Dense Multi-path U-Net for Ischemic Stroke Lesion Segmentation in Multiple Image Modalities

APIS: a paired CT-MRI dataset for ischemic stroke segmentation - methods and challenges

Stroke, the second leading cause of mortality globally, predominantly results from ischemic conditions. Immediate attention and diagnosis, related to the characterization of brain lesions, play a crucial role in patient prognosis. Standard stroke protocols include an initial evaluation from a non-contrast CT to discriminate between hemorrhage and ischemia. However, non-contrast CTs lack sensitivity in detecting subtle ischemic changes in this phase. Alternatively, diffusion-weighted MRI studies provide enhanced capabilities, yet are constrained by limited availability and higher costs. Hence, we idealize new approaches that integrate ADC stroke lesion findings into CT, to enhance the analysis and accelerate stroke patient management. This study details a public challenge where scientists applied top computational strategies to delineate stroke lesions on CT scans, utilizing paired ADC information. Also, it constitutes the first effort to build a paired dataset with NCCT and ADC studies of acute ischemic stroke patients. Submitted algorithms were validated with respect to the references of two expert radiologists. The best achieved Dice score was 0.2 over a test study with 36 patient studies. Despite all the teams employing specialized deep learning tools, results reveal limitations of computational approaches to support the segmentation of small lesions with heterogeneous density.

Hybrid Contextual Semantic Network for Accurate Segmentation and Detection of Small-Size Stroke Lesions From MRI

Stroke is a cerebrovascular disease with high mortality and disability rates. The occurrence of the stroke typically produces lesions of different sizes, with the accurate segmentation and detection of small-size stroke lesions being closely related to the prognosis of patients. However, the large lesions are usually correctly identified, the small-size lesions are usually ignored. This article provides a hybrid contextual semantic network (HCSNet) that can accurately and simultaneously segment and detect small-size stroke lesions from magnetic resonance images. HCSNet inherits the advantages of the encoder-decoder architecture and applies a novel hybrid contextual semantic module that generates high-quality contextual semantic features from the spatial and channel contextual semantic features through the skip connection layer. Moreover, a mixing-loss function is proposed to optimize HCSNet for unbalanced small-size lesions. HCSNet is trained and evaluated on 2D magnetic resonance images produced from the Anatomical Tracings of Lesions After Stroke challenge (ATLAS R2.0). Extensive experiments demonstrate that HCSNet outperforms several other state-of-the-art methods in its ability to segment and detect small-size stroke lesions. Visualization and ablation experiments reveal that the hybrid semantic module improves the segmentation and detection performance of HCSNet.

DXM-TransFuse U-net: Dual cross-modal transformer fusion U-net for automated nerve identification

Accurate nerve identification is critical during surgical procedures to prevent damage to nerve tissues. Nerve injury can cause long-term adverse effects for patients, as well as financial overburden. Birefringence imaging is a noninvasive technique derived from polarized images that have successfully identified nerves that can assist during intraoperative surgery. Furthermore, birefringence images can be processed under 20 ms with a GPGPU implementation, making it a viable image modality option for real-time processing. In this study, we first comprehensively investigate the usage of birefringence images combined with deep learning, which can automatically detect nerves with gains upwards of 14% over its color image-based (RGB) counterparts on the F2 score. Additionally, we develop a deep learning network framework using the U-Net architecture with a Transformer based fusion module at the bottleneck that leverages both birefringence and RGB modalities. The dual-modality framework achieves 76.12 on the F2 score, a gain of 19.6 % over single-modality networks using only RGB images. By leveraging and extracting the feature maps of each modality independently and using each modality's information for cross-modal interactions, we aim to provide a solution that would further increase the effectiveness of imaging systems for enabling noninvasive intraoperative nerve identification.

Cross-Attention and Deep Supervision UNet for Lesion Segmentation of Chronic Stroke

Stroke is an acute cerebrovascular disease with high incidence, high mortality, and high disability rate. Determining the location and volume of the disease in MR images promotes accurate stroke diagnosis and surgical planning. Therefore, the automatic recognition and segmentation of stroke lesions has important clinical significance for large-scale stroke imaging analysis. There are some problems in the segmentation of stroke lesions, such as imbalance of the front and back scenes, uncertainty of position, and unclear boundary. To meet this challenge, this paper proposes a cross-attention and deep supervision UNet (CADS-UNet) to segment chronic stroke lesions from T1-weighted MR images. Specifically, we propose a cross-spatial attention module, which is different from the usual self-attention module. The location information interactively selects encode features and decode features to enrich the lost spatial focus. At the same time, the channel attention mechanism is used to screen the channel characteristics. Finally, combined with deep supervision and mixed loss, the model is supervised more accurately. We compared and verified the model on the authoritative open dataset "Anatomical Tracings of Lesions After Stroke" (Atlas), which fully proved the effectiveness of our model.

Automated Extraction of Cerebral Infarction Region in Head MR Image Using Pseudo Cerebral Infarction Image by CycleGAN

Since recognizing the location and extent of infarction is essential for diagnosis and treatment, many methods using deep learning have been reported. Generally, deep learning requires a large amount of training data. To overcome this problem, we generated pseudo patient images using CycleGAN, which performed image transformation without paired images. Then, we aimed to improve the extraction accuracy by using the generated images for the extraction of cerebral infarction regions. First, we used CycleGAN for data augmentation. Pseudo-cerebral infarction images were generated from healthy images using CycleGAN. Finally, U-Net was used to segment the cerebral infarction region using CycleGAN-generated images. Regarding the extraction accuracy, the Dice index was 0.553 for U-Net with CycleGAN, which was an improvement over U-Net without CycleGAN. Furthermore, the number of false positives per case was 3.75 for U-Net without CycleGAN and 1.23 for U-Net with CycleGAN, respectively. The number of false positives was reduced by approximately 67% by introducing the CycleGAN-generated images to training cases. These results indicate that utilizing CycleGAN-generated images was effective and facilitated the accurate extraction of the infarcted regions while maintaining the detection rate.

