Spinal cord gray matter segmentation using deep dilated convolutions

Perone, Christian S.; Calabrese, Evan; Cohen‐Adad, Julien

doi:10.1038/s41598-018-24304-3

Cited by 110 publications

(88 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A well-known fact regarding the Dice loss is that it usually produces predictions concentrated around the upper and lower bounds of the probability distribution, with very low entropy. As in (Perone & Cohen-Adad, 2018b), we used a high threshold value (0.99) for the Dice predictions to produce a balanced model. We have found, however, that the domain adaptation method also regularizes the network predictions, shifting the Dice probability distribution outside of the probability bounds.…”

Section: Behavior Of Dice Loss and Thresholdingmentioning

confidence: 99%

Unsupervised domain adaptation for medical imaging segmentation with self-ensembling

et al. 2019

Self Cite

View full text Add to dashboard Cite

Recent advances in deep learning methods have come to define the state-of-the-art for many medical imaging applications, surpassing even human judgment in several tasks. Those models, however, when trained to reduce the empirical risk on a single domain, fail to generalize when applied to other domains, a very common scenario in medical imaging due to the variability of images and anatomical structures, even across the same imaging modality. In this work, we extend the method of unsupervised domain adaptation using self-ensembling for the semantic segmentation task and explore multiple facets of the method on a small and realistic publiclyavailable magnetic resonance (MRI) dataset. Through an extensive evaluation, we show that self-ensembling can indeed improve the generalization of the models even when using a small amount of unlabelled data.

show abstract

Section: Behavior Of Dice Loss and Thresholdingmentioning

confidence: 99%

Unsupervised domain adaptation for medical imaging segmentation with self-ensembling

et al. 2019

Self Cite

View full text Add to dashboard Cite

show abstract

“…3.2) for the challenge models. To address the high class imbalance between background, WM and GM, similar to [5] we added a GM Dice loss (DL), but also included DLs for all the other label classes using the generalized Dice loss (GDL) formulation of Sudre et al [8].…”

Section: Methodsmentioning

confidence: 99%

“…This can be explained, because the GM boundary is part of the WM boundaries and thus influences the WM scores, and furthermore the outer WM boundary is already well delineated even without any DL through the good CSF-WM contrast. Choosing a DL as a surrogate for GM DSC only, as proposed in [5], is thus justifiable. While the SCGM challenge results only provide GM segmentation accuracy, for the AMIRA dataset we additionally also provide WM segmentation results.…”

Section: Amira Segmentation Modelmentioning

confidence: 99%

See 1 more Smart Citation

Spinal Cord Gray Matter-White Matter Segmentation on Magnetic Resonance AMIRA Images with MD-GRU

Horvath

Tsagkas

Andermatt

et al. 2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

The small butterfly shaped structure of spinal cord (SC) gray matter (GM) is challenging to image and to delinate from its surrounding white matter (WM). Segmenting GM is up to a point a trade-off between accuracy and precision. We propose a new pipeline for GM-WM magnetic resonance (MR) image acquisition and segmentation. We report superior results as compared to the ones recently reported in the SC GM segmentation challenge and show even better results using the averaged magnetization inversion recovery acquisitions (AMIRA) sequence. Scan-rescan experiments with the AMIRA sequence show high reproducibility in terms of Dice coefficient, Hausdorff distance and relative standard deviation. We use a recurrent neural network (RNN) with multi-dimensional gated recurrent units (MD-GRU) to train segmentation models on the AMIRA dataset of 855 slices. We added a generalized dice loss to the cross entropy loss that MD-GRU uses and were able to improve the results.

show abstract

“…As the shape of the data of each layer of a CNN is generally unrestricted, taking the processing stream from one CNN and adding it to the function of another CNN is possible (e.g. [15][16][17]). Based on this idea, this study combines a CNN for segmentation with another simple CNN designed for general non-linear reconstruction so that the finally generated myelin volume index (GenMVI) is more specific to the characteristics of the tissue in each pixel.…”

Section: Introductionmentioning

confidence: 99%

The Utility of a Convolutional Neural Network for Generating a Myelin Volume Index Map from Rapid Simultaneous Relaxometry Imaging

et al. 2020

View full text Add to dashboard Cite

Background and Purpose: A current algorithm to obtain a synthetic myelin volume fraction map (SyMVF) from rapid simultaneous relaxometry imaging (RSRI) has a potential problem, that it does not incorporate information from surrounding pixels. The purpose of this study was to develop a method that utilizes a convolutional neural network (CNN) to overcome this problem. Methods: RSRI and magnetization transfer images from 20 healthy volunteers were included. A CNN was trained to reconstruct RSRI-related metric maps into a myelin volume-related index (generated myelin volume index: GenMVI) map using the myelin volume index map calculated from magnetization transfer images (MTMVI) as reference. The SyMVF and GenMVI maps were statistically compared by testing how well they correlated with the MTMVI map. The correlations were evaluated based on: (i) averaged values obtained from 164 atlas-based ROIs, and (ii) pixel-based comparison for ROIs defined in four different tissue types (cortical and subcortical gray matter, white matter, and whole brain). Results: For atlas-based ROIs, the overall correlation with the MTMVI map was higher for the GenMVI map than for the SyMVF map. In the pixel-based comparison, correlation with the MTMVI map was stronger for the GenMVI map than for the SyMVF map, and the difference in the distribution for the volunteers was significant (Wilcoxon sign-rank test, P<.001) in all tissue types. Conclusion: The proposed method is useful, as it can incorporate more specific information about local tissue properties than the existing method.

show abstract

Spinal cord gray matter segmentation using deep dilated convolutions

Cited by 110 publications

References 61 publications

Unsupervised domain adaptation for medical imaging segmentation with self-ensembling

Unsupervised domain adaptation for medical imaging segmentation with self-ensembling

Spinal Cord Gray Matter-White Matter Segmentation on Magnetic Resonance AMIRA Images with MD-GRU

The Utility of a Convolutional Neural Network for Generating a Myelin Volume Index Map from Rapid Simultaneous Relaxometry Imaging

Contact Info

Product

Resources

About