacob Laframboise and Tamas Ungi and Kyle R. Sunderland and Boris Zevin and Gabor Fichtinger},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {SPIE Medical Imaging},
publisher = {SPIE},
address = {Houston, United States},
abstract = {<p><strong>Purpose:</strong> Surgical training could be improved by automatic detection of workflow steps. A platform to collect and organize tracking and video data would enable rapid development of deep learning solutions for surgical training. The purpose of this research is to demonstrate 3D Slicer / PLUS Toolkit as a platform for video annotation by identifying and annotating tools interacting with tissues in simulated hernia repair. <strong>Methods:</strong> Tracking data from an optical tracker and video data from a camera are collected by PLUS and 3D Slicer. To demonstrate the platform in use, we identify tissues during a surgical procedure using a neural network. The tracking data is used to identify what tool is in use. The solution is deployed with a custom Slicer module. <strong>Results:</strong> This platform allowed the collection and storage of enough tracked video data for training a convolutional neural network (CNN) to detect interactions with tissues and tools. The CNN was trained on this data and applied to new data with a testing accuracy of 98%. The model’s predictions can be weighted over several frames with a custom Slicer module to improve accuracy. <strong>Conclusion:</strong> We found the 3D Slicer and PLUS Toolkit platform to be a viable platform for training and deploying a solution that combines automatic video processing and optical tool tracking. We designed a proof of concept model to identify tissues with a trained CNN in real time along with tracking of surgical tools.</p>},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
<p><strong>Purpose:</strong> Surgical training could be improved by automatic detection of workflow steps. A platform to collect and organize tracking and video data would enable rapid development of deep learning solutions for surgical training. The purpose of this research is to demonstrate 3D Slicer / PLUS Toolkit as a platform for video annotation by identifying and annotating tools interacting with tissues in simulated hernia repair. <strong>Methods:</strong> Tracking data from an optical tracker and video data from a camera are collected by PLUS and 3D Slicer. To demonstrate the platform in use, we identify tissues during a surgical procedure using a neural network. The tracking data is used to identify what tool is in use. The solution is deployed with a custom Slicer module. <strong>Results:</strong> This platform allowed the collection and storage of enough tracked video data for training a convolutional neural network (CNN) to detect interactions with tissues and tools. The CNN was trained on this data and applied to new data with a testing accuracy of 98%. The model’s predictions can be weighted over several frames with a custom Slicer module to improve accuracy. <strong>Conclusion:</strong> We found the 3D Slicer and PLUS Toolkit platform to be a viable platform for training and deploying a solution that combines automatic video processing and optical tool tracking. We designed a proof of concept model to identify tissues with a trained CNN in real time along with tracking of surgical tools.</p>
Wu, Victoria; Ungi, Tamas; Sunderland, Kyle R.; Pigeau, Grace; Schonewille, Abigael; Fichtinger, Gabor
Using multiple frame U-net for automated segmentation of spinal ultrasound images Conference
18th Annual Imaging Network Ontario (ImNO) Symposium, 2020.
@conference{Wu2020b,
title = {Using multiple frame U-net for automated segmentation of spinal ultrasound images},
author = {Victoria Wu and Tamas Ungi and Kyle R. Sunderland and Grace Pigeau and Abigael Schonewille and Gabor Fichtinger},
url = {https://www.imno.ca/sites/default/files/ImNO2020Proceedings.pdf
https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/Wu2020b.pdf},
year = {2020},
date = {2020-01-01},
urldate = {2020-01-01},
booktitle = {18th Annual Imaging Network Ontario (ImNO) Symposium},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Sunderland, Kyle R.; Pinter, Csaba; Lasso, Andras; Fichtinger, Gabor
Fractional labelmaps for computing accurate dose volume histograms Conference
SPIE Medical Imaging, International Society for Optics and Photonics International Society for Optics and Photonics, 2017.
