Fan Fu

Fan Fu

Auto-generated profile, based on publicly available information

Fan Fu

Shanghai Jiaotong University School of Medicine

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

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

Representative Publications

View All Publications
24 September 2020 in Nature Communications
Verified
Journal Article
Open access
PDF file
FFFan Fu
Fan Fu
JWJianyong Wei
Jianyong Wei
MZMiao Zhang
Miao Zhang
FYFan Yu
Fan Yu
YXYueting Xiao
Yueting Xiao
JLJie Lu
Jie Lu
The computed tomography angiography (CTA) postprocessing manually recognized by technologists is extremely labor intensive and error prone. We propose an artificial intelligence reconstruction system supported by an optimized physiological anatomical-based 3D convolutional neural network that can automatically achieve CTA reconstruction in healthcare services. This system is trained and tested with 18,766 head and neck CTA scans from 5 tertiary hospitals in China collected between June 2017 and November 2018. The overall reconstruction accuracy of the independent testing dataset is 0.931. It is clinically applicable due to its consistency with manually processed images, which achieves a qualification rate of 92.1%. This system reduces the time consumed from 14.22 ± 3.64 min to 4.94 ± 0.36 min, the number of clicks from 115.87 ± 25.9 to 4 and the labor force from 3 to 1 technologist after five months application. Thus, the system facilitates clinical workflows and provides an opportunity for clinical technologists to improve humanistic patient care.

1 May 2023 in Radiology
Verified
Journal Article
Closed access
FFFan Fu
Fan Fu
YSYi Shan
Yi Shan
GYGuang Yang
Guang Yang
CZChao Zheng
Chao Zheng
MZMiao Zhang
Miao Zhang
JLJie Lu
Jie Lu
Background Studies have rarely investigated stenosis detection from head and neck CT angiography scans because accurate interpretation is time consuming and labor intensive. Purpose To develop an automated convolutional neural network-based method for accurate stenosis detection and plaque classification in head and neck CT angiography images and compare its performance with that of radiologists. Materials and Methods A deep learning (DL) algorithm was constructed and trained with use of head and neck CT angiography images that were collected retrospectively from four tertiary hospitals between March 2020 and July 2021. CT scans were partitioned into training, validation, and independent test sets at a ratio of 7:2:1. An independent test set of CT angiography scans was collected prospectively between October 2021 and December 2021 in one of the four tertiary centers. Stenosis grade categories were as follows: mild stenosis (<50%), moderate stenosis (50%-69%), severe stenosis (70%-99%), and occlusion (100%). The stenosis diagnosis and plaque classification of the algorithm were compared with the ground truth of consensus by two radiologists (with more than 10 years of experience). The performance of the models was analyzed in terms of accuracy, sensitivity, specificity, and areas under the receiver operating characteristic curve. Results There were 3266 patients (mean age ± SD, 62 years ± 12; 2096 men) evaluated. The consistency between radiologists and the DL-assisted algorithm on plaque classification was 85.6% (320 of 374 cases [95% CI: 83.2, 88.6]) on a per-vessel basis. Moreover, the artificial intelligence model assisted in visual assessment, such as increasing confidence in the degree of stenosis. This reduced the time needed for diagnosis and report writing of radiologists from 28.8 minutes ± 5.6 to 12.4 minutes ± 2.0 (P < .001). Conclusion A deep learning algorithm for head and neck CT angiography interpretation accurately determined vessel stenosis and plaque classification and had equivalent diagnostic performance when compared with experienced radiologists. © RSNA, 2023 Supplemental material is available for this article.

5 June 2018 in Acta Radiologica
Verified
Journal Article
Closed access
FFFan Fu
Fan Fu
BSBinbin Sui
Binbin Sui
LLLiping Liu
Liping Liu
YSYaping Su
Yaping Su
SSShengjun Sun
Shengjun Sun
YLYingying Li
Yingying Li
Positive “dynamic spot sign” has been proven to be a potential risk factor for acute intracerebral hemorrhage (ICH) expansion, but local perfusion change has not been quantitatively investigated. To quantitatively evaluate perfusion changes at the ICH area using computed tomography perfusion (CTP) imaging. Fifty-three patients with spontaneous ICH were recruited. Unenhanced computed tomography (NCCT), CTP within 6 h, and follow-up NCCT were performed for 21 patients in the “spot sign”-positive group and 32 patients in the control group. Cerebral perfusion change was quantitatively measured on regional cerebral blood flow/regional cerebral blood volume (rCBF/rCBV) maps. Regions of interest (ROIs) were set at the “spot-sign” region and the whole hematoma area for “spot-sign”-positive cases, and at one of the highest values of three interested areas and the whole hematoma area for the control group. Hematoma expansion was determined by follow-up NCCT. For the “spot-sign”-positive group, the average rCBF (rCBV) values at the “spot-sign” region and the whole hematoma area were 21.34 ± 15.24 mL/min/100 g (21.64 ± 21.48 mL/100g) and 5.78 ± 6.32 mL/min/100 g (6.07 ± 5.45 mL/100g); for the control group, the average rCBF (rCBV) values at the interested area and whole hematoma area were 2.50 ± 1.83 mL/min/100 g (3.13 ± 1.96 mL/100g) and 3.02 ± 1.80 mL/min/100 g (3.40 ± 1.44 mL/100g), respectively. Average rCBF and rCBV values of the “spot-sign” region were significantly different from other regions (P < 0.001; P = 0.004). The average volumes of hematoma expansion in the “spot-sign”-positive and control groups were 25.24 ± 19.38 mL and −0.41 ± 1.34 mL, respectively. The higher perfusion change at ICH on CTP images may reflect the contrast extravasation and be associated with the hematoma expansion.

