Lin Shi

Lin Shi

Auto-generated profile, based on publicly available information

Lin Shi

Department of Imaging and Interventional Radiology, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong, China

Research Focus & Metrics

Academic Career Length

22Years (2005-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 May 2016 in IEEE Transactions on Medical Imaging
Verified
Journal Article
Closed access
QDQi Dou
Qi Dou
HCHao Chen
Hao Chen
LYLequan Yu
Lequan Yu
LZLei Zhao
Lei Zhao
Jing QinJing Qin
Jing Qin
PHPheng-Ann Heng
Pheng-Ann Heng
Cerebral microbleeds (CMBs) are small haemorrhages nearby blood vessels. They have been recognized as important diagnostic biomarkers for many cerebrovascular diseases and cognitive dysfunctions. In current clinical routine, CMBs are manually labelled by radiologists but this procedure is laborious, time-consuming, and error prone. In this paper, we propose a novel automatic method to detect CMBs from magnetic resonance (MR) images by exploiting the 3D convolutional neural network (CNN). Compared with previous methods that employed either low-level hand-crafted descriptors or 2D CNNs, our method can take full advantage of spatial contextual information in MR volumes to extract more representative high-level features for CMBs, and hence achieve a much better detection accuracy. To further improve the detection performance while reducing the computational cost, we propose a cascaded framework under 3D CNNs for the task of CMB detection. We first exploit a 3D fully convolutional network (FCN) strategy to retrieve the candidates with high probabilities of being CMBs, and then apply a well-trained 3D CNN discrimination model to distinguish CMBs from hard mimics. Compared with traditional sliding window strategy, the proposed 3D FCN strategy can remove massive redundant computations and dramatically speed up the detection process. We constructed a large dataset with 320 volumetric MR scans and performed extensive experiments to validate the proposed method, which achieved a high sensitivity of 93.16% with an average number of 2.74 false positives per subject, outperforming previous methods using low-level descriptors or 2D CNNs by a significant margin. The proposed method, in principle, can be adapted to other biomarker detection tasks from volumetric medical data.

30 March 2018 in IEEE Transactions on Medical Imaging
Verified
Journal Article
Closed access
RZRongzhao Zhang
Rongzhao Zhang
LZLei Zhao
Lei Zhao
WLWutao Lou
Wutao Lou
JAJill M. Abrigo
Jill M. Abrigo
VMVincent C. T. Mok
Vincent C. T. Mok
LSLin Shi
Lin Shi
Acute ischemic stroke is recognized as a common cerebral vascular disease in aging people. Accurate diagnosis and timely treatment can effectively improve the blood supply of the ischemic area and reduce the risk of disability or even death. Understanding the location and size of infarcts plays a critical role in the diagnosis decision. However, manual localization and quantification of stroke lesions are laborious and time-consuming. In this paper, we propose a novel automatic method to segment acute ischemic stroke from diffusion weighted images (DWIs) using deep 3-D convolutional neural networks (CNNs). Our method can efficiently utilize 3-D contextual information and automatically learn very discriminative features in an end-to-end and data-driven way. To relieve the difficulty of training very deep 3-D CNN, we equip our network with dense connectivity to enable the unimpeded propagation of information and gradients throughout the network. We train our model with Dice objective function to combat the severe class imbalance problem in data. A DWI data set containing 242 subjects (90 for training, 62 for validation, and 90 for testing) with various types of acute ischemic stroke was constructed to evaluate our method. Our model achieved high performance on various metrics (Dice similarity coefficient: 79.13%, lesionwise precision: 92.67%, and lesionwise F1 score: 89.25%), outperforming the other state-of-the-art CNN methods by a large margin. We also evaluated the model on ISLES2015-SSIS data set and achieved very competitive performance, which further demonstrated its generalization capacity. The proposed method is fast and accurate, demonstrating a good potential in clinical routines.

12 September 2017 in Journal of Cerebral Blood Flow & Metabolism
Verified
Journal Article
Closed access
LZLei Zhao
Lei Zhao
JBJ Matthijs Biesbroek
J Matthijs Biesbroek
LSLin Shi
Lin Shi
WLWenyan Liu
Wenyan Liu
HKHugo J Kuijf
Hugo J Kuijf
AWAdrian Wong
Adrian Wong
Lesion location is an important determinant for post-stroke cognitive impairment. Although several ‘strategic’ brain regions have previously been identified, a comprehensive map of strategic brain regions for post-stroke cognitive impairment is lacking due to limitations in sample size and methodology. We aimed to determine strategic brain regions for post-stroke cognitive impairment by applying multivariate lesion-symptom mapping in a large cohort of 410 acute ischemic stroke patients. Montreal Cognitive Assessment at three to six months after stroke was used to assess global cognitive functioning and cognitive domains (memory, language, attention, executive and visuospatial function). The relation between infarct location and cognition was assessed in multivariate analyses at the voxel-level and the level of regions of interest using support vector regression. These two assumption-free analyses consistently identified the left angular gyrus, left basal ganglia structures and the white matter around the left basal ganglia as strategic structures for global cognitive impairment after stroke. A strategic network involving several overlapping and domain-specific cortical and subcortical structures was identified for each of the cognitive domains. Future studies should aim to develop even more comprehensive infarct location-based models for post-stroke cognitive impairment through multicenter studies including thousands of patients.

