More Like this

1. Brain Hematoma Segmentation Using Active Learning and an Active Contour Model

   https://doi.org/10.1007/978-3-030-17935-9_35  

   Yao, Heming; Williamson, Craig; Gryak, Jonathan; Najarian, Kayvan (April 2019, Bioinformatics and Biomedical Engineering)

   Traumatic brain injury (TBI) is a massive public health problem worldwide. Accurate and fast automatic brain hematoma segmentation is important for TBI diagnosis, treatment and outcome prediction. In this study, we developed a fully automated system to detect and segment hematoma regions in head Computed Tomography (CT) images of patients with acute TBI. We first over-segmented brain images into superpixels and then extracted statistical and textural features to capture characteristics of superpixels. To overcome the shortage of annotated data, an uncertainty-based active learning strategy was designed to adaptively and iteratively select the most informative unlabeled data to be annotated for training a Support Vector Machine classifier (SVM). Finally, the coarse segmentation from the SVM classifier was incorporated into an active contour model to improve the accuracy of the segmentation. From our experiments, the proposed active learning strategy can achieve a comparable result with 5 times fewer labeled data compared with regular machine learning. Our proposed automatic hematoma segmentation system achieved an average Dice coefficient of 0.60 on our dataset, where patients are from multiple health centers and at multiple levels of injury. Our results show that the proposed method can effectively overcome the challenge of limited and highly varied dataset. 

   more » « less
2. Noninvasive and reliable automated detection of spreading depolarization in severe traumatic brain injury using scalp EEG

   https://doi.org/10.1038/s43856-023-00344-3  

   Chamanzar, Alireza; Elmer, Jonathan; Shutter, Lori; Hartings, Jed; Grover, Pulkit (December 2023, Communications Medicine)

   Abstract BackgroundSpreading depolarizations (SDs) are a biomarker and a potentially treatable mechanism of worsening brain injury after traumatic brain injury (TBI). Noninvasive detection of SDs could transform critical care for brain injury patients but has remained elusive. Current methods to detect SDs are based on invasive intracranial recordings with limited spatial coverage. In this study, we establish the feasibility of automated SD detection through noninvasive scalp electroencephalography (EEG) for patients with severe TBI. MethodsBuilding on our recent WAVEFRONT algorithm, we designed an automated SD detection method. This algorithm, with learnable parameters and improved velocity estimation, extracts and tracks propagating power depressions using low-density EEG. The dataset for testing our algorithm contains 700 total SDs in 12 severe TBI patients who underwent decompressive hemicraniectomy (DHC), labeled using ground-truth intracranial EEG recordings. We utilize simultaneously recorded, continuous, low-density (19 electrodes) scalp EEG signals, to quantify the detection accuracy of WAVEFRONT in terms of true positive rate (TPR), false positive rate (FPR), as well as the accuracy of estimating SD frequency. ResultsWAVEFRONT achieves the best average validation accuracy using Delta band EEG: 74% TPR with less than 1.5% FPR. Further, preliminary evidence suggests WAVEFRONT can estimate how frequently SDs may occur. ConclusionsWe establish the feasibility, and quantify the performance, of noninvasive SD detection after severe TBI using an automated algorithm. The algorithm, WAVEFRONT, can also potentially be used for diagnosis, monitoring, and tailoring treatments for worsening brain injury. Extension of these results to patients with intact skulls requires further study. 

   more » « less
3. Acoustofluidic separation enables early diagnosis of traumatic brain injury based on circulating exosomes

   https://doi.org/10.1038/s41378-021-00244-3  

   Wang, Zeyu; Wang, Haichen; Becker, Ryan; Rufo, Joseph; Yang, Shujie; Mace, Brian E.; Wu, Mengxi; Zou, Jun; Laskowitz, Daniel T.; Huang, Tony Jun (March 2021, Microsystems & Nanoengineering)