7 November 2025 in JACC: Cardiovascular Imaging
Verified
Journal Article
Closed access
FFFan Fu
Fan Fu
JZJun Zhu
Jun Zhu
HMHongping Meng
Hongping Meng
XCXiaoyue Chen
Xiaoyue Chen
DLDong Lin
Dong Lin
BLBiao Li
Biao Li

27 January 2025 in Acta Radiologica
Verified
Journal Article
Closed access
XZXuening Zhao
Xuening Zhao
XWXiaochen Wang
Xiaochen Wang
SWSihui Wang
Sihui Wang
LCLingxu Chen
Lingxu Chen
MYMengyuan Yuan
Mengyuan Yuan
SSShengjun Sun
Shengjun Sun
Background: Preoperative identification of T-box pituitary transcription factor 19 (TPIT) lineage silent adenomas in non-functioning pituitary adenomas (NFPAs) is important. Purpose: To compare the clinical, laboratory, and radiological features of the three cell lineages of adenomas in NFPAs and evaluate the diagnostic efficacy of multiple microcysts and clivus invasion on magnetic resonance imaging (MRI) for TPIT lineage adenomas in NFPAs. Material and Methods: A total of 405 patients with NFPA were retrospectively enrolled, including steroidogenic factor 1 (SF-1) lineage adenomas (n = 204), TPIT lineage adenomas (n = 111), and pituitary transcription factor 1 (PIT-1) lineage adenomas (n = 90). The clinical, laboratory, and radiological features of the three lineages adenomas were compared. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of multiple microcysts, clivus invasion, and their combination were calculated to diagnose TPIT lineage adenomas in NFPAs. Results: Among the three lineages of NFPAs, patients with SF-1 lineage were older than those with TPIT and PIT-1 lineages ( P < 0.001). TPIT lineage adenomas were most common in women ( P < 0.001) and had the highest tumor volume ( P < 0.001), and incidence of clivus invasion ( P < 0.001). The multiple microcysts and clivus invasion in the diagnosis of TPIT lineage adenomas in NFPAs had high specificity (88.44% vs. 98.64%) and accuracy (77.28%). Conclusion: The MRI findings of multiple microcysts and clivus invasion can help diagnose TPIT lineage adenomas in NFPAs with high specificity.

17 March 2026 in Journal of the American Heart Association
Verified
Journal Article
Open access
QLQian Li
Qian Li
FFFan Fu
Fan Fu
JCJie Chen
Jie Chen
YZYu Zhang
Yu Zhang
HZHairong Zheng
Hairong Zheng
NZNa Zhang
Na Zhang
Background: The aim of this study was to evaluate an integrated fluorine‐18 fluorodeoxyglucose positron emission tomography/magnetic resonance imaging (PET/MRI) model combining conventional and radiomic features for the noninvasive identification of symptomatic extracranial carotid atherosclerotic plaques. Methods: A total of 200 patients with advanced carotid plaques (78 symptomatic, 122 asymptomatic) underwent hybrid fluorine‐18 fluorodeoxyglucose PET/MRI. Six MRI morphological features (intraplaque hemorrhage, lipid‐rich necrotic core, ruptured fibrous cap, calcification, surface ulceration, and plaque enhancement) and 3 PET metabolic parameters—metabolic uptake, maximum standardized uptake value, and target‐to‐background ratio—were evaluated. A total of 1991 radiomics were extracted from each of the MRI and PET images, respectively. For each modality, 3 types of models were developed: conventional, radiomics, and combined (incorporating conventional and radiomic features). Diagnostic performance was evaluated using multivariate logistic regression and a linear support vector classifier. Results: In MRI‐based analysis, surface ulceration and plaque enhancement emerged as independent predictors of symptomatic plaques. In PET‐based analysis, standardized uptake value and metabolic uptake were identified as significant independent predictors. The PET conventional model outperformed the MRI conventional model in discriminative ability (areas under the curve: training, 0.899 versus 0.755; internal test, 0.926 versus 0.822; temporal validation, 0.878 versus 0.702). The integrated PET/MRI combined model achieved the best performance, with areas under the curve of 0.962 (training), 0.967 (internal test), and 0.926 (temporal validation), significantly outperforming PET‐ and MRI‐based combined models (all P <0.05). Conclusions: The fluorine‐18 fluorodeoxyglucose PET/MRI combined model integrating morphological, metabolic, and radiomic features outperformed other models, supporting its potential as a noninvasive, high‐precision tool for cerebrovascular risk stratification.