GGGuanzhong Gong

Guanzhong Gong

*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


Guanzhong Gong has a total of 102 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Optimization of temporal phase in multi-phase contrast-enhanced MRI for gross tumor volume determination in pancreatic cancer radiotherapy

    by Kangning Meng, Yujie Guo, Yong Yin, Ya Su, Lizhen Wang, Ruozheng Wang, Guanzhong Gong

    Background: Pancreatic cancer is characterized by an insidious onset and rapid progression, and the accurate determination of the gross tumor volume (GTV) constitutes a critical prerequisite for ensuring the efficacy of radiotherapy. Multiphase contrast-enhanced magnetic resonance imaging (CE-MRI) enables the dynamic visualization of tumor hemodynamic perfusion characteristics; however, the tissue discrimination capability varies substantially across different enhancement phases. To date, there is no universal consensus on the optimal imaging phase for GTV determination in pancreatic cancer. This study aimed to quantitatively analyze the differences in imaging and GTV determination of pancreatic cancer using multiphase CE-MRI, thereby providing a basis for selecting the optimal phase for GTV determination. Methods: Thirty patients with advanced pancreatic cancer [American Joint Committee on Cancer (AJCC) stage III–IV] who underwent magnetic resonance (MR) simulation were retrospectively enrolled in this study. MR T1-weighted images (T1WI) and contrast-enhanced T1-weighted images (CE-T1WI) were obtained at 15 s, 45 s, 75 s, 150 s, and >20 min after contrast injection. The GTV was determined from these different image sequences and named GTV-T1WI, GTV-15 s, GTV-45 s, GTV-75 s, GTV-150 s, and GTV-20 min. Differences in mean signal intensity (SI), SI contrast, volume, and shape among the different GTVs were compared. Normal pancreatic tissue was defined as a 1 cm3 region of interest of the parenchyma, strictly excluding blood vessels and pancreatic ducts. Results: The mean SI of the GTV was lower than that of the normal pancreatic tissue at each time phase (P1WI-15 s, the SI contrast between GTV and normal pancreatic tissue was the highest at 0.45±0.10, significantly superior to the T1WI (0.34±0.13, P-15 s volume was 21.02±12.43 cm3. Compared with CE-T1WI-15 s, the SI contrast between the GTV and normal pancreatic tissue on T1WI and CE-T1WI-45 s to CE-T1WI-20 min decreased by 22.42–77.43% (P-15 s, the volume of GTV-T1WI and GTV-45 s–GTV-20 min decreased by −14.10–22.75%. Except for GTV-15 s and GTV-45 s, GTV-15 s and GTV-75 s, and GTV-45 s and GTV-75 s, the differences in GTV volumes in the other phases were statistically significant (P-15 s. The Dice similarity coefficients (DSCs) of GTV-T1WI, GTV-45 s, GTV-75 s, GTV-150 s, and GTV-20 min were 0.74±0.10, 0.79±0.11, 0.76±0.13, 0.72±0.15, and 0.64±0.13, respectively. Conclusions: The CE-T1WI-15 s sequence demonstrated significant improvements in SI contrast and boundary definition. Consequently, it holds significant potential as an optimal sequence for GTV determination in pancreatic cancer radiotherapy, warranting further validation in larger cohorts.

  • Open Access

    SwitchNet: Adaptive Distribution Switching in UNet for Brain Lesion Segmentation

    by Jiahao Chen, Jingwen Guan, Bowen Xin, Yichao Hao, Guanzhong Gong, Rolf Bjerkvig, Jian Wang, Xiuying Wang

