JCJie-Zhi Cheng

Jie-Zhi Cheng

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Jie-Zhi Cheng has a total of 77 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Domain generalization for mammographic image analysis with contrastive learning

    by Zheren Li, Zhiming Cui, Lichi Zhang, Sheng Wang, Chenjin Lei, Xi Ouyang, Dongdong Chen, Xiangyu Zhao, Chunling Liu, Zaiyi Liu, et al.


  • Open Access

    Editorial: Workflow optimisation for radiological imaging

    by Jie-Zhi Cheng, Minjeong Kim, Xin Yang

    Deep learning techniques have effectively addressed numerous technical challenges in medical image analysis, leading to the flourishing advancement of artificial intelligence (AI) applications in medicine. The coverage of these applications may include the improvement of the image diagnostic workup, imagingworkflow optimization, burden alleviation of repetitive medical tasks, and comprehensive analysis for treatment planning. In recent years, numerous commercial AI applications have been introduced into clinical practice, demonstrating clear benefits. Concurrently, the literature has seen a surge in promising clinical validation studies for AI. Many of these significant studies have shown AI’s efficacy, supported by large datasets from multiple institutions and countries. To advance research on AI applications in radiology, we organized this special topic to explore new AI and machine learning methods for enhancing medicalworkflows, procedures, and diagnostic processes related to radiological imaging. We received numerous high-quality manuscripts, and after a rigorous review process, 16 of them were ultimately published. The published articles in this issue can be categorized into several research topics of computer-aided diagnosis,cancer staging and metastasis prediction, quantitative analysis, and prognosis prediction. Each research topic is further summarized as follows.Computer-aided diagnosis (CAD)Li et al. utilized deep learning and radiomics techniques to diagnose Parkinson’s disease based on T1-weighted magnetic resonance imaging. The VB-net demarcated the brain into 109 regions, from which a total of 2,264 radiomics features were extracted. The support vector machine was further applied toidentify Parkinson’s disease with high accuracy. Yang et al. focused on the early and rapid diagnosis of mild cognitive impairment, which is crucial for improving the prognosis of Alzheimer’s disease. The structures of bilateral hippocampus and parahippocampal gyrus were segmented from MR images for the Gaussian process models for the final MCI identification. Wei et al. developed a novel prediction model that integrated multi-modal radiomics features derived from diffusion-weighted imaging and apparent diffusion coefficient images, along with multi-clinical features for the diagnosis of acute ischemic stroke. Lin proposed a deep learning-based diagnostic system to achieve early detection of esophageal cancer from non-contrast chest computed tomography (CT) images. For the differential diagnosis, Feng et al. extensively compared the performance of different machine learning approaches for the staging of testicular lesions in MR images. Lu et al. developed a deep learning radiomics model on multimodal images of ultrasonography, mammography, and MRI, which is further augmented with transfer learning, for the malignancy identification of breast lesions. Cui et al. developed a CT-based radiomics model to distinguishbetween laryngeal squamous cell carcinoma and squamous cell hyperplasia. The applications of CAD is quite extensive, ranging from brain diseases to cancer problems on various body parts.Cancer staging, metastasis predictionWu et al. manifested the early stage of cervical cancers with radiomics features from the T2-weighted images and apparent diffusion coefficient maps. Tian et al. developed a segmentation-classification paradigm for rectal cancer (RC) lesions for the assessment of different T-stages.For the metastasis prediction, Ma et al. developed a clinical-radiomics nomogram, which leveraged deep learning techniques, preoperative MR images, and clinical characteristics, to achieve accurate prediction of lymph node metastasis in RC. Fu et al. developed multitask prediction models for High-grade serous ovarian cancer with the radiomics profiling on preoperative CT scans. Specifically, the radiomics features were derived from manual ROIs, and then classification models were constructed with the feature selection scheme to predict the assessments of R0 resection, lymph node invasion, and distant metastasisstatus. The staging and metastasis prediction for cancers are very informative for treatment planning.Quantitative analysisSun et al. employed the ResUNet to delineate the testicular from the MR scans for the measurement of testicular volume, which is an essential for the evaluation of functions and pathologies. Jiang et al. utilized the radiomics analysis on dual-energy CT images of the pancreas to establish a quantitative biomarker for type 2 diabetes mellitus. Deep learning algorithms were applied to segment pancreas, and were sequenced with the later classifiers of random forest, support vector machine and logistic regression. Meng et al. combined the clinical features, CT imaging signs, and radiomics features for the differentiation of Mycoplasma pneumonia in adults and children. The radiomics features are well-correlated with CT imaging signs, which provided a representation of different CT imaging signs as potential quantitative biomarkers. Jia et al. found that periplaque pericoronary adipose tissue (PCAT) was more valuable thanproximal PCAT for the evaluation of coronary atherosclerosis. Spectral parameters of peri-plaque PCAT could further improve the diagnostic accuracy. The quantitative parameters w.r.t. peri-plaque PCAT attenuation were shownto be a potential biomarkers for plaque vulnerability and the hemodynamic characteristics of coronary atherosclerosis. The quantitative measurement can replace tedious manual drawings, whereas the biomarker will be very beneficial for disease diagnosis and treatment decision.Prognosis predictionXue et al. predicted the prognosis of early acute pancreatitis with a clinical–radiomics model on CT scans. The prediction of disease prognosis is very helpful for the determination of treatment approaches.The AI applications in medicine is booming, and can potentially benefit various aspects of clinical...


