Anatomical Modeling of Brain Vasculature in Two-Photon Microscopy by Generalizable Deep Learning

Evaluating segment anything model (SAM) on MRI scans of brain tumors

Addressing the challenge of automatically segmenting anatomical structures from brain images has been a long-standing problem, attributed to subject- and image-based variations and constraints in available data annotations. The Segment Anything Model (SAM), developed by Meta, is a foundational model trained to provide zero-shot segmentation outputs with or without interactive user inputs, demonstrating notable performance on various objects and image domains without explicit prior training. This study evaluated SAM's performance in brain tumor segmentation using two publicly available Magnetic Resonance Imaging (MRI) datasets. The study analyzed SAM's standalone segmentation as well as its performance when provided user interaction through point prompts and bounding box inputs. SAM exhibited versatility across configurations and datasets, with the bounding box consistently outperforming others in achieving superior localized precision, with average Dice scores of 0.68 for TCGA and 0.56 for BRATS, along with average IoU values of 0.89 and 0.65, respectively, especially for tumors with low-to-medium curvature. Inconsistencies were observed, particularly in relation to variations in tumor size, shape, and textural features. The conclusion drawn from the study is that while SAM can automate medical image segmentation, further training and careful implementation are necessary for diagnostic purposes, especially with challenging cases such as MRI scans of brain tumors.

Microvascular plasticity in mouse stroke model recovery: Anatomy statistics, dynamics measured by longitudinal in vivo two-photon angiography, network vectorization

This manuscript quantitatively investigates remodeling dynamics of the cortical microvascular network (thousands of connected capillaries) following photothrombotic ischemia (cubic millimeter volume, imaged weekly) using a novel in vivo two-photon angiography and high throughput vascular vectorization method. The results suggest distinct temporal patterns of cerebrovascular plasticity, with acute remodeling peaking at one week post-stroke. The network architecture then gradually stabilizes, returning to a new steady state after four weeks. These findings align with previous literature on neuronal plasticity, highlighting the correlation between neuronal and neurovascular remodeling. Quantitative analysis of neurovascular networks using length- and strand-based statistical measures reveals intricate changes in network anatomy and topology. The distance and strand-length statistics show significant alterations, with a peak of plasticity observed at one week post-stroke, followed by a gradual return to baseline. The orientation statistic plasticity peaks at two weeks, gradually approaching the (conserved across subjects) stroke signature. The underlying mechanism of the vascular response (angiogenesis vs. tissue deformation), however, is yet unexplored. Overall, the combination of chronic two-photon angiography, vascular vectorization, reconstruction/visualization, and statistical analysis enables both qualitative and quantitative assessments of neurovascular remodeling dynamics, demonstrating a method for investigating cortical microvascular network disorders and the therapeutic modes of action thereof.

Deep learning-based vessel extraction in 3D confocal microscope images of cleared human glioma tissues

Comprehensive visualization and accurate extraction of tumor vasculature are essential to study the nature of glioma. Nowadays, tissue clearing technology enables 3D visualization of human glioma vasculature at micron resolution, but current vessel extraction schemes cannot well cope with the extraction of complex tumor vessels with high disruption and irregularity under realistic conditions. Here, we developed a framework, FineVess, based on deep learning to automatically extract glioma vessels in confocal microscope images of cleared human tumor tissues. In the framework, a customized deep learning network, named 3D ResCBAM nnU-Net, was designed to segment the vessels, and a novel pipeline based on preprocessing and post-processing was developed to refine the segmentation results automatically. On the basis of its application to a practical dataset, we showed that the FineVess enabled extraction of variable and incomplete vessels with high accuracy in challenging 3D images, better than other traditional and state-of-the-art schemes. For the extracted vessels, we calculated vascular morphological features including fractal dimension and vascular wall integrity of different tumor grades, and verified the vascular heterogeneity through quantitative analysis.

Accurate and robust auto-segmentation of head and neck organ-at-risks based on a novel CNN fine-tuning workflow

PURPOSE

Obvious inconsistencies in auto-segmentations exist among various AI software. In this study, we have developed a novel convolutional neural network (CNN) fine-tuning workflow to achieve precise and robust localized segmentation.

METHODS

The datasets include Hubei Cancer Hospital dataset, Cetuximab Head and Neck Public Dataset, and Québec Public Dataset. Seven organs-at-risks (OARs), including brain stem, left parotid gland, esophagus, left optic nerve, optic chiasm, mandible, and pharyngeal constrictor, were selected. The auto-segmentation results from four commercial AI software were first compared with the manual delineations. Then a new multi-scale lightweight residual CNN model with an attention module (named as HN-Net) was trained and tested on 40 samples and 10 samples from Hubei Cancer Hospital, respectively. To enhance the network's accuracy and generalization ability, the fine-tuning workflow utilized an uncertainty estimation method for automatic selection of candidate samples of worthiness from Cetuximab Head and Neck Public Dataset for further training. The segmentation performances were evaluated on the Hubei Cancer Hospital dataset and/or the entire Québec Public Dataset.

RESULTS

A maximum difference of 0.13 and 0.7 mm in average Dice value and Hausdorff distance value for the seven OARs were observed by four AI software. The proposed HN-Net achieved an average Dice value of 0.14 higher than that of the AI software, and it also outperformed other popular CNN models (HN-Net: 0.79, U-Net: 0.78, U-Net++: 0.78, U-Net-Multi-scale: 0.77, AI software: 0.65). Additionally, the HN-Net fine-tuning workflow by using the local datasets and external public datasets further improved the automatic segmentation with the average Dice value by 0.02.

