Hui Chen

Hui Chen


Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

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

Representative Publications

View All Publications
1 July 2022 in Radiology
Verified
Journal Article
Closed access
HCHui Chen
Hui Chen
BYBo-Wen Yang
Bo-Wen Yang
LQLe Qian
Le Qian
YMYi-Shuang Meng
Yi-Shuang Meng
XBXiang-Hui Bai
Xiang-Hui Bai
WFWeiwei Feng
Weiwei Feng
A deep learning model that used feature fusion to classify benign and malignant ovarian tumors on gray scale and color Doppler US images achieved a similar performance to clinical expert assessment. Background Deep learning (DL) algorithms could improve the classification of ovarian tumors assessed with multimodal US. Purpose To develop DL algorithms for the automated classification of benign ...

1 June 2021 in Frontiers in Oncology
Verified
Journal Article
Open access
PDF file
LQLe Qian
Le Qian
QDQinwen Du
Qinwen Du
MJMeijiao Jiang
Meijiao Jiang
FYFei Yuan
Fei Yuan
HCHui Chen
Hui Chen
WFWeiwei Feng
Weiwei Feng
Aim: This study aimed to compare different ultrasound-based International Ovarian Tumor Analysis (IOTA) prediction models, namely, the Simple Rules (SRs) the Assessment of Different NEoplasias in the adneXa (ADNEX) models, and the Risk of Malignancy Index (RMI), for the pre-operative diagnosis of adnexal mass. Methods: This single-centre diagnostic accuracy study involved 486 patients. All ultrasound examinations were analyzed and the prediction models were applied. Pathology was the clinical reference standard. The diagnostic performances of prediction models were measured by evaluating receiver-operating characteristic curves, sensitivities, specificities, positive and negative predictive values, positive and negative likelihood ratios, and diagnostic odds ratios. Results: To discriminate benign and malignant tumors, areas under the ROC curves (AUCs) for ADNEX models were 0.94 (95% CI: 0.92–0.96) with CA125 and 0.94 (95% CI: 0.91–0.96) without CA125, which were significantly higher than the AUCs for RMI I-III: 0.87 (95% CI: 0.83–0.90), 0.83 (95% CI: 0.80–0.86), and 0.82 (95% CI: 0.78–0.86), (all P < 0.0001). At a cut-off of 10%, the ADNEX model with CA125 had the highest sensitivity (0.93; 95% CI: 0.87–0.97) compared with the other models. The SRs model achieved a sensitivity of 0.93 (95% CI: 0.86–0.97) and a specificity of 0.86 (95% CI: 0.82–0.89) when inconclusive diagnoses (11.7%) were classified as malignant. Conclusion: ADNEX and SRs models were excellent at characterising adnexal masses which were superior to the RMI in Chinese patients.

6 November 2024 in Journal of Ovarian Research
Verified
Journal Article
Open access
PDF file
XHXin He
Xin He
XBXiang-Hui Bai
Xiang-Hui Bai
HCHui Chen
Hui Chen
WFWeiwei Feng
Weiwei Feng
The study aimed to compare the diagnostic efficacy of the machine learning models with expert subjective assessment (SA) in assessing the malignancy risk of ovarian tumors using transvaginal ultrasound (TVUS). The retrospective single-center diagnostic study included 1555 consecutive patients from January 2019 to May 2021. Using this dataset, Residual Network(ResNet), Densely Connected Convolutional Network(DenseNet), Vision Transformer(ViT), and Swin Transformer models were established and evaluated separately or combined with Cancer antigen 125 (CA 125). The diagnostic performance was then compared with SA. Of the 1555 patients, 76.9% were benign, while 23.1% were malignant (including borderline). When differentiating the malignant from ovarian tumors, the SA had an AUC of 0.97 (95% CI, 0.93–0.99), sensitivity of 87.2%, and specificity of 98.4%. Except for Vision Transformer, other machine learning models had diagnostic performance comparable to that of the expert. The DenseNet model had an AUC of 0.91 (95% CI, 0.86–0.95), sensitivity of 84.6%, and specificity of 95.1%. The ResNet50 model had an AUC of 0.91 (0.85–0.95). The Swin Transformer model had an AUC of 0.92 (0.87–0.96), sensitivity of 87.2%, and specificity of 94.3%. There was a statistically significant difference between the Vision Transformer and SA, and between the Vision Transformer and Swin Transformer models (AUC: 0.87 vs. 0.97, P = 0.01; AUC: 0.87 vs. 0.92, P = 0.04). Adding CA125 did not improve the diagnostic performance of the models in distinguishing benign and malignant ovarian tumors. The deep learning model of TVUS can be used in ovarian cancer evaluation, and its diagnostic performance is comparable to that of expert assessment.