@conference{sunderland2017a,
title = {Fractional labelmaps for computing accurate dose volume histograms},
author = {Kyle R. Sunderland and Csaba Pinter and Andras Lasso and Gabor Fichtinger},
url = {https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/Sunderland2017a.pdf},
year = {2017},
date = {2017-01-01},
urldate = {2017-01-01},
booktitle = {SPIE Medical Imaging},
publisher = {International Society for Optics and Photonics},
organization = {International Society for Optics and Photonics},
abstract = {<p><strong>PURPOSE:</strong> In radiation therapy treatment planning systems, structures are represented as parallel 2D contours. For treatment planning algorithms, structures must be converted into labelmap (i.e. 3D image denoting structure inside/outside) representations. This is often done by triangulated a surface from contours, which is converted into a binary labelmap. This surface to binary labelmap conversion can cause large errors in small structures. Binary labelmaps are often represented using one byte per voxel, meaning a large amount of memory is unused. Our goal is to develop a fractional labelmap representation containing non-binary values, allowing more information to be stored in the same amount of memory.</p>
<p><strong>METHODS:</strong> We implemented an algorithm in 3D Slicer, which converts surfaces to fractional labelmaps by creating 216 binary labelmaps, changing the labelmap origin on each iteration. The binary labelmap values are summed to create the fractional labelmap. In addition, an algorithm is implemented in the SlicerRT toolkit that calculates dose volume histograms (DVH) using fractional labelmaps.</p>
<p><strong>RESULTS:</strong> We found that with manually segmented RANDO® head and neck structures, fractional labelmaps represented structure volume up to 19.07% (average 6.81%) more accurately than binary labelmaps, while occupying the same amount of memory. When compared to baseline DVH from treatment planning software, DVH from fractional labelmaps had agreement acceptance percent (1% ΔD, 1% ΔV) up to 57.46% higher (average 4.33%) than DVH from binary labelmaps.</p>
<p><strong>CONCLUSION:</strong> Fractional labelmaps promise to be an effective method for structure representation, allowing considerably more information to be stored in the same amount of memory</p>},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
<p><strong>PURPOSE:</strong> In radiation therapy treatment planning systems, structures are represented as parallel 2D contours. For treatment planning algorithms, structures must be converted into labelmap (i.e. 3D image denoting structure inside/outside) representations. This is often done by triangulated a surface from contours, which is converted into a binary labelmap. This surface to binary labelmap conversion can cause large errors in small structures. Binary labelmaps are often represented using one byte per voxel, meaning a large amount of memory is unused. Our goal is to develop a fractional labelmap representation containing non-binary values, allowing more information to be stored in the same amount of memory.</p>
<p><strong>METHODS:</strong> We implemented an algorithm in 3D Slicer, which converts surfaces to fractional labelmaps by creating 216 binary labelmaps, changing the labelmap origin on each iteration. The binary labelmap values are summed to create the fractional labelmap. In addition, an algorithm is implemented in the SlicerRT toolkit that calculates dose volume histograms (DVH) using fractional labelmaps.</p>
<p><strong>RESULTS:</strong> We found that with manually segmented RANDO® head and neck structures, fractional labelmaps represented structure volume up to 19.07% (average 6.81%) more accurately than binary labelmaps, while occupying the same amount of memory. When compared to baseline DVH from treatment planning software, DVH from fractional labelmaps had agreement acceptance percent (1% ΔD, 1% ΔV) up to 57.46% higher (average 4.33%) than DVH from binary labelmaps.</p>
<p><strong>CONCLUSION:</strong> Fractional labelmaps promise to be an effective method for structure representation, allowing considerably more information to be stored in the same amount of memory</p>
<p><strong>METHODS:</strong> We implemented an algorithm in 3D Slicer, which converts surfaces to fractional labelmaps by creating 216 binary labelmaps, changing the labelmap origin on each iteration. The binary labelmap values are summed to create the fractional labelmap. In addition, an algorithm is implemented in the SlicerRT toolkit that calculates dose volume histograms (DVH) using fractional labelmaps.</p>
<p><strong>RESULTS:</strong> We found that with manually segmented RANDO® head and neck structures, fractional labelmaps represented structure volume up to 19.07% (average 6.81%) more accurately than binary labelmaps, while occupying the same amount of memory. When compared to baseline DVH from treatment planning software, DVH from fractional labelmaps had agreement acceptance percent (1% ΔD, 1% ΔV) up to 57.46% higher (average 4.33%) than DVH from binary labelmaps.</p>
<p><strong>CONCLUSION:</strong> Fractional labelmaps promise to be an effective method for structure representation, allowing considerably more information to be stored in the same amount of memory</p>
Sunderland, Kyle R.; Pinter, Csaba; Lasso, Andras; Fichtinger, Gabor
Effects of voxelization on dose volume histogram accuracy Conference
SPIE Medical Imaging, International Society for Optics and Photonics International Society for Optics and Photonics, 2016.