   Abstract Traumatic brain injury (TBI) is a global cause of morbidity and mortality. Initial management and risk stratification of patients with TBI is made difficult by the relative insensitivity of screening radiographic studies as well as by the absence of a widely available, noninvasive diagnostic biomarker. In particular, a blood-based biomarker assay could provide a quick and minimally invasive process to stratify risk and guide early management strategies in patients with mild TBI (mTBI). Analysis of circulating exosomes allows the potential for rapid and specific identification of tissue injury. By applying acoustofluidic exosome separation—which uses a combination of microfluidics and acoustics to separate bioparticles based on differences in size and acoustic properties—we successfully isolated exosomes from plasma samples obtained from mice after TBI. Acoustofluidic isolation eliminated interference from other blood components, making it possible to detect exosomal biomarkers for TBI via flow cytometry. Flow cytometry analysis indicated that exosomal biomarkers for TBI increase in the first 24 h following head trauma, indicating the potential of using circulating exosomes for the rapid diagnosis of TBI. Elevated levels of TBI biomarkers were only detected in the samples separated via acoustofluidics; no changes were observed in the analysis of the raw plasma sample. This finding demonstrated the necessity of sample purification prior to exosomal biomarker analysis. Since acoustofluidic exosome separation can easily be integrated with downstream analysis methods, it shows great potential for improving early diagnosis and treatment decisions associated with TBI. 

   more » « less
4. A Topological-Attention ConvLSTM Network and Its Application to EM Images

   https://doi.org/10.1007/978-3-030-87193-2_21  

   Yang, Jiaqi; Hu, Xiaoling; Chen, Chao; Tsai, Chialing (October 2021, Medical Image Computing and Computer Assisted Intervention (MICCAI))

   Structural accuracy of segmentation is important for fine-scale structures in biomedical images. We propose a novel Topological-Attention ConvLSTM Network (TACLNet) for 3D anisotropic image segmentation with high structural accuracy. We adopt ConvLSTM to leverage contextual information from adjacent slices while achieving high efficiency. We propose a Spatial Topological-Attention (STA) module to effectively transfer topologically critical information across slices. Furthermore, we propose an Iterative Topological-Attention (ITA) module that provides a more stable topologically critical map for segmentation. Quantitative and qualitative results show that our proposed method outperforms various baselines in terms of topology-aware evaluation metrics. 

   more » « less
5. Colon Segmentation Using Guided Sequential Episodic Training and Contrastive Learning

   https://doi.org/10.1007/978-3-031-78201-5_5  

   Harb, Samir; Ali, Asem; Yousuf, Mohamed; Elshazly, Salwa; Farag, Aly (December 2024, Springer Nature Switzerland)

   Abstract. Accurate colon segmentation on abdominal CT scans is crucial for various clinical applications. In this work, we propose an accurate AQ1 approach to colon segmentation from abdomen CT scans. Our architecture incorporates 3D contextual information via sequential episodic training (SET). In each episode, we used two consecutive slices, in a CT scan, as support and query samples in addition to other slices that did not include colon regions as negative samples. Choosing consecutive slices is a proper assumption for support and query samples, as the anatomy of the body does not have abrupt changes. Unlike traditional few-shot segmentation (FSS) approaches, we use the episodic training strategy in a supervised manner. In addition, to improve the discriminability of the learned features of the model, an embedding space is developed using contrastive learning. To guide the contrastive learning process, we use AQ2 an initial labeling that is generated by a Markov random field (MRF)- based approach. Finally, in the inference phase, we first detect the rec tum, which can be accurately extracted using the MRF-based approach, and then apply the SET on the remaining slices. Experiments on our private dataset of 98 CT scans and a public dataset of 30 CT scans illustrate that the proposed FSS model achieves a remarkable validation dice coefficient (DC) of 97.3% (Jaccard index, JD 94. 5%) compared to the classical FSS approaches 82.1% (JD 70.3%). Our findings highlight the efficacy of sequential episodic training in accurate 3D medical imaging segmentation. The codes for the proposed models are available at https://github.com/Samir-Farag/ICPR2024. 

   more » « less