Curv‐Net: Curvilinear structure segmentation network based on selective kernel and multi‐Bi‐ConvLSTM

Abstract

Purpose Accurately segmenting curvilinear structures, for example, retinal blood vessels or nerve fibers, in the medical image is essential to the clinical diagnosis of many diseases. Recently, deep learning has become a popular technology to deal with the image segmentation task, and it has obtained remarkable achievement. However, the existing methods still have many problems when segmenting the curvilinear structures in medical images, such as losing the details of curvilinear structures, producing many false‐positive segmentation results. To mitigate these problems, we propose a novel end‐to‐end curvilinear structure segmentation network called Curv‐Net. Methods Curv‐Net is an effective encoder–decoder architecture constructed based on selective kernel (SK) and multibidirectional convolutional LSTM (multi‐Bi‐ConvLSTM). To be specific, we first employ the SK module in the convolutional layer to adaptively extract the multi‐scale features of the input image, and then we design a multi‐Bi‐ConvLSTM as the skip concatenation to fuse the information learned in the same stage and propagate the feature information from the deep stages to the shallow stages, which can enable the feature captured by Curv‐Net to contain more detail information and high‐level semantic information simultaneously to improve the segmentation performance. Results The effectiveness and reliability of our proposed Curv‐Net are verified on three public datasets: two color fundus datasets (DRIVE and CHASE_DB1) and one corneal nerve fiber dataset (CCM‐2). We calculate the accuracy (ACC), sensitivity (SE), specificity (SP), Dice similarity coefficient (Dice), and area under the receiver (AUC) for the DRIVE and CHASE_DB1 datasets. The ACC, SE, SP, Dice, and AUC of the DRIVE dataset are 0.9629, 0.8175, 0.9858, 0.8352, and 0.9810, respectively. For the CHASE_DB1 dataset, the values are 0.9810, 0.8564, 0.9899, 0.8143, and 0.9832, respectively. To validate the corneal nerve fiber segmentation performance of the proposed Curv‐Net, we test it on the CCM‐2 dataset and calculate Dice, SE, and false discovery rate (FDR) metrics. The Dice, SE, and FDR achieved by Curv‐Net are 0.8114 ± 0.0062, 0.8903 ± 0.0113, and 0.2547 ± 0.0104, respectively. Conclusions Curv‐Net is evaluated on three public datasets. Extensive experimental results demonstrate that Curv‐Net outperforms the other superior curvilinear structure segmentation methods.

YoloCurvSeg: You Only Label One Noisy Skeleton for Vessel-style Curvilinear Structure Segmentation

“…For example, Ronneberger et al [41] propose U-Net, which has been widely used in numerous medical image segmentation tasks. Existing curvilinear structure segmentation works focus on well-designed network architectures by introducing multi-scale [11], [42], multi-task [6], [9], [43], or various attention mechanisms [10], [44] and well-playing morphological and topological properties by introducing GANs or morphology-/topology-preserving loss functions [13], [14]. Still, data availability and annotation quality are the main limitations of these methods.…”

“…For DRIVE, we utilize the original division of 20 training samples and 20 testing samples. For CHASEDB1, we follow the division in [2], [11], with the first 20 images serving as the training set and the remaining 8 used for testing. For OCTA500, we respectively utilize 200, 10 and 90 samples as the training, validation and testing sets.…”

“…In recent years, benefiting from the development of deep learning (DL), many DL-based segmentation algorithms for curvilinear structures have been proposed and have shown overwhelming performance compared to traditional (e.g., matched filter-based and morphological processing-based [7], [8]) methods. Most existing works are dedicated to designing sophisticated network architectures [9]- [11] or deploying strategies to preserve curvilinear structures' topology by employing generative adversarial networks (GANs) [2], [12] or topology-preserving loss functions [13], [14]. These methods are typically fully-supervised, wherein large-scale well-annotated datasets are required.…”

MRFDCNet: Multireceptive Field Dense Connection Network for Real-Time Semantic Segmentation

“…MDCCNet [23] designs a multiscale deep context convolutional network that combines multiscale features and restores object boundaries through dense connected CRF. Curv-Net [24] proposes a new U-shape network, which is composed of SK module and multi-Bi-ConvLSTM. e SK module is used to extract multiscale features, and multi-Bi-ConvLSTM is used to fuse feature information of deep and shallow stages.…”

YoloCurvSeg: You Only Label One Noisy Skeleton for Vessel-style Curvilinear Structure Segmentation

“…In recent years, benefiting from the development of deep learning (DL), many DL-based segmentation algorithms for curvilinear structures have been proposed and have shown overwhelming performance compared to traditional (e.g., matched filter-based and morphological processing-based (Nguyen et al, 2013;Singh and Srivastava, 2016)) methods. Most existing works are dedicated to designing sophisticated network architectures (Peng et al, 2021;Mou et al, 2021;He et al, 2022) and deploying strategies to preserve curvilinear structures' topology by employing generative adversarial networks (GANs) (Lin et al, 2021c;Son et al, 2019) or topologypreserving loss functions (Cheng et al, 2021a;Shit et al, 2021). These methods are typically fully-supervised, wherein largescale well-annotated datasets are required.…”