    Automatic brain lesion segmentation enhances diagnostic efficiency by enabling detailed texture analysis and precise delineation of tumor subregions. Multimodal MRI has improved segmentation accuracy by combining complementary information from different modalities. Conventional methods either fuse all modalities uniformly, obscuring how individual modalities contribute to specific segmentation subtasks, or predefine modality-to-subregion mappings based on prior medical knowledge. The former limits interpretability on modality contribution during training, while the latter relies on parameter-heavy architectures like cascaded subnetworks, making models struggle to adapt to varying modalities. To address these challenges, this paper proposes SwitchNet, a novel model that integrates interpretability into the training process and optimizes parameter efficiency without relying on predefined modality selection. First, we propose Adaptive Encoder and Decoder Blocks employing dynamic switching mechanisms to efficiently allocate feature space and prioritize critical subtasks. These blocks enable the model to automatically identify and utilize the most informative modalities. By strategically allocating parameters to modalities, our model optimizes overall parameter complexity while maintaining strong performance. Second, we propose a Guide-Contribution Mechanism to provide interpretability during training by quantitatively revealing the contributions of individual modalities to the segmentation process. This mechanism offers valuable insights into how the model delineates tumor subregions. SwitchNet was validated on three benchmark datasets, including BraTS 2023, ISLES 2022, and UCSF-PDGM, achieving competitive segmentation performance while significantly enhancing interpretability and maintaining parameter efficiency without extra cost. These results highlight its potential for efficient tumor segmentation and clinical explainability.


  • Learnable dendrite neural P systems and applications in survival prediction of glioblastoma patients

    by Xiu Yin, Xiyu Liu, Shulei Chang, Bosheng Song, Guanzhong Gong, Jiaxing Yin, Dengwang Li, Jie Xue


  • Unsupervised anomaly detection in brain MRI via disentangled anatomy learning

    by Tao Yang, Xiuying Wang, Hao Liu, Guanzhong Gong, Lian-Ming Wu, Yu-Ping Wang, Lisheng Wang


  • Denoising-enhanced pancreatic segmentation using diverse kernel mutual adaptive learning

    by Qi Li, Lianghui Cheng, Zhen Xia, Pu Huang, Xiyu Liu, Guanzhong Gong, Dengwang Li, Jie Xue


  • Open Access

    Imaging manifestations of brain metastases in ultra-long delay-enhanced MRI > 60 min

    by Shanshan Du, Yujie Guo, Yong Yin, Rui Liu, Kangning Meng, Ya Su, Lizhen Wang, Ruozheng Wang, Guanzhong Gong

  • Open Access

    Spatial-demographic analysis model for brain metastases distribution

    by Lin Zhang, Tongtong Che, Bowen Xin, Shuyu Li, Guanzhong Gong, Xiuying Wang

    Purpose: The distribution analysis of the morphologic characteristics and spatial relations among brain metastases (BMs) to guide screening and early diagnosis. Material and Methods: This retrospective study analysed 4314 BMs across 30 brain regions from MRIs of 304 patients. This paper proposed a unified analysis model based on persistent homology (PH) and graph modelling to provide a comprehensive portrait of BMs distribution. Spatial relationships are quantified through dynamic multiple-scale graphs constructed with Rips filtration. The multi-scale centrality importance and clustering coefficients are extracted to decode BMs spatial relations. Morphologic BMs characteristics are further analysed by varying radius and volume values that are considered as clinically influential factors. Finally, two-tailed proportional hypothesis testing is used for BM statistical distribution analysis. Results: For spatial analysis, results have shown a statistical increase in the proportions of high-level centrality BMs at the left cerebellum (p<0.01). BMs rapidly form graphs with high clustering rather than those with high centrality. For demographic analysis, the cerebellum and frontal are the top high-frequency areas of BMs with 0-4 and 5-10 radii. Statistical increases in the proportions of BMs at cerebellum (p<0.01). Conclusion: Results indicate that distributions of both BMs spatial relations and demographics are statistically non-random. This research offers novel insights into the BMs distribution analysis, providing physicians with the BMs demographic to guide screening and early diagnosis.


