DSCA: A Digital Subtraction Angiography Sequence Dataset and Spatio-Temporal Model for Cerebral Artery Segmentation

TL;DR

This work tackles automated cerebral artery segmentation in Digital Subtraction Angiography by introducing the DSCA dataset and DSANet, a spatio-temporal network that leverages both MinIP imagery and multi-frame sequences. DSANet incorporates a dedicated Temporal Encoding Branch, TemporalFormer blocks for global frame-level context, and a Spatio-Temporal Fusion module to integrate spatial and temporal cues, achieving a Dice score of 0.9033 on CA segmentation. The dataset comprises 224 DSA sequences across ICA, ECA, and VA from 58 patients, with pixel-wise MinIP annotations distinguishing BV and MAT, enabling clinically meaningful analysis. The results show DSANet outperforms state-of-the-art methods, particularly in small-vessel connectivity and MAT discrimination, and the public DSCA dataset is poised to accelerate clinical cerebrovascular analysis and broader DSA research.

Abstract

Cerebrovascular diseases (CVDs) remain a leading cause of global disability and mortality. Digital Subtraction Angiography (DSA) sequences, recognized as the gold standard for diagnosing CVDs, can clearly visualize the dynamic flow and reveal pathological conditions within the cerebrovasculature. Therefore, precise segmentation of cerebral arteries (CAs) and classification between their main trunks and branches are crucial for physicians to accurately quantify diseases. However, achieving accurate CA segmentation in DSA sequences remains a challenging task due to small vessels with low contrast, and ambiguity between vessels and residual skull structures. Moreover, the lack of publicly available datasets limits exploration in the field. In this paper, we introduce a DSA Sequence-based Cerebral Artery segmentation dataset (DSCA), the publicly accessible dataset designed specifically for pixel-level semantic segmentation of CAs. Additionally, we propose DSANet, a spatio-temporal network for CA segmentation in DSA sequences. Unlike existing DSA segmentation methods that focus only on a single frame, the proposed DSANet introduces a separate temporal encoding branch to capture dynamic vessel details across multiple frames. To enhance small vessel segmentation and improve vessel connectivity, we design a novel TemporalFormer module to capture global context and correlations among sequential frames. Furthermore, we develop a Spatio-Temporal Fusion (STF) module to effectively integrate spatial and temporal features from the encoder. Extensive experiments demonstrate that DSANet outperforms other state-of-the-art methods in CA segmentation, achieving a Dice of 0.9033.

Figures (11)

• Figure 1: (a) is a DSA sequence, which includes 17 frames in the arterial phase. (b) is the MinIP image of the arterial phase of the DSA sequence.
• Figure 2: The annotating examples of the DSCA dataset. The first to third rows are ICA, ECA, and VA. The 1st and 2nd columns are antero-posterior view, and the 3rd and 4th columns are lateral view.
• Figure 3: Architecture of our proposed DSANet. It is primarily composed of two encoding branches, one decoding branch, and two modules: TemporalFormer (TF), Spatio-Temporal Fusion (STF). The details of the dimension transformation process of TF are shown in Fig. 4
• Figure 4: The details of the dimension transformation process in TemporalFormer, where c is 512.
• Figure 5: Segmentation results on the DSCA dataset. Green is MAT, and red is BV. Enlarged viewing for better clarity.