Comparing 3D, 2.5D, and 2D Approaches to Brain Image Auto-Segmentation

Avesta, Arman; Hossain, S. M. Khaled; Lin, Ming De; Aboian, Mariam; Krumholz, Harlan M.; Aneja, Sanjay

doi:10.3390/bioengineering10020181

Cited by 23 publications

(8 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our more favorable results in segmenting the hippocampus are likely because of the 3D structure of our CapsNet, which can use the contextual information in the volume of the image rather than just a section of the image to better segment the complex shape of the hippocampus. 39 Our study has several limitations. Our models were tested on only 3 brain structures that are commonly segmented on brain MRIs, meaning that our findings may not generalize across other imaging modalities and anatomic structures.…”

Section: Discussionmentioning

confidence: 91%

3D Capsule Networks for Brain Image Segmentation

Avesta¹,

Hui²,

Aboian³

et al. 2023

AJNR Am J Neuroradiol

Self Cite

View full text Add to dashboard Cite

BACKGROUND AND PURPOSE: Current autosegmentation models such as UNets and nnUNets have limitations, including the inability to segment images that are not represented during training and lack of computational efficiency. 3D capsule networks have the potential to address these limitations. MATERIALS AND METHODS:We used 3430 brain MRIs, acquired in a multi-institutional study, to train and validate our models. We compared our capsule network with standard alternatives, UNets and nnUNets, on the basis of segmentation efficacy (Dice scores), segmentation performance when the image is not well-represented in the training data, performance when the training data are limited, and computational efficiency including required memory and computational speed. RESULTS:The capsule network segmented the third ventricle, thalamus, and hippocampus with Dice scores of 95%, 94%, and 92%, respectively, which were within 1% of the Dice scores of UNets and nnUNets. The capsule network significantly outperformed UNets in segmenting images that were not well-represented in the training data, with Dice scores 30% higher. The computational memory required for the capsule network is less than one-tenth of the memory required for UNets or nnUNets. The capsule network is also .25% faster to train compared with UNet and nnUNet. CONCLUSIONS:We developed and validated a capsule network that is effective in segmenting brain images, can segment images that are not well-represented in the training data, and is computationally efficient compared with alternatives.ABBREVIATIONS: CapsNet ¼ capsule network; Conv1 ¼ first network layer made of convolutional operators; ConvCaps3 ¼ third network layer made of convolutional capsules; ConvCaps4 ¼ fourth network layer made of convolutional capsules; DeconvCaps8 ¼ eighth network layer made of deconvolutional capsules; FinalCaps13 ¼ final thirteenth network layer made of capsules; FinalCaps13 ¼ final layer; GPU ¼ graphics processing unit; PrimaryCaps2 ¼ second network layer made of primary capsules N euroanatomic image segmentation is an important component in the management of various neurologic disorders. [1][2][3] Accurate segmentation of anatomic structures on brain MRIs is an essential step in a variety of neurosurgical and radiation therapy procedures. 1,[3][4][5][6] Manual segmentation is time-consuming and is prone to intra-and interobserver variability. 7,8 With the advent of deep learning to automate various image-analysis tasks, 9,10 there has been increasing enthusiasm for using deep learning for brain image autosegmentation. [11][12][13][14] UNets are among the most popular and successful deep learning autosegmentation algorithms. 11,15-17 Despite the broad success of UNets in segmenting anatomic structures across various imaging modalities, they have well-described limitations. UNets perform best on images that closely resemble the images used for training but underperform on images that contain variant

show abstract

Section: Discussionmentioning

confidence: 91%

3D Capsule Networks for Brain Image Segmentation

Avesta¹,

Hui²,

Aboian³

et al. 2023

AJNR Am J Neuroradiol

Self Cite

View full text Add to dashboard Cite

show abstract

“…Postprocessing was carried out in a 3D Slicer module (http://www.slicer.org) to increase speed, remove noise, and maximize contours of segmentations. The code for this is available 22 . The average DSC for the ossicles was 0.84 in the training data set and 0.85 for the test data set across all algorithms.…”

Section: Resultsmentioning

confidence: 99%

Deep Learning for Automated Image Segmentation of the Middle Ear: A Scoping Review

Ross,

Tanna,

Lilaonitkul

et al. 2024

Otolaryngol.--head neck surg.