17 April 2021 in Journal of Ovarian Research
Verified
Journal Article
Open access
PDF file
MJMei-Jiao Jiang
Mei-Jiao Jiang
QLQian Le
Qian Le
BYBo-Wen Yang
Bo-Wen Yang
FYFei Yuan
Fei Yuan
HCHui Chen
Hui Chen
Background: Ovarian sex cord stromal tumours (OSCSTs) are rare ovarian tumours and include different histopathologic subtypes. This study aimed to analyse the clinical and sonographic characteristics of different histopathologic OSCST subtypes. Methods: A total of 63 patients with surgically proven OSCSTs were enrolled in this retrospective study to analyse their clinical and sonographic features. Ultrasound examinations and predictive models were performed before surgery. The clinical and sonographic findings were compared according to the type of OSCST based on the histopathological diagnosis. Results: The mean age of 63 patients was 52.17 years (range: 17–78 years). Eighteen patients experienced irregular vaginal bleeding (28.57% 18/63), 7 patients exhibited abnormal body hair (11.11%). 2 patients (3.17%) showed an increased level of CA125, and 25 patients (39.68%, 25/63) showed an increased level of testosterone. Forty-two patients had ovarian thecoma-fibroma groups (OTFGs). Six patients had Sertoli-Leydig cell tumours (S-LCTs), 4 patients had Leydig cell tumours (LCTs), 8 patients had ovarian granulosa cell tumours (OGCTs), 2 patients had ovarian steroid cell tumours, not otherwise specified (OSCTs-NOS), and one patient had sclerosing stromal tumours (SSTs). The mean diameter of the tumour was 47.9 mm (range: 10–258 mm). Forty-seven masses were hypoechoic (74.60%). Twenty-eight masses had posterior echo attenuation, 22 masses exhibited abundant Doppler flow signals (34.92%), and one patient had ascites (1.59%). The diagnostic accuracy of the Simple Rules (SR) and the Assessment of Different NEoplasias in the adneXa (ADNEX) model in distinguishing benign and malignant OSCSTs was 44% (30/63) and 84% (53/63), respectively. The diagnostic accuracy of the SR for OTFGs, S-LCTs & LCTs & OSCTs-NOS, OGCTs, and SSTs was 47.6% (20/42), 16.67% (2/12), 100% (8/8), and 0% (0/1), respectively. The diagnostic accuracy of the ADNEX model for OTFGs, S-LCTs & LCTs & OSCTs-NOS, OGCTs, and SSTs was 93% (31/42), 58.33% (7/12), 75% (6/8), and 100% (1/1), respectively. Conclusions: OSCSTs generally appear as a solid mass on ultrasound. Posterior echo attenuation indicates an OTFG. A solid mass with abundant Doppler flow signals indicates an S-LCT, LCT, OSCT-NOS or OGCT. Current predictive models are not very effective, but symptoms, sonographic features and serum hormones are helpful for diagnosis.