CONCLUSION

The delineations of commercial AI software need to be carefully reviewed, and localized further training is necessary for clinical practice. The proposed fine-tuning workflow could be feasibly adopted to implement an accurate and robust auto-segmentation model by using local datasets and external public datasets.

Measurements of cerebral microvascular blood flow, oxygenation, and morphology in a mouse model of whole-brain irradiation-induced cognitive impairment by two-photon microscopy and optical coherence tomography: evidence for microvascular injury in the cerebral white matter

Whole-brain irradiation (WBI, also known as whole-brain radiation therapy) is a mainstay treatment modality for patients with multiple brain metastases. It is also used as a prophylactic treatment for microscopic tumors that cannot be detected by magnetic resonance imaging. WBI induces a progressive cognitive decline in ~ 50% of the patients surviving over 6 months, significantly compromising the quality of life. There is increasing preclinical evidence that radiation-induced injury to the cerebral microvasculature and accelerated neurovascular senescence plays a central role in this side effect of WBI. To better understand this side effect, male C57BL/6 mice were first subjected to a clinically relevant protocol of fractionated WBI (5 Gy, two doses per week, for 4 weeks). Nine months post the WBI treatment, we applied two-photon microscopy and Doppler optical coherence tomography to measure capillary red-blood-cell (RBC) flux, capillary morphology, and microvascular oxygen partial pressure (PO2) in the cerebral somatosensory cortex in the awake, head-restrained, WPI-treated mice and their age-matched controls, through a cover-glass-sealed chronic cranial window. Thanks to the extended penetration depth with the fluorophore - Alexa680, measurements of capillary blood flow properties (e.g., RBC flux, speed, and linear density) in the cerebral subcortical white matter were enabled. We found that the WBI-treated mice exhibited a significantly decreased capillary RBC flux in the white matter. WBI also caused a significant reduction in capillary diameter, as well as a large (although insignificant) reduction in segment density at the deeper cortical layers (e.g., 600-700 μm), while the other morphological properties (e.g., segment length and tortuosity) were not obviously affected. In addition, we found that PO2 measured in the arterioles and venules, as well as the calculated oxygen saturation and oxygen extraction fraction, were not obviously affected by WBI. Lastly, WBI was associated with a significant increase in the erythrocyte-associated transients of PO2, while the changes of other cerebral capillary PO2 properties (e.g., capillary mean-PO2, RBC-PO2, and InterRBC-PO2) were not significant. Collectively, our findings support the notion that WBI results in persistent cerebral white matter microvascular impairment, which likely contributes to the WBI-induced brain injury and cognitive decline. Further studies are warranted to assess the WBI-induced changes in brain tissue oxygenation and malfunction of the white matter microvasculature as well.

All answers are in the images: A review of deep learning for cerebrovascular segmentation

Cerebrovascular imaging is a common examination. Its accurate cerebrovascular segmentation become an important auxiliary method for the diagnosis and treatment of cerebrovascular diseases, which has received extensive attention from researchers. Deep learning is a heuristic method that encourages researchers to derive answers from the images by driving datasets. With the continuous development of datasets and deep learning theory, it has achieved important success for cerebrovascular segmentation. Detailed survey is an important reference for researchers. To comprehensively analyze the newest cerebrovascular segmentation, we have organized and discussed researches centered on deep learning. This survey comprehensively reviews deep learning for cerebrovascular segmentation since 2015, it mainly includes sliding window based models, U-Net based models, other CNNs based models, small-sample based models, semi-supervised or unsupervised models, fusion based models, Transformer based models, and graphics based models. We organize the structures, improvement, and important parameters of these models, as well as analyze development trends and quantitative assessment. Finally, we have discussed the challenges and opportunities of possible research directions, hoping that our survey can provide researchers with convenient reference.

Three-dimensional microvascular network reconstruction from in vivo images with adaptation of the regional inhomogeneity in the signal-to-noise ratio

OBJECTIVE

Quantification of angiographic images with two-photon laser scanning fluorescence microscopy (2PLSM) relies on proper segmentation of the vascular images. However, the images contain inhomogeneities in the signal-to-noise ratio (SNR) arising from regional effects of light scattering and absorption. The present study developed a semiautomated quantification method for volume images of 2PLSM angiography by adjusting the binarization threshold according to local SNR along the vessel centerlines.

METHODS

A phantom model made with fluorescent microbeads was used to incorporate a region-dependent binarization threshold.

RESULTS

The recommended SNR for imaging was found to be 4.2-10.6 that provide the true size of imaged objects if the binarization threshold was fixed at 50% of SNR. However, angiographic images in the mouse cortex showed variable SNR up to 45 over the depths. To minimize the errors caused by variable SNR and a spatial extent of the imaged objects in an axial direction, the microvascular networks were three-dimensionally reconstructed based on the cross-sectional diameters measured along the vessel centerline from the XY-plane images with adapted binarization threshold. The arterial volume was relatively constant over depths of 0-500 µm, and the capillary volume (1.7% relative to the scanned volume) showed the larger volumes than the artery (0.8%) and vein (0.6%).

CONCLUSIONS

The present methods allow consistent segmentation of microvasculature by adapting the local inhomogeneity in the SNR, which will be useful for quantitative comparison of the microvascular networks, such as under disease conditions where SNR in the 2PLSM images varies over space and time.