Automatic Temporal Segmentation of Vessels of the Brain Using 4D ASL MRA Images

CAVE: Cerebral artery-vein segmentation in digital subtraction angiography

Cerebral X-ray digital subtraction angiography (DSA) is a widely used imaging technique in patients with neurovascular disease, allowing for vessel and flow visualization with high spatio-temporal resolution. Automatic artery-vein segmentation in DSA plays a fundamental role in vascular analysis with quantitative biomarker extraction, facilitating a wide range of clinical applications. The widely adopted U-Net applied on static DSA frames often struggles with disentangling vessels from subtraction artifacts. Further, it falls short in effectively separating arteries and veins as it disregards the temporal perspectives inherent in DSA. To address these limitations, we propose to simultaneously leverage spatial vasculature and temporal cerebral flow characteristics to segment arteries and veins in DSA. The proposed network, coined CAVE, encodes a 2D+time DSA series using spatial modules, aggregates all the features using temporal modules, and decodes it into 2D segmentation maps. On a large multi-center clinical dataset, CAVE achieves a vessel segmentation Dice of 0.84 (±0.04) and an artery-vein segmentation Dice of 0.79 (±0.06). CAVE surpasses traditional Frangi-based k-means clustering (P < 0.001) and U-Net (P < 0.001) by a significant margin, demonstrating the advantages of harvesting spatio-temporal features. This study represents the first investigation into automatic artery-vein segmentation in DSA using deep learning. The code is publicly available at https://github.com/RuishengSu/CAVE_DSA.

Accurate and robust segmentation of cerebral vasculature on four-dimensional arterial spin labeling magnetic resonance angiography using machine-learning approach

Segmentation of cerebral vasculature on MR vascular images is of great significance for clinical application and research. However, the existing cerebrovascular segmentation approaches are limited due to insufficient image contrast and complicated algorithms. This study aims to explore the potential of the emerging four-dimensional arterial spin labeling magnetic resonance angiography (4D ASL-MRA) technique for fast and accurate cerebrovascular segmentation with a simple machine-learning approach. Nine temporal features were extracted from the intensity-time signal of each voxel, and eight spatial features from the neighboring voxels. Then, the unsupervised outlier detection algorithm, i.e. Isolation Forest, is used for segmentation of the vascular voxels based on the extracted features. The total length of the centerlines of the intracranial arterial vasculature, the dice similarity coefficient (DSC), and the average Hausdorff Distance (AVGHD) on the cross-sections of small- to large-sized vessels were calculated to evaluate the performance of the segmentation approach on 4D ASL-MRA of 18 subjects. Experiments show that the temporal information on 4D ASL-MRA can largely improve the segmentation performance. In addition, the proposed segmentation approach outperforms the traditional methods that were performed on the 3D image (i.e. the temporal average intensity projection of 4D ASL-MRA) and the previously proposed frame-wise approach. In conclusion, this study demonstrates that accurate and robust segmentation of cerebral vasculature is achievable on 4D ASL-MRA by using a simple machine-learning approach with appropriate features.

Visual Prompting Based Incremental Learning for Semantic Segmentation of Multiplex Immuno-Flourescence Microscopy Imagery

Deep learning approaches are state-of-the-art for semantic segmentation of medical images, but unlike many deep learning applications, medical segmentation is characterized by small amounts of annotated training data. Thus, while mainstream deep learning approaches focus on performance in domains with large training sets, researchers in the medical imaging field must apply new methods in creative ways to meet the more constrained requirements of medical datasets. We propose a framework for incrementally fine-tuning a multi-class segmentation of a high-resolution multiplex (multi-channel) immuno-flourescence image of a rat brain section, using a minimal amount of labelling from a human expert. Our framework begins with a modified Swin-UNet architecture that treats each biomarker in the multiplex image separately and learns an initial "global" segmentation (pre-training). This is followed by incremental learning and refinement of each class using a very limited amount of additional labeled data provided by a human expert for each region and its surroundings. This incremental learning utilizes the multi-class weights as an initialization and uses the additional labels to steer the network and optimize it for each region in the image. In this way, an expert can identify errors in the multi-class segmentation and rapidly correct them by supplying the model with additional annotations hand-picked from the region. In addition to increasing the speed of annotation and reducing the amount of labelling, we show that our proposed method outperforms a traditional multi-class segmentation by a large margin.

A Machine Learning Method for Automated In Vivo Transparent Vessel Segmentation and Identification Based on Blood Flow Characteristics

In vivo transparent vessel segmentation is important to life science research. However, this task remains very challenging because of the fuzzy edges and the barely noticeable tubular characteristics of vessels under a light microscope. In this paper, we present a new machine learning method based on blood flow characteristics to segment the global vascular structure in vivo. Specifically, the videos of blood flow in transparent vessels are used as input. We use the machine learning classifier to classify the vessel pixels through the motion features extracted from moving red blood cells and achieve vessel segmentation based on a region-growing algorithm. Moreover, we utilize the moving characteristics of blood flow to distinguish between the types of vessels, including arteries, veins, and capillaries. In the experiments, we evaluate the performance of our method on videos of zebrafish embryos. The experimental results indicate the high accuracy of vessel segmentation, with an average accuracy of 97.98%, which is much more superior than other segmentation or motion-detection algorithms. Our method has good robustness when applied to input videos with various time resolutions, with a minimum of 3.125 fps.