Projects:NerveSegmentation - Revision history

Adalca: /* Results */

2012-11-28T19:44:25Z

Results

Adalca: /* Results */

2012-11-28T19:32:47Z

Results

Adalca: /* Results */

2012-11-28T19:31:14Z

Results

Adalca: /* Data */

2012-11-28T19:21:29Z

Data

Adalca: /* Data */

2012-11-28T19:19:44Z

Data

Adalca: /* Results */

2011-10-28T18:02:26Z

Results

Adalca at 17:51, 28 October 2011

2011-10-28T17:51:15Z

Adalca: /* Publications */

2011-10-28T17:43:58Z

Publications

Adalca: /* Key Investigators */

2011-10-28T17:41:18Z

Key Investigators

Adalca: /* Results */

2011-10-28T17:35:55Z

Results

@@ Line 40: / Line 40: @@
 nerve, as we see the margins of the nerve underneath the segmentation.]]
-We used our method on MRI scans of the spine in ten nerve bundles
+We evaluate our algorithm on synthetic and patient data. We show that it can
-from five subjects. We obtain both expert and automatic
+fully segment the nerves from the initial input point up to the end of the ganglia,
-segmentations of nerve bundles and ganglia that were deemed traceable,
+and provide accurate estimates of the nerve thickness. To quantify the accuracy of our
-and evaluate tracing accuracy for all bundles.
+method, we measure the distances between the desired and automatic nerve surfaces.
-The mean distance between automatically extracted centerlines and their
+First, we generate synthetic nerve tracks via our particle model. We vary the dynamics
-manual counterparts is within 1 voxel, 90% of automatically extracted
+parameters, such as the radius and control points, beyond ranges that are observed in
-centerlines are within 0.9 voxels of the expert centerline, and 90% of expert
+real images, to allow for tracks with more irregular behavior. The image intensity is
-centerlines are within 2.2 voxels of the algorithm centerline. Visually,
+then formed by adding white noise to the predicted nerve image. We superimpose the nerve
-we find very good centerline alignment (≤ 1 voxel) inside the vertebral canal.
+tracks on backgrounds with no noise (blank), Gaussian noise, Perlin noise, and random sections of MRI volumes, which present varying
-The greater disagreement is present in the ganglia, where the algorithm tends to
+degrees of segmentation dificulty.
-We find that since the edges of nerve bundles and ganglia are subject to
+We evaluate the algorithm on 20 images with each background. Typical diameters measure between
-the partial volume effects and are often near other anatomical structures, the
+four and ten voxels, yielding many partial volume voxels between nerve and background.
-algorithm slightly over-estimates the extent of the nerves in some regions of
+We also construct and test a second synthetic dataset of the same size, where the
-the image. As mentioned above, inside the thick ganglia the algorithm under-segments
+initial nerve intensity is generated from smoothed binary map of expert nerve
-due to the ganglia’s more irregular shape. The mean distance from
+segmentation in patient data. The nerve and background noise models are applied in
-the automatically extracted surface to the expert surface is 1.1 voxels
+the same manner as in the first synthetic set. The nerves range from four to 25 voxels
-and the 90th percentile is at 2.3 voxels. Visual inspection reveals that
+in diameter, and present with far more irregular shape.
-the expert segmentation can be irregular and include small deformations or
+The median distance between the automatically extracted and true nerve surfaces for each nerve generated from our particle model is
-protrusions, especially in pathologies. Since the algorithm attempts to maintain
+.8±0.4 voxels, with a 90th percentile of 2.1±0.7 voxels. Most of the algorithm errors are in the edge, partial-volume voxels. For nerves generated from expert segmentations,
-the estimation within the ganglia, the correct outer surface voxels in areas of
+the shape irregularity results in an increased median distance between the automatically extracted and true nerve surfaces of 1.3 ± 0.3 voxels, with the 90th percentile of
-small irregularities will be more distant from the algorithm prediction. This
+.1 ± 0.4 voxels.
-We also evaluated the algorithm on the nerves following the ganglia, where
-they split up into several thinner peripheral nerves. Here, the algorithm often
-loses some nerves due to loss of contrast and the small radii of the nerves.
+We further demonstrate our method on MRI spine scans of 12 nerve bundle segmen-
-When the algorithm continues to track, we observe a fully estimated path (the
+tations from six subjects. The scans were acquired 3D Wide-band Steady State Free
-segmentation follows the nerve), but with over-segmentation as the filter is driven
+Precession sequence. These
-by stronger edges from the neighboring anatomical structures.
+include four nerves in two pathologies where the nerves have been displaced by disk
 = Conclusion =