@conference{sunderland2016a,
title = {Effects of voxelization on dose volume histogram accuracy},
author = {Kyle R. Sunderland and Csaba Pinter and Andras Lasso and Gabor Fichtinger},
url = {https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/sunderland2016a_0.pdf},
year = {2016},
date = {2016-01-01},
urldate = {2016-01-01},
booktitle = {SPIE Medical Imaging},
pages = {97862O–97862O},
publisher = {International Society for Optics and Photonics},
organization = {International Society for Optics and Photonics},
abstract = {<p><strong>PURPOSE:</strong> In radiotherapy treatment planning systems, structures of interest such as targets and organs at risk are stored as 2D contours on evenly spaced planes. In order to be used in various algorithms, contours must be converted into binary labelmap volumes using voxelization. The voxelization process results in lost information, which has little effect on the volume of large structures, but has significant impact on small structures, which contain few voxels. Volume differences for segmented structures affects metrics such as dose volume histograms (DVH), which are used for treatment planning. Our goal is to evaluate the impact of voxelization on segmented structures, as well as how factors like voxel size affects metrics, such as DVH.</p>
<p><strong>METHODS:</strong> We create a series of implicit functions, which represent simulated structures. These structures are sampled at varying resolutions, and compared to labelmaps with high sub-millimeter resolutions. We generate DVH and evaluate voxelization error for the same structures at different resolutions by calculating the agreement acceptance percentage between the DVH.</p>
<p><strong>RESULTS:</strong> We implemented tools for analysis as modules in the SlicerRT toolkit based on the 3D Slicer platform. We found that there were large DVH variation from the baseline for small structures or for structures located in regions with a high dose gradient, potentially leading to the creation of suboptimal treatment plans.</p>
<p><strong>CONCLUSION:</strong> This work demonstrates that labelmap and dose volume voxel size is an important factor in DVH accuracy, which must be accounted for in order to ensure the development of accurate treatment plans. </p>},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
<p><strong>PURPOSE:</strong> In radiotherapy treatment planning systems, structures of interest such as targets and organs at risk are stored as 2D contours on evenly spaced planes. In order to be used in various algorithms, contours must be converted into binary labelmap volumes using voxelization. The voxelization process results in lost information, which has little effect on the volume of large structures, but has significant impact on small structures, which contain few voxels. Volume differences for segmented structures affects metrics such as dose volume histograms (DVH), which are used for treatment planning. Our goal is to evaluate the impact of voxelization on segmented structures, as well as how factors like voxel size affects metrics, such as DVH.</p>
<p><strong>METHODS:</strong> We create a series of implicit functions, which represent simulated structures. These structures are sampled at varying resolutions, and compared to labelmaps with high sub-millimeter resolutions. We generate DVH and evaluate voxelization error for the same structures at different resolutions by calculating the agreement acceptance percentage between the DVH.</p>
<p><strong>RESULTS:</strong> We implemented tools for analysis as modules in the SlicerRT toolkit based on the 3D Slicer platform. We found that there were large DVH variation from the baseline for small structures or for structures located in regions with a high dose gradient, potentially leading to the creation of suboptimal treatment plans.</p>
<p><strong>CONCLUSION:</strong> This work demonstrates that labelmap and dose volume voxel size is an important factor in DVH accuracy, which must be accounted for in order to ensure the development of accurate treatment plans. </p>
<p><strong>METHODS:</strong> We create a series of implicit functions, which represent simulated structures. These structures are sampled at varying resolutions, and compared to labelmaps with high sub-millimeter resolutions. We generate DVH and evaluate voxelization error for the same structures at different resolutions by calculating the agreement acceptance percentage between the DVH.</p>
<p><strong>RESULTS:</strong> We implemented tools for analysis as modules in the SlicerRT toolkit based on the 3D Slicer platform. We found that there were large DVH variation from the baseline for small structures or for structures located in regions with a high dose gradient, potentially leading to the creation of suboptimal treatment plans.</p>
<p><strong>CONCLUSION:</strong> This work demonstrates that labelmap and dose volume voxel size is an important factor in DVH accuracy, which must be accounted for in order to ensure the development of accurate treatment plans. </p>
Sunderland, Kyle R.; Pinter, Csaba; Lasso, Andras; Fichtinger, Gabor
Analysis of dose volume histogram deviations using different voxelization parameters Conference
14th Annual Imaging Network Ontario Symposium (ImNO), Toronto, Canada, 2016.