28 September 2022 in Diagnostics
Verified
Journal Article
Open access
PDF file
YYYanhua Yang
Yanhua Yang
JLJing Li
Jing Li
HCHui Chen
Hui Chen
WFWeiwei Feng
Weiwei Feng
Approximately 10% (176 million) of women of reproductive age worldwide suffer from endometriosis, which has a high rate of postoperative recurrence. The objective of this study was to investigate the risk factors of severe endometriosis and establish a preoperative prediction model. A retrospective analysis of a database established between January 2020 and March 2022 including 491 women with a pathology-based endometriosis diagnosis was conducted. Subjects were divided into two groups: the non-severe group (ASRM ≤ 40) and the severe group (ASRM > 40). Age ≥ 40 years, bilateral lesions, pelvic nodules, adenomyosis, APTT, CA125 ≥ 34.5 U/mL, D-dimer ≥ 0.34 mg/L, and maximum cyst diameter ≥ 58 mm were independent correlation factors for severe endometriosis. The logistic regression equation for these factors showed good diagnostic efficiency (AUC = 0.846), which was similar to the model with intraoperative indicators (AUC = 0.865). Patients with severe endometriosis also had a shorter APTT and higher D-dimer and PLT, indicating hypercoagulability. In conclusion, we constructed a simple and feasible formula involving parameters that are preoperatively accessible to predict the severity of endometriosis. This study is of reference value for determining the timing of and alternatives to surgery. At the same time, attention should be paid to the primary prevention of venous thrombosis and cardiovascular metabolic diseases in patients with severe endometriosis.

25 March 2026 in Journal of Ovarian Research
Verified
Journal Article
Open access
JSJie Shen
Jie Shen
YFYun-Han Fang
Yun-Han Fang
JDJia-Yi Ding
Jia-Yi Ding
RJRu-Shi Jiao
Ru-Shi Jiao
YNYa-Ning Niu
Ya-Ning Niu
HCHui Chen
Hui Chen
High-grade serous ovarian cancer (HGSOC) is characterized by high incidence and mortality rates, yet effective detection methods remain limited. In the present study, deep learning models capable of identifying HGSOC while concurrently providing segmentation results of tumor regions and intratumoral solid components were developed and validated. This retrospective single-center diagnostic study included 395 patients (and their combined 1745 ultrasound images) with pathologically confirmed primary malignant ovarian tumors who underwent preoperative ultrasound examination between January 2018 and December 2022. The dataset was split into training (n = 237; 148 HGSOC, 89 Non-HGSOC), validation (n = 79; 49 HGSOC, 30 Non-HGSOC), and test (n = 79; 50 HGSOC, 29 Non-HGSOC) sets at a 6:2:2 ratio. Four models were constructed, all based on Residual Network-18 (ResNet-18) for feature extraction from grayscale and color Doppler ultrasound (US) images. Model A was built solely on image features, while Models B, C, and D further integrated additional features via a multilayer perceptron (MLP) as follows: Model B incorporated cancer antigen 125 (CA125); Model C combined CA125 and human epididymis protein 4 (HE4); and Model D integrated CA125, HE4, and age. All the models employed a dual-branch decoder for classification and segmentation to generate the corresponding outputs. The discriminative ability of the models was assessed using the area under the receiver operating characteristic curve (AUC), while classification performance was evaluated with the sensitivity and specificity. The Dice surface coefficient (DSC) was determined to evaluate segmentation performance. Among the 395 patients, 62.5% had HGSOC, while 37.5% were diagnosed with other pathological types. In the identification of HGSOC, the four models yielded AUC values of 0.72 (Model A), 0.81 (Model B), 0.80 (Model C), and 0.84 (Model D). Compared with Model A, which solely used image features, Model D resulted in the most significant increase in the AUC value (p = 0.03). The sensitivity across the models ranged from 0.66 to 0.86, with Model B yielding the highest value (0.86); the specificity values ranged from 0.63 to 0.93, with Model D yielding the highest value (0.93). The DSC values ranged from 0.70 to 0.86. Deep learning models can be used to distinguish HGSOC and concurrently generate real-time segmentation results for relevant regions. The online version contains supplementary material available at 10.1186/s13048-026-02070-5.