HMHui Ma

Hui Ma

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

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hui Ma has a total of 17 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Turbo Gradient and Spin-Echo BLADE-DWI for Extraocular Muscles in Thyroid-Associated Ophthalmopathy

    by Qing Fu, Dingxi Liu, Hui Ma, Kun Zhou, Ting Yin, Chuansheng Zheng, Ziqiao Lei

    Purpose: To investigate feasibility and diagnostic performance of turbo gradient and spin-echo BLADE (proprietary name for Periodically Rotated Overlapping ParallEL Lines with Enhanced Reconstruction [PROPELLER] in Siemens MR systems)-diffusion weighted imaging (TGSE-BLADE-DWI) for depicting extraocular muscle (EOM) involvement and activity in thyroid-associated ophthalmopathy (TAO), and to compare TGSE-BLADE-DWI with readout-segmented echo-planar imaging (RESOLVE). Materials and methods: Thirty-five patients with identified TAO underwent the two DWI scans. Two radiologists visually scored the image quality of the two DWIs with respect to the susceptibility artifacts and geometric distortions on a three-point scale. The maximum size (Sizemax) of EOMs and corresponding ADCs (cADCs) of each patient were compared between the active and inactive phases. The clinical activity score (CAS) was used as a reference to assess the diagnostic performance of EOM ADCs for grading TAO activity. ROC analysis, Pearson correlation, and Wilcoxon signed-rank test were used for statistical analyses. Results: For scores of EOMs, the image quality of TGSE-BLADE-DWI was significantly higher than that of RESOLVE. There were no statistically significant differences between the AUCs of the two DWIs, Sizemax, or cADCs between the active and inactive phases. TGSE-BLADE-DWI ADCs were significantly higher than the RESOLVE ADCs in the right superior rectus, right lateral rectus, left superior rectus, and left inferior rectus. There were no statistically significant correlations between the cADC or Sizemax, and CAS. The highest AUC was 0.697 for RESOLVE and 0.657 for TGSE-BLADE-DWI. The best performing ADC threshold was 1.85 × 10−3 mm2/s with 85.7% sensitivity, 58.8% specificity and 66.67% accuracy for RESOLVE and 1.99 × 10−3 mm2/s with 79.0% sensitivity, and 55.6% specificity and 65.27% accuracy for TGSE-BLADE-DWI. Conclusion: Compared to RESOLVE, TGSE-BLADE-DWI provided improved image quality with fewer susceptibility artifacts and geometric distortions for EOM visualization and showed an equivalent performance in detecting active TAO.


  • Top Cited Papers
    Open Access

    Deep Learning-based Detection for COVID-19 from Chest CT using Weak Label

    by Chuansheng Zheng, Xianbo Deng, Qiang Fu, Qiang Zhou, Jiapei Feng, Hui Ma, Wenyu Liu, Xinggang Wang

    Accurate and rapid diagnosis of COVID-19 suspected cases plays a crucial role in timely quarantine and medical treatment. Developing a deep learning-based model for automatic COVID-19 detection on chest CT is helpful to counter the outbreak of SARS-CoV-2. A weakly-supervised deep learning-based software system was developed using 3D CT volumes to detect COVID-19. For each patient, the lung region was segmented using a pre-trained UNet; then the segmented 3D lung region was fed into a 3D deep neural network to predict the probability of COVID-19 infectious. 499 CT volumes collected from Dec. 13, 2019, to Jan. 23, 2020, were used for training and 131 CT volumes collected from Jan 24, 2020, to Feb 6, 2020, were used for testing. The deep learning algorithm obtained 0.959 ROC AUC and 0.976 PR AUC. There was an operating point with 0.907 sensitivity and 0.911 specificity in the ROC curve. When using a probability threshold of 0.5 to classify COVID-positive and COVID-negative, the algorithm obtained an accuracy of 0.901, a positive predictive value of 0.840 and a very high negative predictive value of 0.982. The algorithm took only 1.93 seconds to process a single patient’s CT volume using a dedicated GPU. Our weakly-supervised deep learning model can accurately predict the COVID-19 infectious probability in chest CT volumes without the need for annotating the lesions for training. The easily-trained and highperformance deep learning algorithm provides a fast way to identify COVID-19 patients, which is beneficial to control the outbreak of SARS-CoV-2. The developed deep learning software is available at https://github.com/sydney0zq/covid-19-detection .


  • Multislice Spiral Computed Tomography Imaging in Congenital Inner Ear Malformations

    by Hui Ma, Ping Han, Bo Liang, Zhi-Liang Tian, Zi-Qiao Lei, Wei-Jia Kong, Gan-Sheng Feng

    The purpose of this study is to evaluate the usefulness of multislice spiral computed tomography (CT) in the diagnosis of congenital inner ear malformations. Forty-four patients with sensorineural hearing loss were examined on a Somatom Sensation 16 (Siemens) CT scanner. The 3-dimensional reconstructions and multiplanar reformation (MPR) were performed using the volume-rendering technique (VRT) on the workstation. Of the 44 patients examined for this study, 25 patients were found to be normal and 19 patients (36 ears) were diagnosed with congenital inner ear malformations. Of the malformations, the axial, MPR, and VRT images can all display the site and degree in 33 of the ears. Volume-rendering technique images were superior to the axial images in displaying the malformations in 3 ears with small lateral semicircular canal malformations. The common malformations were Michel deformity (1 ear), common cavity deformity (3 ears), incomplete partition I (3 ears), incomplete partition II (Mondini deformity) (5 ears), vestibular and semicircular canal malformations (14 ears), enlarged vestibular aqueduct (16 ears, 6 of which had other malformations), and internal auditory canal malformation (8 ears, all accompanied by other malformations). Multislice spiral CT allows a comprehensively assessment of various congenital inner ear malformations through high-quality MPR and VRT reconstructions. Volume-rendering technique images can display the site and degree of the malformation 3-dimensionally and intuitionisticly. This is very useful to the cochlear implantation.