Zujun Hou

Bio: Zujun Hou is an academic researcher. The author has contributed to research in topics: Medicine & Artificial intelligence. The author has an hindex of 1, co-authored 1 publications receiving 208 citations.

A Review on MR Image Intensity Inhomogeneity Correction.

Abstract: Intensity inhomogeneity (IIH) is often encountered in MR imaging, and a number of techniques have been devised to correct this artifact. This paper attempts to review some of the recent developments in the mathematical modeling of IIH field. Low-frequency models are widely used, but they tend to corrupt the low-frequency components of the tissue. Hypersurface models and statistical models can be adaptive to the image and generally more stable, but they are also generally more complex and consume more computer memory and CPU time. They are often formulated together with image segmentation within one framework and the overall performance is highly dependent on the segmentation process. Beside these three popular models, this paper also summarizes other techniques based on different principles. In addition, the issue of quantitative evaluation and comparative study are discussed.

A Weakly Supervised Deep Learning Approach for Leakage Detection in Fluorescein Angiography Images

Abstract: Purpose The purpose of this study was to design an automated algorithm that can detect fluorescence leakage accurately and quickly without the use of a large amount of labeled data. Methods A weakly supervised learning-based method was proposed to detect fluorescein leakage without the need for manual annotation of leakage areas. To enhance the representation of the network, a residual attention module (RAM) was designed as the core component of the proposed generator. Moreover, class activation maps (CAMs) were used to define a novel anomaly mask loss to facilitate more accurate learning of leakage areas. In addition, sensitivity, specificity, accuracy, area under the curve (AUC), and dice coefficient (DC) were used to evaluate the performance of the methods. Results The proposed method reached a sensitivity of 0.73 ± 0.04, a specificity of 0.97 ± 0.03, an accuracy of 0.95 ± 0.05, an AUC of 0.86 ± 0.04, and a DC of 0.87 ± 0.01 on the HRA data set; a sensitivity of 0.91 ± 0.02, a specificity of 0.97 ± 0.02, an accuracy of 0.96 ± 0.03, an AUC of 0.94 ± 0.02, and a DC of 0.85 ± 0.03 on Zhao's publicly available data set; and a sensitivity of 0.71 ± 0.04, a specificity of 0.99 ± 0.06, an accuracy of 0.87 ± 0.06, an AUC of 0.85 ± 0.02, and a DC of 0.78 ± 0.04 on Rabbani's publicly available data set. Conclusions The experimental results showed that the proposed method achieves better performance on fluorescence leakage detection and can detect one image within 1 second and thus has great potential value for clinical diagnosis and treatment of retina-related diseases, such as diabetic retinopathy and malarial retinopathy. Translational Relevance The proposed weakly supervised learning-based method that automates the detection of fluorescence leakage can facilitate the assessment of retinal-related diseases.

A comparative study between deep learning and radiomics models in grading liver tumors using hepatobiliary phase contrast-enhanced MR images

Abstract: To compare a deep learning model with a radiomics model in differentiating high-grade (LR-3, LR-4, LR-5) liver imaging reporting and data system (LI-RADS) liver tumors from low-grade (LR-1, LR-2) LI-RADS tumors based on the contrast-enhanced magnetic resonance images.Magnetic resonance imaging scans of 361 suspected hepatocellular carcinoma patients were retrospectively reviewed. Lesion volume segmentation was manually performed by two radiologists, resulting in 426 lesions from the training set and 83 lesions from the test set. The radiomics model was constructed using a support vector machine (SVM) with pre-defined features, which was first selected using Chi-square test, followed by refining using binary least absolute shrinkage and selection operator (LASSO) regression. The deep learning model was established based on the DenseNet. Performance of the models was quantified by area under the receiver-operating characteristic curve (AUC), accuracy, sensitivity, specificity and F1-score.A set of 8 most informative features was selected from 1049 features to train the SVM classifier. The AUCs of the radiomics model were 0.857 (95% confidence interval [CI] 0.816-0.888) for the training set and 0.879 (95% CI 0.779-0.935) for the test set. The deep learning method achieved AUCs of 0.838 (95% CI 0.799-0.871) for the training set and 0.717 (95% CI 0.601-0.814) for the test set. The performance difference between these two models was assessed by t-test, which showed the results in both training and test sets were statistically significant.The deep learning based model can be trained end-to-end with little extra domain knowledge, while the radiomics model requires complex feature selection. However, this process makes the radiomics model achieve better performance in this study with smaller computational cost and more potential on model interpretability.