  • Open Access

    Domain Generalization for Mammographic Image Analysis with Contrastive Learning

    by Zheren Li, Zhiming Cui, Lichi Zhang, Sheng Wang, Chenjin Lei, Xi Ouyang, Dongdong Chen, Xiangyu Zhao, Yajia Gu, Zaiyi Liu, et al.

    The deep learning technique has been shown to be effectively addressed several image analysis tasks in the computer-aided diagnosis scheme for mammography. The training of an efficacious deep learning model requires large data with diverse styles and qualities. The diversity of data often comes from the use of various scanners of vendors. But, in practice, it is impractical to collect a sufficient amount of diverse data for training. To this end, a novel contrastive learning is developed to equip the deep learning models with better style generalization capability. Specifically, the multi-style and multi-view unsupervised self-learning scheme is carried out to seek robust feature embedding against style diversity as a pretrained model. Afterward, the pretrained network is further fine-tuned to the downstream tasks, e.g., mass detection, matching, BI-RADS rating, and breast density classification. The proposed method has been evaluated extensively and rigorously with mammograms from various vendor style domains and several public datasets. The experimental results suggest that the proposed domain generalization method can effectively improve performance of four mammographic image tasks on the data from both seen and unseen domains, and outperform many state-of-the-art (SOTA) generalization methods.


  • Domain Generalization for Mammography Detection via Multi-style and Multi-view Contrastive Learning

    by Zheren Li, Zhiming Cui, Sheng Wang, Yuji Qi, Xi Ouyang, Qitian Chen, Yuezhi Yang, Zhong Xue, Dinggang Shen, Jie-Zhi Cheng


  • Learning Hierarchical Attention for Weakly-Supervised Chest X-Ray Abnormality Localization and Diagnosis

    by Xi Ouyang, Srikrishna Karanam, Ziyan Wu, Terrence Chen, Jiayu Huo, Xiang Sean Zhou, Qian Wang, Jie-Zhi Cheng

    We consider the problem of abnormality localization for clinical applications. While deep learning has driven much recent progress in medical imaging, many clinical challenges are not fully addressed, limiting its broader usage. While recent methods report high diagnostic accuracies, physicians have concerns trusting these algorithm results for diagnostic decision-making purposes because of a general lack of algorithm decision reasoning and interpretability. One potential way to address this problem is to further train these models to localize abnormalities in addition to just classifying them. However, doing this accurately will require a large amount of disease localization annotations by clinical experts, a task that is prohibitively expensive to accomplish for most applications. In this work, we take a step towards addressing these issues by means of a new attention-driven weakly supervised algorithm comprising a hierarchical attention mining framework that unifies activation- and gradient-based visual attention in a holistic manner. Our key algorithmic innovations include the design of explicit ordinal attention constraints, enabling principled model training in a weakly-supervised fashion, while also facilitating the generation of visual-attention-driven model explanations by means of localization cues. On two large-scale chest X-ray datasets (NIH ChestX-ray14 and CheXpert), we demonstrate significant localization performance improvements over the current state of the art while also achieving competitive classification performance.


  • Self-co-attention neural network for anatomy segmentation in whole breast ultrasound

    by Baiying Lei, Shan Huang, Hang Li, Ran Li, Cheng Bian, Yi-Hong Chou, Jing Qin, Peng Zhou, Xuehao Gong, Jie-Zhi Cheng


  • Open Access

    mr2NST: Multi-Resolution and Multi-Reference Neural Style Transfer for Mammography

    by Sheng Wang, Jiayu Huo, Xi Ouyang, Jifei Che, Xuhua Ren, Zhong Xue, Qian Wang, Jie-Zhi Cheng

    Computer-aided diagnosis with deep learning techniques has been shown to be helpful for the diagnosis of the mammography in many clinical studies. However, the image styles of different vendors are very distinctive, and there may exist domain gap among different vendors that could potentially compromise the universal applicability of one deep learning model. In this study, we explicitly address style variety issue with the proposed multi-resolution and multi-reference neural style transfer (mr2NST) network. The mr2NST can normalize the styles from different vendors to the same style baseline with very high resolution. We illustrate that the image quality of the transferred images is comparable to the quality of original images of the target domain (vendor) in terms of NIMA scores. Meanwhile, the mr2NST results are also shown to be helpful for the lesion detection in mammograms.

  • Deep multilevel contextual networks for biomedical image segmentation

    by Hao Chen, Qi Dou, Xiaojuan Qi, Jie-Zhi Cheng, Pheng-Ann Heng

  • Corrections to “Accurate Cervical Cell Segmentation From Overlapping Clumps in Pap Smear Images” [Jan 17 288-300]

    by Youyi Song, Ee-Leng Tan, Xudong Jiang, Jie-Zhi Cheng, Baiying Lei, Tianfu Wang

    In [1] , Baiying Lei was indicated as the corresponding author. Tianfu Wang and Baiying Lei should have been indicated as the corresponding authors.