View full text Add to dashboard Cite

ObjectiveConvolutional neural networks (CNNs) have revolutionized medical image segmentation in recent years. This scoping review aimed to carry out a comprehensive review of the literature describing automated image segmentation of the middle ear using CNNs from computed tomography (CT) scans.Data SourcesA comprehensive literature search, generated jointly with a medical librarian, was performed on Medline, Embase, Scopus, Web of Science, and Cochrane, using Medical Subject Heading terms and keywords. Databases were searched from inception to July 2023. Reference lists of included papers were also screened.Review MethodsTen studies were included for analysis, which contained a total of 866 scans which were used in model training/testing. Thirteen different architectures were described to perform automated segmentation. The best Dice similarity coefficient (DSC) for the entire ossicular chain was 0.87 using ResNet. The highest DSC for any structure was the incus using 3D‐V‐Net at 0.93. The most difficult structure to segment was the stapes, with the highest DSC of 0.84 using 3D‐V‐Net.ConclusionsNumerous architectures have demonstrated good performance in segmenting the middle ear using CNNs. To overcome some of the difficulties in segmenting the stapes, we recommend the development of an architecture trained on cone beam CTs to provide improved spatial resolution to assist with delineating the smallest ossicle.Implications for PracticeThis has clinical applications for preoperative planning, diagnosis, and simulation.

show abstract

“…To prepare training data sets, synapses were manually demarcated as the presynaptic membrane, rigid synaptic cleft, and postsynaptic membrane. Networks that use slice-to-slice context have been shown to have higher accuracy 34 , therefore we chose to train 2.5D and 3D architectures to detect features using several sections of context, as synapses and vesicle clouds can be tracked through approximately 3-10+ sections in our datasets (Figure 1A-C, details are described in Methods).…”

Section: Resultsmentioning

confidence: 99%

A multi-faceted analysis of synapses reveals the role of neuroligin-1 cleavage in presynaptic vesicle accumulation in the lateral amygdala

Thomas,

Anderson,

Alexis

et al. 2023

Preprint

View full text Add to dashboard Cite

Neuroligin-1 (NLGN1) is a cell adhesion molecule found at excitatory glutamatergic synapses in the brain which regulates synaptic function and maturation. Extracellular cleavage of NLGN1 by proteases has been shown to control vesicle release in cultured neurons, but nothing is known about the underlying changes to synapse structure that accompany this, or how synapse function is affected in brain tissue. We found that prevention of NLGN1 cleavage through mutation to the extracellular stalk domain increases synaptic vesicle docking and miniature excitatory post-synaptic current frequency at synapses of the lateral amygdala. Using a novel volume electron microscopy (vEM) analysis pipeline based on deep learning extraction of thousands of synapses and vesicles clouds and subsequent spatial analyses, we found that the total pool of synaptic vesicles shifts closer to the synapse in mutants. Furthermore, we observed an increased frequency of incomplete synapses that lack vesicle accumulation, pointing towards disruption of synaptic pruning and accumulation of putatively non-functioning synapses. Our study provides evidence of a structural and functional role of NLGN1 cleavage in native brain tissue, and establishes a foundation for vEM analysis of synapse-vesicle spatial relationships in other animal models of dysfunction and disease.

show abstract

Comparing 3D, 2.5D, and 2D Approaches to Brain Image Auto-Segmentation

Cited by 23 publications

References 36 publications

3D Capsule Networks for Brain Image Segmentation

3D Capsule Networks for Brain Image Segmentation

Deep Learning for Automated Image Segmentation of the Middle Ear: A Scoping Review

A multi-faceted analysis of synapses reveals the role of neuroligin-1 cleavage in presynaptic vesicle accumulation in the lateral amygdala

Contact Info

Product

Resources

About