Evaluation of tracer kinetic parameters in cervical cancer using dynamic contrast-enhanced MRI as biomarkers in terms of biological relevance, diagnostic performance and inter-center variability

Abstract: Objectives This study assessed the clinical value of parameters derived from dynamic contrast-enhanced (DCE) MRI with respect to correlation with angiogenesis and proliferation of cervical cancer, performance of diagnosis and reproducibility of DCE-MRI parameters across MRI scanners. Materials and Methods A total of 113 patients with cervical carcinoma from two centers were included in this retrospective study. The DCE data were centralized and processed using five tracer kinetic models (TKMs) (Tofts, Ex-Tofts, ATH, SC, and DP), yielding the following parameters: volume transfer constant (Ktrans), extravascular extracellular volume (Ve), fractional volume of vascular space (Vp), blood flow (Fp), and permeability surface area product (PS). CD34 counts and Ki-67 PI (proliferation index) of cervical cancer and normal cervix tissue were obtained using immunohistochemical staining in Center 1. Results CD34 count and Ki-67 PI in cervical cancer were significantly higher than in normal cervix tissue (p<0.05). Parameter Ve from each TKM was significantly smaller in cervical cancer tissue than in normal cervix tissue (p<0.05), indicating the higher proliferation of cervical cancer cells. Ve of each TKM attained the largest AUC to diagnose cervical cancer. The distributions of DCE parameters for both cervical cancer and normal cervix tissue were not significantly different between two centers (P>0.05). Conclusion Parameter Ve was similar to the expression of Ki-67 in revealing the proliferation of tissue cells, attained good performance in diagnosis of cervical cancer, and demonstrated consistent findings on measured values across centers.

Dictionary learning LASSO for feature selection with application to hepatocellular carcinoma grading using contrast enhanced magnetic resonance imaging

Abstract: Introduction The successful use of machine learning (ML) for medical diagnostic purposes has prompted myriad applications in cancer image analysis. Particularly for hepatocellular carcinoma (HCC) grading, there has been a surge of interest in ML-based selection of the discriminative features from high-dimensional magnetic resonance imaging (MRI) radiomics data. As one of the most commonly used ML-based selection methods, the least absolute shrinkage and selection operator (LASSO) has high discriminative power of the essential feature based on linear representation between input features and output labels. However, most LASSO methods directly explore the original training data rather than effectively exploiting the most informative features of radiomics data for HCC grading. To overcome this limitation, this study marks the first attempt to propose a feature selection method based on LASSO with dictionary learning, where a dictionary is learned from the training features, using the Fisher ratio to maximize the discriminative information in the feature. Methods This study proposes a LASSO method with dictionary learning to ensure the accuracy and discrimination of feature selection. Specifically, based on the Fisher ratio score, each radiomic feature is classified into two groups: the high-information and the low-information group. Then, a dictionary is learned through an optimal mapping matrix to enhance the high-information part and suppress the low discriminative information for the task of HCC grading. Finally, we select the most discrimination features according to the LASSO coefficients based on the learned dictionary. Results and discussion The experimental results based on two classifiers (KNN and SVM) showed that the proposed method yielded accuracy gains, compared favorably with another 5 state-of-the-practice feature selection methods.

N4ITK: Improved N3 Bias Correction

Abstract: A variant of the popular nonparametric nonuniform intensity normalization (N3) algorithm is proposed for bias field correction. Given the superb performance of N3 and its public availability, it has been the subject of several evaluation studies. These studies have demonstrated the importance of certain parameters associated with the B-spline least-squares fitting. We propose the substitution of a recently developed fast and robust B-spline approximation routine and a modified hierarchical optimization scheme for improved bias field correction over the original N3 algorithm. Similar to the N3 algorithm, we also make the source code, testing, and technical documentation of our contribution, which we denote as ?N4ITK,? available to the public through the Insight Toolkit of the National Institutes of Health. Performance assessment is demonstrated using simulated data from the publicly available Brainweb database, hyperpolarized 3He lung image data, and 9.4T postmortem hippocampus data.

Minimization of Region-Scalable Fitting Energy for Image Segmentation

Abstract: Intensity inhomogeneities often occur in real-world images and may cause considerable difficulties in image segmentation. In order to overcome the difficulties caused by intensity inhomogeneities, we propose a region-based active contour model that draws upon intensity information in local regions at a controllable scale. A data fitting energy is defined in terms of a contour and two fitting functions that locally approximate the image intensities on the two sides of the contour. This energy is then incorporated into a variational level set formulation with a level set regularization term, from which a curve evolution equation is derived for energy minimization. Due to a kernel function in the data fitting term, intensity information in local regions is extracted to guide the motion of the contour, which thereby enables our model to cope with intensity inhomogeneity. In addition, the regularity of the level set function is intrinsically preserved by the level set regularization term to ensure accurate computation and avoids expensive reinitialization of the evolving level set function. Experimental results for synthetic and real images show desirable performances of our method.