@conference{Sunderland2016b,
title = {Analysis of dose volume histogram deviations using different voxelization parameters},
author = {Kyle R. Sunderland and Csaba Pinter and Andras Lasso and Gabor Fichtinger},
url = {https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/Sunderland2016b.pdf},
year = {2016},
date = {2016-01-01},
urldate = {2016-01-01},
booktitle = {14th Annual Imaging Network Ontario Symposium (ImNO)},
address = {Toronto, Canada},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
Sunderland, Kyle R.; Woo, Boyeong; Pinter, Csaba; Fichtinger, Gabor
Reconstruction of surfaces from planar contours through contour interpolation Conference
SPIE Medical Imaging 2015, vol. 9415, 2015.
@conference{Sunderland2015,
title = {Reconstruction of surfaces from planar contours through contour interpolation},
author = {Kyle R. Sunderland and Boyeong Woo and Csaba Pinter and Gabor Fichtinger},
url = {http://dx.doi.org/10.1117/12.2081436
https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/Sunderland2015-manuscript.pdf
https://labs.cs.queensu.ca/perklab/wp-content/uploads/sites/3/2024/02/Sunderland2015-poster.pdf},
doi = {10.1117/12.2081436},
year = {2015},
date = {2015-01-01},
urldate = {2015-01-01},
booktitle = {SPIE Medical Imaging 2015},
volume = {9415},
pages = {94151R-94151R-8},
abstract = {<p>Segmented structures such as targets or organs at risk are typically stored as 2D contours contained on evenly spaced cross sectional images (slices). Contour interpolation algorithms are implemented in radiation oncology treatment planning software to turn 2D contours into a 3D surface, however the results differ between algorithms, causing discrepancies in analysis. Our goal was to create an accurate and consistent contour interpolation algorithm that can handle issues such as keyhole contours, rapid changes, and branching. This was primarily motivated by radiation therapy research using the open source SlicerRT extension for the 3D Slicer platform. The implemented algorithm triangulates the mesh by minimizing the length of edges spanning the contours with dynamic programming. The first step in the algorithm is removing keyholes from contours. Correspondence is then found between contour layers and branching patterns are determined. The final step is triangulating the contours and sealing the external contours. The algorithm was tested on contours segmented on computed tomography (CT) images. Some cases such as inner contours, rapid changes in contour size, and branching were handled well by the algorithm when encountered individually. There were some special cases in which the simultaneous occurrence of several of these problems in the same location could cause the algorithm to produce suboptimal mesh. An open source contour interpolation algorithm was implemented in SlicerRT for reconstructing surfaces from planar contours. The implemented algorithm was able to generate qualitatively good 3D mesh from the set of 2D contours for most tested structures.</p>},
keywords = {},
pubstate = {published},
tppubtype = {conference}
}
<p>Segmented structures such as targets or organs at risk are typically stored as 2D contours contained on evenly spaced cross sectional images (slices). Contour interpolation algorithms are implemented in radiation oncology treatment planning software to turn 2D contours into a 3D surface, however the results differ between algorithms, causing discrepancies in analysis. Our goal was to create an accurate and consistent contour interpolation algorithm that can handle issues such as keyhole contours, rapid changes, and branching. This was primarily motivated by radiation therapy research using the open source SlicerRT extension for the 3D Slicer platform. The implemented algorithm triangulates the mesh by minimizing the length of edges spanning the contours with dynamic programming. The first step in the algorithm is removing keyholes from contours. Correspondence is then found between contour layers and branching patterns are determined. The final step is triangulating the contours and sealing the external contours. The algorithm was tested on contours segmented on computed tomography (CT) images. Some cases such as inner contours, rapid changes in contour size, and branching were handled well by the algorithm when encountered individually. There were some special cases in which the simultaneous occurrence of several of these problems in the same location could cause the algorithm to produce suboptimal mesh. An open source contour interpolation algorithm was implemented in SlicerRT for reconstructing surfaces from planar contours. The implemented algorithm was able to generate qualitatively good 3D mesh from the set of 2D contours for most tested structures.</p>