HD-RDS-UNet: Leveraging Spatial-Temporal Correlation between the Decoder Feature Maps for Lymphoma Segmentation

Lymphoma is a group of malignant tumors originated in the lymphatic system. Automatic and accurate lymphoma segmentation in PET/CT volumes is critical yet challenging in the clinical practice. Recently, UNet-like architectures are widely used for medical image segmentation. The pure UNet-like architectures model the spatial correlation between the feature maps very well, whereas they discard the critical temporal correlation. Some prior work combines UNet with recurrent neural networks (RNNs) to utilize the spatial and temporal correlation simultaneously. However, it is inconvenient to incorporate some advanced techniques for UNet to RNNs, which hampers their further improvements. In this paper, we propose a recurrent dense siamese decoder architecture, which simulates RNNs and can densely utilize the spatial-temporal correlation between the decoder feature maps following a UNet approach. We combine it with a modified hyper dense encoder. Therefore, the proposed model is a UNet with a hyper dense encoder and a recurrent dense siamese decoder (HD-RDS-UNet). To stabilize the training process, we propose a weighted Dice loss with stable gradient and self-adaptive parameters. We perform patient-independent fivefold cross-validation on 3D volumes collected from whole-body PET/CT scans of patients with lymphomas. The experimental results show that the volume-wise average Dice score and sensitivity are 85.58% and 94.63%, respectively. The patient-wise average Dice score and sensitivity are 85.85% and 95.01%, respectively. The different configurations of HD-RDS-UNet consistently show superiority in the performance comparison. Besides, a trained HD-RDS-UNet can be easily pruned, resulting in significantly reduced inference time and memory usage, while keeping very good segmentation performance.

"Keep it simple, scholar": an experimental analysis of few-parameter segmentation networks for retinal vessels in fundus imaging

PURPOSE

With the recent development of deep learning technologies, various neural networks have been proposed for fundus retinal vessel segmentation. Among them, the U-Net is regarded as one of the most successful architectures. In this work, we start with simplification of the U-Net, and explore the performance of few-parameter networks on this task.

METHODS

We firstly modify the model with popular functional blocks and additional resolution levels, then we switch to exploring the limits for compression of the network architecture. Experiments are designed to simplify the network structure, decrease the number of trainable parameters, and reduce the amount of training data. Performance evaluation is carried out on four public databases, namely DRIVE, STARE, HRF and CHASE_DB1. In addition, the generalization ability of the few-parameter networks are compared against the state-of-the-art segmentation network.

RESULTS

We demonstrate that the additive variants do not significantly improve the segmentation performance. The performance of the models are not severely harmed unless they are harshly degenerated: one level, or one filter in the input convolutional layer, or trained with one image. We also demonstrate that few-parameter networks have strong generalization ability.

CONCLUSION

It is counter-intuitive that the U-Net produces reasonably good segmentation predictions until reaching the mentioned limits. Our work has two main contributions. On the one hand, the importance of different elements of the U-Net is evaluated, and the minimal U-Net which is capable of the task is presented. On the other hand, our work demonstrates that retinal vessel segmentation can be tackled by surprisingly simple configurations of U-Net reaching almost state-of-the-art performance. We also show that the simple configurations have better generalization ability than state-of-the-art models with high model complexity. These observations seem to be in contradiction to the current trend of continued increase in model complexity and capacity for the task under consideration.

Neuroimaging and deep learning for brain stroke detection - A review of recent advancements and future prospects

BACKGROUND AND OBJECTIVE

In recent years, deep learning algorithms have created a massive impact on addressing research challenges in different domains. The medical field also greatly benefits from the use of improving deep learning models which save time and produce accurate results. This research aims to emphasize the impact of deep learning models in brain stroke detection and lesion segmentation. This is achieved by discussing the state of the art approaches proposed by the recent works in this field.

METHODS

In this study, the advancements in stroke lesion detection and segmentation were focused. The survey analyses 113 research papers published in different academic research databases. The research articles have been filtered out based on specific criteria to obtain the most prominent insights related to stroke lesion detection and segmentation.

RESULTS

The features of the stroke lesion vary based on the type of imaging modality. To develop an effective method for stroke lesion detection, the features need to be carefully extracted from the input images. This review takes an attempt to categorize and discuss the different deep architectures employed for stroke lesion detection and segmentation, based on the underlying imaging modality. This further assists in understanding the relevance of the two-deep neural network components in medical image analysis namely Convolutional Neural Network (CNN) and Fully Convolutional Network (FCN). It hints at other possible deep architectures that can be proposed for better results towards stroke lesion detection. Also, the emerging trends and breakthroughs in stroke detection have been detailed in this evaluation.

CONCLUSION

This work concludes by examining the technical and non-technical challenges faced by researchers and indicate the future implications in stroke detection. It could support the bio-medical researchers to propose better solutions for stroke lesion detection.