  • Open Access

    Hippocampal sparing in whole-brain radiotherapy for brain metastases: controversy, technology and the future

    by Rui Liu, GuanZhong Gong, KangNing Meng, ShanShan Du, Yong Yin

    Whole-brain radiotherapy (WBRT) plays an irreplaceable role in the treatment of brain metastases (BMs), but cognitive decline after WBRT seriously affects patients’ quality of life. The development of cognitive dysfunction is closely related to hippocampal injury, but standardized criteria for predicting hippocampal injury and dose limits for hippocampal protection have not yet been developed. This review systematically reviews the clinical efficacy of hippocampal avoidance - WBRT (HA-WBRT), the controversy over dose limits, common methods and characteristics of hippocampal imaging and segmentation, differences in hippocampal protection by common radiotherapy (RT) techniques, and the application of artificial intelligence (AI) and radiomic techniques for hippocampal protection. In the future, the application of new techniques and methods can improve the consistency of hippocampal dose limit determination and the prediction of the occurrence of cognitive dysfunction in WBRT patients, avoiding the occurrence of cognitive dysfunction in patients and thus benefiting more patients with BMs.


  • Open Access

    An Effective Approach to Improve the Automatic Segmentation and Classification Accuracy of Brain Metastasis by Combining Multi-phase Delay Enhanced MR Images

    by Mingming Chen, Yujie Guo, Pengcheng Wang, Qi Chen, Lu Bai, Shaobin Wang, Ya Su, Lizhen Wang, Guanzhong Gong

    The objective of this study is to analyse the diffusion rule of the contrast media in multi-phase delayed enhanced magnetic resonance (MR) T1 images using radiomics and to construct an automatic classification and segmentation model of brain metastases (BM) based on support vector machine (SVM) and Dpn-UNet. A total of 189 BM patients with 1047 metastases were enrolled. Contrast-enhanced MR images were obtained at 1, 3, 5, 10, 18, and 20 min following contrast medium injection. The tumour target volume was delineated, and the radiomics features were extracted and analysed. BM segmentation and classification models in the MR images with different enhancement phases were constructed using Dpn-UNet and SVM, and differences in the BM segmentation and classification models with different enhancement times were compared. (1) The signal intensity for BM decreased with time delay and peaked at 3 min. (2) Among the 144 optimal radiomics features, 22 showed strong correlation with time (highest R-value = 0.82), while 41 showed strong correlation with volume (highest R-value = 0.99). (3) The average dice similarity coefficients of both the training and test sets were the highest at 10 min for the automatic segmentation of BM, reaching 0.92 and 0.82, respectively. (4) The areas under the curve (AUCs) for the classification of BM pathology type applying single-phase MRI was the highest at 10 min, reaching 0.674. The AUC for the classification of BM by applying the six-phase image combination was the highest, reaching 0.9596, and improved by 42.3% compared with that by applying single-phase images at 10 min. The dynamic changes of contrast media diffusion in BM can be reflected by multi-phase delayed enhancement based on radiomics, which can more objectively reflect the pathological types and significantly improve the accuracy of BM segmentation and classification.


  • Hypergraph-Based Numerical Neural-Like P Systems for Medical Image Segmentation

    by Jie Xue, Liwen Ren, Bosheng Song, Yujie Guo, Jie Lu, Xiyu Liu, Guanzhong Gong, Dengwang Li

    Neural-like P systems are membrane computing models inspired by natural computing and are viewed as third-generation neural network models. Although real neurons have complex structures, classical neural-like P systems simplify the structures and corresponding mechanisms to two-dimensional graphs or tree-based firing and forgetting communications, which limit the real applications of these models. In this paper, we propose a hypergraph-based numerical neural-like (HNN) P system containing five types of neurons to describe the high-order correlations among neuron structures. Three new kinds of communication mechanisms among neurons are also proposed to address numerical variables and functions. Based on the new neural-like P system, a tumor/organ segmentation model for medical images is developed. The experimental results indicate that the proposed models outperform the state-of-the-art methods based on two hippocampal datasets and a multiple brain metastases dataset, thus verifying the effectiveness of the HNN P system in correctly segmenting tumors/organs.