Measurement-oriented deep-learning workflow for improved segmentation of myelin and axons in high-resolution images of human cerebral white matter

Janjic, Predrag; Petrovski, Kristijan; Dolgoski, Blagoja; Smiley, John F.; Zdravkovski, Panche; Pavlovski, Goran; Jakjovski, Zlatko; Davceva, Natasha; Poposka, Verica; Stankov, Aleksandar; Rosoklija, Gorazd; Petrushevska, Gordana; Kocarev, Ljupčo; Dwork, Andrew J.

doi:10.1016/j.jneumeth.2019.108373

“…These tools are useful for processing large datasets but require image pre-processing, adjustment of parameters in MATLAB, or additional training of the neural network using manually labeled images from user-specific datasets, which can be hurdles for implementation. Janjic et al developed a high-fidelity tool using deep neural networks, which suits researchers with a computer science background seeking an advanced g-ratio quantification tool (Janjic et al, 2019). While there are other useful tools available, MyelTracer's intuitive GUI makes it a powerful tool allowing researchers to quantify myelin without computer science knowledge.…”

Section: Discussionmentioning

confidence: 99%

“…Several non-automated, semi-automated, and fully automated toolboxes are currently available. They include G-ratio for ImageJ, AxonSeg, and AxonDeepSeg (Goebbels et al, 2010;More et al, 2011;Bégin et al, 2014;Zaimi et al, 2016Zaimi et al, , 2018Janjic et al, 2019). Collectively, these tools have facilitated different quantitative measurements of myelin ultrastructure, such as the number of axons and g-ratio.…”

Section: Introductionmentioning

confidence: 99%

MyelTracer: A Semi-Automated Software for Myeling-Ratio Quantification

Kaiser

¹

,

Allen

²

,

Kwon

³

et al. 2021

eNeuro

View full text Add to dashboard Cite

show abstract

“…Their algorithm was trained using these light micrographs but was not tested outside of their initial database. Janjic et al [ 15 ] developed a segmentation program using electron micrographs of human prefrontal cortex. While both of these programs showed excellent agreement when compared to manual analysis of micrographs from the same dataset, the robustness of the algorithms was not tested.…”

Section: Discussionmentioning

confidence: 99%

“…Deep learning, also known as computer vision, is a subset of machine learning that can use convolutional neural networks, where each network hierarchically defines specific features of images and does not require structured numerical input data [13,14]. Various programs have leveraged deep learning for nerve analysis, but their generalizability has not been investigated for implementation outside of the research groups in which they were developed [8][9][10][11][14][15][16] and this has limited their widespread use.…”

Section: Introductionmentioning

confidence: 99%

A simple and robust method for automating analysis of naïve and regenerating peripheral nerves

Wong

¹

,

Hricz

²

,

Malapati

³

et al. 2021

View full text Add to dashboard Cite

Background Manual axon histomorphometry (AH) is time- and resource-intensive, which has inspired many attempts at automation. However, there has been little investigation on implementation of automated programs for widespread use. Ideally such a program should be able to perform AH across imaging modalities and nerve states. AxonDeepSeg (ADS) is an open source deep learning program that has previously been validated in electron microscopy. We evaluated the robustness of ADS for peripheral nerve axonal histomorphometry in light micrographs prepared using two different methods. Methods Axon histomorphometry using ADS and manual analysis (gold-standard) was performed on light micrographs of naïve or regenerating rat median nerve cross-sections prepared with either toluidine-resin or osmium-paraffin embedding protocols. The parameters of interest included axon count, axon diameter, myelin thickness, and g-ratio. Results Manual and automatic ADS axon counts demonstrated good agreement in naïve nerves and moderate agreement on regenerating nerves. There were small but consistent differences in measured axon diameter, myelin thickness and g-ratio; however, absolute differences were small. Both methods appropriately identified differences between naïve and regenerating nerves. ADS was faster than manual axon analysis. Conclusions Without any algorithm retraining, ADS was able to appropriately identify critical differences between naïve and regenerating nerves and work with different sample preparation methods of peripheral nerve light micrographs. While there were differences between absolute values between manual and ADS, ADS performed consistently and required much less time. ADS is an accessible and robust tool for AH that can provide consistent analysis across protocols and nerve states.

show abstract

“…Deep learning, also known as computer vision, is a subset of machine learning that can use convolutional neural networks, where each network hierarchically defines specific features of images and does not require structured numerical input data [9,10]. Various programs have leveraged deep learning for nerve analysis, but their generalizability has not been investigated for implementation outside of the research groups in which they were developed [4][5][6][7][10][11][12]. AxonDeepSeg (ADS) is a novel deep learning program for fully-automated AH originally developed using both scanning-and transmission-electron micrographs of the brain and spinal cord [7].…”

Section: Introductionmentioning

confidence: 99%

A simple and robust method for automating analysis of naïve and regenerating peripheral nerves

Wong

¹

,

Hricz

²

,

Malapati

³

et al. 2021

Preprint

2

0

View full text Add to dashboard Cite

Background Manual axon histomorphometry (AH) is time- and resource-intensive, which has inspired many attempts at automation. However, there has been little investigation on implementation of automated programs for widespread use. Ideally such a program should be able to perform AH across imaging modalities and nerve states. AxonDeepSeg (ADS) is an open source deep learning program that has previously been validated in electron microscopy. We evaluated the robustness of ADS for peripheral nerve axonal histomorphometry in light micrographs prepared using two different methods. Methods Axon histomorphometry using ADS and manual analysis (gold-standard) was performed on light micrographs of naïve or regenerating rat median nerve cross-sections prepared with either toluidine-resin or osmium-paraffin embedding protocols. The parameters of interest included axon count, axon diameter, myelin thickness, and g-ratio. Results Manual and automatic ADS axon counts demonstrated good agreement in naïve nerves and moderate agreement on regenerating nerves. There were small but consistent differences in measured axon diameter, myelin thickness and g-ratio; however, absolute differences were small. Both methods appropriately identified differences between naïve and regenerating nerves. ADS was faster than manual axon analysis. Conclusions Without any algorithm retraining, ADS was able to appropriately identify critical differences between naïve and regenerating nerves and work with different sample preparation methods of peripheral nerve light micrographs. While there were differences between absolute values between manual and ADS, ADS performed consistently and required much less time. ADS is an accessible and robust tool for AH that can provide consistent analysis across protocols and nerve states.

show abstract

Measurement-oriented deep-learning workflow for improved segmentation of myelin and axons in high-resolution images of human cerebral white matter

Cited by 13 publications

References 24 publications

MyelTracer: A Semi-Automated Software for Myeling-Ratio Quantification

MyelTracer: A Semi-Automated Software for Myeling-Ratio Quantification

A simple and robust method for automating analysis of naïve and regenerating peripheral nerves

A simple and robust method for automating analysis of naïve and regenerating peripheral nerves

Contact Info

Product

Resources

About