Limits...
Segmentation of bones in magnetic resonance images of the wrist.

Włodarczyk J, Czaplicka K, Tabor Z, Wojciechowski W, Urbanik A - Int J Comput Assist Radiol Surg (2014)

Bottom Line: The automated segmentation results were compared with gold-standard manual segmentations using a few well-established metrics: area under ROC curve AUC, mean similarity MS and mean absolute distance MAD.The mean (standard deviation) values of AUC, MS and MAD were 0.97 (0.04), 0.93 (0.09) and 1.23 (0.28), respectively.The results of the present study demonstrate that automated segmentation of wrist bones is feasible.

View Article: PubMed Central - PubMed

Affiliation: Jagiellonian University, Reymonta 4, 30-059,  Krakow, Poland, justyna.wlodarczyk@uj.edu.pl.

ABSTRACT

Purpose: Rheumatoid arthritis (RA) is a disease characterized by progressive and irreversible destruction of bones and joints. According to current recommendations, magnetic resonance imaging (MRI) is used to asses three main signs of RA based on manual evaluation of MR images: synovitis, bone edema and bone erosions. The key feature of a future computer-assisted diagnostic system for evaluation RA lesions is accurate segmentation of 15 wrist bones. In the present paper, we focus on developing a wrist bones segmentation framework.

Method: The segmentation procedure consisted of three stages: segmentation of the distal parts of ulna and radius, segmentation of the proximal parts of metacarpal bones and segmentation of carpal bones. At every stage, markers of bones were determined first, using an atlas-based approach. Then, given markers of bones and a marker of background, a watershed from markers algorithm was applied to find the final segmentation.

Results: The MR data for 37 cases were analyzed. The automated segmentation results were compared with gold-standard manual segmentations using a few well-established metrics: area under ROC curve AUC, mean similarity MS and mean absolute distance MAD. The mean (standard deviation) values of AUC, MS and MAD were 0.97 (0.04), 0.93 (0.09) and 1.23 (0.28), respectively.

Conclusion: The results of the present study demonstrate that automated segmentation of wrist bones is feasible. The proposed algorithm can be the first stage for the detection of early lesions like bone edema or synovitis.

Show MeSH

Related in: MedlinePlus

Flowchart of the wrist bones segmentation algorithm: boxes with gray background and thin borders denote atlas input, boxes with white background and thick borders denote partial segmentation results, boxes with gray background and thick borders denote final segmentation results, a box with black background denote a step in which user interaction is required
© Copyright Policy - OpenAccess
Related In: Results  -  Collection


getmorefigures.php?uid=PMC4379414&req=5

Fig2: Flowchart of the wrist bones segmentation algorithm: boxes with gray background and thin borders denote atlas input, boxes with white background and thick borders denote partial segmentation results, boxes with gray background and thick borders denote final segmentation results, a box with black background denote a step in which user interaction is required

Mentions: The flowchart diagram of the wrist bones segmentation algorithm is presented in Fig. 2. The core of the algorithm is the atlas-based segmentation, which requires registration of an actual image with an atlas image. This is a natural choice since the anatomy of analyzed objects is fixed to a large extent. Because registration in three dimensions is computationally expensive, we chose to divide the segmentation problem into subproblems. Thus, the algorithm consists of four modules: segmentation of the hand mask, segmentation of the distal parts of the ulna and radius, segmentation of the metacarpal bases and segmentation of carpals. We use registration in all steps apart from mask segmentation: In the first two cases, we use a relatively cheap 2D algorithm and a 3D algorithm in the third step. The segmentation procedure uses the “watershed from markers” algorithm [13]. At the input of this algorithm, one must provide a magnitude of gradient image and a marker image. The gradient image can be interpreted as a topographic relief, where the gradient value at a voxel is equivalent to its altitude in the relief. The watershed of a relief corresponds to the limits of the adjacent catchment basins, and each catchment basin contains exactly one marker. The gradient components were calculated using Prewitt’s filter for an original MR image blurred with a Gaussian filter with fixed half-size equal to 1 voxel. The atlas-based segmentation provides a method for automated selection of markers, enabling robust watershed-based segmentation of the wrist bones. The markers constructed by our algorithm can be an input for other procedures (e.g., region growing or deformable models) but given the results of the evaluation of watershed-based segmentation, we expect in such cases only marginal if any improvement of the segmentation results.Fig. 2


Segmentation of bones in magnetic resonance images of the wrist.