@@ Line 34: / Line 34: @@
 [[File:NerveSegResults3D.png|800px|thumb|center|
-Patient data results. Left: Rendered segmentation results { a herniated disk (manually
+Patient data results. Left: Rendered segmentation results a herniated disk (manually
 segmented, yellow) is impinging on the nerve tracts (segmented with our algorithm with minimal user
 input, green). Right: Slice that includes a section of a nerve and a ganglion. The yellow outline

@@ Line 21: / Line 21: @@
 = Results =
-[[File:NerveSegResults.png|800px|thumb|center|
-Left: Example segmentation results – the rightmost nerve shows results without
-post-processing pruning, while the left segmentation was processed after completion
-of tracking. Right: Cumulative distribution
+[[File:NerveSegResultsGraphs.png|800px|thumb|center|
-functions of distances (in voxels) between the automatically extracted and manual
+Left: Summary of synthetic results for the eight types of synthetic nerves and backgrounds. The
-segmentations: centerline distances on top and surfaces on the bottom. Blue plots indicate
+red bars indicate the average of the median distance between the automatic and the manual outline. The
-differences from algorithm to expert segmentation voxels, red plots correspond to
+blue bars indicate the 90th percentile. For each category, Model indicates that we generated the nerve
-expert-to-algorithm distances. Error bars indicate standard deviations. Dotted vertical
+via our particle model, and Expert indicates nerves generated from the smoothed expert segmentation
-lines indicate mean (lower) and 90th percentiles distances (higher).]]
+of real nerves. Right: Summary of the 12 segmentation results on real patient data. The red bars indicate the
 We used our method on MRI scans of the spine in ten nerve bundles

@@ Line 9: / Line 9: @@
 = Data =
-[[File:NerveSegMRE2.png|800px|thumb|center|We show four slices of an example High-Resolution MR scan. All arrows point to the same nerve bundle. Blue arrows show examples of poor contrast between the nerve and the surrounding tissue; orange arrows indicate the thickening of the neural tract into a ganglion.
+[[File:NerveSegMRE2.png|800px|thumb|center|We show four slices of an example High-Resolution MR scan. The blue outline in the first figure shows a manual segmentation of a nerve for reference. All arrows point to the same nerve bundle. Blue arrows show examples of poor contrast between the nerve and the surrounding tissue; orange arrows indicate the thickening of the neural tract into a ganglion.
 ]]

@@ Line 9: / Line 9: @@
 = Data =
-[[File:NerveSegMREg.png|800px|thumb|center|We show three slices of an example High-Resolution MR scan. All arrows point to the same nerve bundle. Blue arrows show examples of poor contrast between the nerve and the surrounding tissue; orange arrows indicate the thickening of the neural tract into a ganglion.
+[[File:NerveSegMRE2.png|800px|thumb|center|We show four slices of an example High-Resolution MR scan. All arrows point to the same nerve bundle. Blue arrows show examples of poor contrast between the nerve and the surrounding tissue; orange arrows indicate the thickening of the neural tract into a ganglion.
 ]]

@@ Line 74: / Line 74: @@
 As shown in the results, the proposed segmentation may slightly over-segment (usually by at most two voxels) in thin areas and under-segment in thick areas, but will give a very good estimation of the nerve core and location.
 =Key Investigators=
 * MIT: Adrian Dalca, Polina Golland
 * BWH: Giovanna Danagoulian, Ehud Schmidt, Ron Kikinis
 = Publications =
 *[http://www.na-mic.org/publications/pages/display?search=Projects%3ANerveSegmentation&submit=Search&words=all&title=checked&keywords=checked&authors=checked&abstract=checked&sponsors=checked&searchbytag=checked| NA-MIC Publications Database on nerve segmentation]

← Older revision		Revision as of 17:43, 28 October 2011
Line 82:		Line 82:

	= Publications =		= Publications =
−	*[http://www.na-mic.org/publications/pages/display?search=Projects%3ANerveSegmentation&submit=Search&words=all&title=checked&keywords=checked&authors=checked&abstract=checked&sponsors=checked&searchbytag=checked\| NA-MIC Publications Database on ~~brain tumor~~ segmentation ~~and modeling~~]	+	*[http://www.na-mic.org/publications/pages/display?search=Projects%3ANerveSegmentation&submit=Search&words=all&title=checked&keywords=checked&authors=checked&abstract=checked&sponsors=checked&searchbytag=checked\| NA-MIC Publications Database on nerve segmentation]