Automatic Graph-Based Modeling of Brain Microvessels Captured With Two-Photon Microscopy

Damseh, Rafat; Pouliot, Philippe; Gagnon, Louise; Sakadžić, Sava; Boas, David A.; Cheriet, Foued; Lesage, Frédéric

doi:10.1109/jbhi.2018.2884678

Cited by 33 publications

(43 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Segmentation Performance Analysis. To evaluate our segmentation approach, we first visually compared the predicted vessel segmentation from our DNN with the ground truth, a traditional Hessian matrix approach [13], and a recently developed DNN model [8] in Figure 3. With increasing depth, our prediction result maintains greater overlap of the prediction and ground truth compared to other methods, out to 606µm [ Fig.…”

Section: Resultsmentioning

confidence: 99%

“…Our method has good qualitative performance with well connected vasculature and apt segmentation for both large and small vessels, demonstrating its ability to generalize to other 2PM imaging setups without retraining. For comparison, the supervised learning method by Damseh et al [8] is unable to generalize well. (E) MIPs for longitudinal x-z cross sections, each representing 20 discrete slices along the y-axis.…”

Section: Resultsmentioning

confidence: 99%

“…In the presence of these challenges, a number of techniques have been employed for vascular segmentation, including methods based on the Hessian matrix [12,13], tracing [14], optimally oriented flux [15], and geometric flow [16]. However, in practice, these methods demonstrate limited segmentation quality [8].…”

Section: Introductionmentioning

confidence: 99%

“…In recent years, techniques based on deep learning have shown significant improvement over traditional methods for 2PM vascular segmentation [8][9][10]17]. One of the first works in this line was done by Teikari et.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Anatomical modeling of brain vasculature in two-photon microscopy by generalizable deep learning

Tahir

Kura

Zhu

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Segmentation of blood vessels from two-photon microscopy (2PM) angiograms of brains has important applications in hemodynamic analysis and disease diagnosis. Here we develop a generalizable deep-learning technique for accurate 2PM vascular segmentation of sizable regions in mouse brains acquired from multiple 2PM setups. In addition, the technique is computationally efficient, making it ideal for large-scale neurovascular analysis. Introduction: Vascular segmentation from 2PM angiograms is usually an important first step in hemodynamic modeling of brain vasculature. Existing state-of-the-art segmentation methods based on deep learning either lack the ability to generalize to data from various imaging systems, or are computationally infeasible for large-scale angiograms. In this work, we present a method which improves upon both these limitations by being generalizable to various imaging systems, and also being able to segment very large-scale angiograms. Methods: We employ a computationally efficient deep learning framework based on a semi-supervised learning strategy, whose effectiveness we demonstrate on experimentally acquired in-vivo angiograms from mouse brains of dimensions up to 808x808x702 micrometers. Results: After training on data from only one 2PM microscope, we perform vascular segmentation on data from another microscope without any network tuning. Our method demonstrates 10x faster computation in terms of voxels-segmented-persecond and 3x larger depth compared to the state-of-the-art. Conclusion: Our work provides a generalizable and computationally efficient anatomical modeling framework for the brain vasculature, which consists of deeplearning based vascular segmentation followed by graphing. It paves the way for future modeling and analysis of hemodynamic response at much greater scales that were inaccessible before.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Anatomical modeling of brain vasculature in two-photon microscopy by generalizable deep learning

Tahir

Kura

Zhu

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Researchers demonstrated individual capillary tracking in living mouse cortex over several weeks, but estimates of vascular plasticity lacked statistical significance due to the small, manually segmented, sample size (Cudmore et al, 2017). Alternatively, convolutional neural networks allow computers to learn this manual task from example (Damseh et al, 2018;Haft-Javaherian et al, 2019). However, deep learners that are trained by humans in voxel-by-voxel classification have human biases and are not intrinsically robust to input image properties such as resolution and noise level.…”

Section: Introductionmentioning

confidence: 99%

Segmentation-less, automated vascular vectorization robustly extracts neurovascular network statistics from in vivo two-photon images

Mihelic

Sikora

Hassan

et al. 2020

Preprint

View full text Add to dashboard Cite

Recent advances in two-photon microscopy (2PM) have allowed large scale imaging and analysis of cortical blood vessel networks in living mice. However, extracting a network graph and vector representations for vessels remain bottlenecks in many applications. Vascular vectorization is algorithmically difficult because blood vessels have many shapes and sizes, the samples are often unevenly illuminated, and large image volumes are required to achieve good statistical power. State-of-the-art, three-dimensional, vascular vectorization approaches require a segmented/binary image, relying on manual or supervised-machine annotation. Therefore, voxel-by-voxel image segmentation is biased by the human annotator/trainer. Furthermore, segmented images oftentimes require remedial morphological filtering before skeletonization/vectorization. To address these limitations, we propose a vectorization method to extract vascular objects directly from unsegmented images. The Segmentation-Less, Automated, Vascular Vectorization (SLAVV) source code in MATLAB is openly available on GitHub. This novel method uses simple models of vascular anatomy, efficient linear filtering, and lowcomplexity vector extraction algorithms to remove the image segmentation requirement, replacing it with manual or automated vector classification. SLAVV is demonstrated on three in vivo 2PM image volumes of microvascu- lar networks (capillaries, arterioles and venules) in the mouse cortex. Vectorization performance is proven robust to the choice of plasma-or endotheliallabeled contrast, and processing costs are shown to scale with input image volume. Fully-automated SLAVV performance is evaluated on various, simulated 2PM images based on the large, [1.4, 0.9, 0.6] mm input image, and performance metrics show greater robustness to image quality than an intensity-based thresholding approach.

show abstract

Challenges and opportunities of integrating imaging and mathematical modelling to interrogate biological processes

Berg¹,

Holroyd²,

Walsh³

et al. 2022

The International Journal of Biochemistry & Cell Biology

View full text Add to dashboard Cite

Automatic Graph-Based Modeling of Brain Microvessels Captured With Two-Photon Microscopy

Cited by 33 publications

References 43 publications

Anatomical modeling of brain vasculature in two-photon microscopy by generalizable deep learning

Anatomical modeling of brain vasculature in two-photon microscopy by generalizable deep learning

Segmentation-less, automated vascular vectorization robustly extracts neurovascular network statistics from in vivo two-photon images

Challenges and opportunities of integrating imaging and mathematical modelling to interrogate biological processes

Contact Info

Product

Resources

About