Włodarczyk J, Czaplicka K, Tabor Z, Wojciechowski W, Urbanik A - Int J Comput Assist Radiol Surg (2014)

Flowchart of the wrist bones segmentation algorithm: boxes with gray background and thin borders denote atlas input, boxes with white background and thick borders denote partial segmentation results, boxes with gray background and thick borders denote final segmentation results, a box with black background denote a step in which user interaction is required
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC4379414&req=5

Fig2: Flowchart of the wrist bones segmentation algorithm: boxes with gray background and thin borders denote atlas input, boxes with white background and thick borders denote partial segmentation results, boxes with gray background and thick borders denote final segmentation results, a box with black background denote a step in which user interaction is required
Mentions: The flowchart diagram of the wrist bones segmentation algorithm is presented in Fig. 2. The core of the algorithm is the atlas-based segmentation, which requires registration of an actual image with an atlas image. This is a natural choice since the anatomy of analyzed objects is fixed to a large extent. Because registration in three dimensions is computationally expensive, we chose to divide the segmentation problem into subproblems. Thus, the algorithm consists of four modules: segmentation of the hand mask, segmentation of the distal parts of the ulna and radius, segmentation of the metacarpal bases and segmentation of carpals. We use registration in all steps apart from mask segmentation: In the first two cases, we use a relatively cheap 2D algorithm and a 3D algorithm in the third step. The segmentation procedure uses the “watershed from markers” algorithm [13]. At the input of this algorithm, one must provide a magnitude of gradient image and a marker image. The gradient image can be interpreted as a topographic relief, where the gradient value at a voxel is equivalent to its altitude in the relief. The watershed of a relief corresponds to the limits of the adjacent catchment basins, and each catchment basin contains exactly one marker. The gradient components were calculated using Prewitt’s filter for an original MR image blurred with a Gaussian filter with fixed half-size equal to 1 voxel. The atlas-based segmentation provides a method for automated selection of markers, enabling robust watershed-based segmentation of the wrist bones. The markers constructed by our algorithm can be an input for other procedures (e.g., region growing or deformable models) but given the results of the evaluation of watershed-based segmentation, we expect in such cases only marginal if any improvement of the segmentation results.Fig. 2

Bottom Line: The automated segmentation results were compared with gold-standard manual segmentations using a few well-established metrics: area under ROC curve AUC, mean similarity MS and mean absolute distance MAD.The mean (standard deviation) values of AUC, MS and MAD were 0.97 (0.04), 0.93 (0.09) and 1.23 (0.28), respectively.The results of the present study demonstrate that automated segmentation of wrist bones is feasible.

View Article: PubMed Central - PubMed

Affiliation: Jagiellonian University, Reymonta 4, 30-059,  Krakow, Poland, justyna.wlodarczyk@uj.edu.pl.

ABSTRACT

Purpose: Rheumatoid arthritis (RA) is a disease characterized by progressive and irreversible destruction of bones and joints. According to current recommendations, magnetic resonance imaging (MRI) is used to asses three main signs of RA based on manual evaluation of MR images: synovitis, bone edema and bone erosions. The key feature of a future computer-assisted diagnostic system for evaluation RA lesions is accurate segmentation of 15 wrist bones. In the present paper, we focus on developing a wrist bones segmentation framework.

Method: The segmentation procedure consisted of three stages: segmentation of the distal parts of ulna and radius, segmentation of the proximal parts of metacarpal bones and segmentation of carpal bones. At every stage, markers of bones were determined first, using an atlas-based approach. Then, given markers of bones and a marker of background, a watershed from markers algorithm was applied to find the final segmentation.

Results: The MR data for 37 cases were analyzed. The automated segmentation results were compared with gold-standard manual segmentations using a few well-established metrics: area under ROC curve AUC, mean similarity MS and mean absolute distance MAD. The mean (standard deviation) values of AUC, MS and MAD were 0.97 (0.04), 0.93 (0.09) and 1.23 (0.28), respectively.

Conclusion: The results of the present study demonstrate that automated segmentation of wrist bones is feasible. The proposed algorithm can be the first stage for the detection of early lesions like bone edema or synovitis.

Show MeSH
Related in: MedlinePlus