28 July 2015 in BioMedical Engineering OnLine
Verified
Journal Article
Open access
PDF file
XSXiaofei Sun
Xiaofei Sun
LSLin Shi
Lin Shi
YLYishan Luo
Yishan Luo
WYWei Yang
Wei Yang
HLHongpeng Li
Hongpeng Li
DWDefeng Wang
Defeng Wang
Intensity normalization is an important preprocessing step in brain magnetic resonance image (MRI) analysis. During MR image acquisition, different scanners or parameters would be used for scanning different subjects or the same subject at a different time, which may result in large intensity variations. This intensity variation will greatly undermine the performance of subsequent MRI processing and population analysis, such as image registration, segmentation, and tissue volume measurement. In this work, we proposed a new histogram normalization method to reduce the intensity variation between MRIs obtained from different acquisitions. In our experiment, we scanned each subject twice on two different scanners using different imaging parameters. With noise estimation, the image with lower noise level was determined and treated as the high-quality reference image. Then the histogram of the low-quality image was normalized to the histogram of the high-quality image. The normalization algorithm includes two main steps: (1) intensity scaling (IS), where, for the high-quality reference image, the intensities of the image are first rescaled to a range between the low intensity region (LIR) value and the high intensity region (HIR) value; and (2) histogram normalization (HN),where the histogram of low-quality image as input image is stretched to match the histogram of the reference image, so that the intensity range in the normalized image will also lie between LIR and HIR. We performed three sets of experiments to evaluate the proposed method, i.e., image registration, segmentation, and tissue volume measurement, and compared this with the existing intensity normalization method. It is then possible to validate that our histogram normalization framework can achieve better results in all the experiments. It is also demonstrated that the brain template with normalization preprocessing is of higher quality than the template with no normalization processing. We have proposed a histogram-based MRI intensity normalization method. The method can normalize scans which were acquired on different MRI units. We have validated that the method can greatly improve the image analysis performance. Furthermore, it is demonstrated that with the help of our normalization method, we can create a higher quality Chinese brain template.

15 July 2019 in IEEE Access
Verified
Journal Article
Open access
PSPan Sun
Pan Sun
DWDefeng Wang
Defeng Wang
VMVincent Ct Mok
Vincent Ct Mok
LSLin Shi
Lin Shi
Radiomics-based researches have shown predictive abilities with machine-learning approaches. However, it is still unknown whether different radiomics strategies affect the prediction performance. The aim of this study was to compare the prediction performance of frequently utilized radiomics feature selection and classification methods in glioma grading. Quantitative radiomics features were extracted from tumor regions in 210 Glioblastoma (GBM) and 75 low-grade glioma (LGG) MRI subjects. Then, the diagnostic performance of sixteen feature selection and fifteen classification methods were evaluated by using two different test modes: ten-fold cross-validation and percentage split. Balanced accuracy and area under the curve (AUC) of the receiver operating characteristic were used to evaluate prediction performance. In addition, the roles of the number of selected features, feature type, MRI modality, and tumor sub-region were compared to optimize the radiomics-based prediction. The results indicated that the combination of feature selection method L 1 -based linear support vector machine (L 1 -SVM) and classifier multi-layer perceptron (MLPC) achieved the best performance in the differentiation of GBM and LGG in both ten-fold cross validation (balanced accuracy:0.944, AUC:0.986) and percentage split (balanced accuracy:0.953, AUC:0.981). For radiomics feature extraction, the enhancing tumor region (ET) combined with necrotic and non-enhancing tumor (NCR/NET) regions in T1 post-contrast (T1-Gd) modality provided more considerable tumor-related phenotypes than other combinations of tumor region and MRI modality. Our comparative investigation indicated that both feature selection methods and machine learning classifiers affected the predictive performance in glioma grading. Also, the cross-combination strategy for comparison of radiomics feature selection and classification methods provided a way of searching optimal machine learning model for future radiomics-based prediction.

1 July 2014 in Magnetic Resonance Imaging
Verified
Journal Article
Closed access
RLRyan Wen Liu
Ryan Wen Liu
LSLin Shi
Lin Shi
HWHuang Wenhua
Huang Wenhua
JXJing Xu
Jing Xu
SYSimon Chun Ho Yu
Simon Chun Ho Yu
DWDefeng Wang
Defeng Wang