1.Combining radiomics and deep learning to predict overall survival in non-small cell lung cancer patients
Yongxin LIU ; Qiusheng WANG ; Huayong JIANG ; Na LU ; Diandian CHEN ; Yanjun YU ; Yanxiang GAO ; Huijuan ZHANG ; Minmin DENG ; Yinglun SUN ; Fuli ZHANG
Chinese Journal of Medical Physics 2025;42(11):1462-1468
Objective To develop a combined model integrating radiomics and 3D deep learning features for improving the predictive efficacy of overall survival in non-small cell lung cancer(NSCLC)patients undergoing radiotherapy,thereby providing a foundation for optimizing individualized radiotherapy strategies.Methods A retrospective analysis was conducted on 522 NSCLC patients from 3 centers.Radiomics features were extracted from the tumor region of interest on radiotherapy planning CT scans,and a 3D-SE-ResNet was constructed to extract deep learning features.Following feature extraction,features were selected via univariate Cox analysis and Lasso-Cox regression,and a combined model was established by fusing the two feature types through principal component analysis.The discriminative ability of the model was evaluated using the concordance index(C-index)and the area under the receiver operating characteristic curve(AUC),while the risk stratification efficacy was verified by Kaplan-Meier survival analysis.Results The predictive performance of deep learning features was significantly superior to that of radiomics features(C-index:0.73 vs 0.65).The combined model achieved the highest predictive performance in the training set,internal test set,and external test set(C-index:0.74,0.69,0.72 respectively),with higher AUC values for predicting 1-year,2-year,and 3-year OS than either single model.Kaplan-Meier analysis showed significant differences in survival between the high-and low-risk groups(Log-rank test,P<0.001),and calibration curves indicated good consistency between predicted and actual survival outcomes.Conclusion The combined model integrating radiomics and 3D deep learning features can accurately predict survival outcomes in NSCLC patients undergoing radiotherapy.The multi-center validation results support its potential application in prognosis stratification for individualized radiotherapy.
2.Combining radiomics and deep learning to predict overall survival in non-small cell lung cancer patients
Yongxin LIU ; Qiusheng WANG ; Huayong JIANG ; Na LU ; Diandian CHEN ; Yanjun YU ; Yanxiang GAO ; Huijuan ZHANG ; Minmin DENG ; Yinglun SUN ; Fuli ZHANG
Chinese Journal of Medical Physics 2025;42(11):1462-1468
Objective To develop a combined model integrating radiomics and 3D deep learning features for improving the predictive efficacy of overall survival in non-small cell lung cancer(NSCLC)patients undergoing radiotherapy,thereby providing a foundation for optimizing individualized radiotherapy strategies.Methods A retrospective analysis was conducted on 522 NSCLC patients from 3 centers.Radiomics features were extracted from the tumor region of interest on radiotherapy planning CT scans,and a 3D-SE-ResNet was constructed to extract deep learning features.Following feature extraction,features were selected via univariate Cox analysis and Lasso-Cox regression,and a combined model was established by fusing the two feature types through principal component analysis.The discriminative ability of the model was evaluated using the concordance index(C-index)and the area under the receiver operating characteristic curve(AUC),while the risk stratification efficacy was verified by Kaplan-Meier survival analysis.Results The predictive performance of deep learning features was significantly superior to that of radiomics features(C-index:0.73 vs 0.65).The combined model achieved the highest predictive performance in the training set,internal test set,and external test set(C-index:0.74,0.69,0.72 respectively),with higher AUC values for predicting 1-year,2-year,and 3-year OS than either single model.Kaplan-Meier analysis showed significant differences in survival between the high-and low-risk groups(Log-rank test,P<0.001),and calibration curves indicated good consistency between predicted and actual survival outcomes.Conclusion The combined model integrating radiomics and 3D deep learning features can accurately predict survival outcomes in NSCLC patients undergoing radiotherapy.The multi-center validation results support its potential application in prognosis stratification for individualized radiotherapy.
3.Correlation between clinicopathologic features and BRAF V600E gene mutation abundance in papillary thyroid carcinoma
Yong ZHAO ; Bo GAO ; Zhuomiaoyu CHEN ; Qiusheng WANG
Chinese Journal of General Surgery 2021;36(9):684-686
Objective:To investigate the correlation between cervical lymph node metastasis,extra thyroidal extension and other clinicopathologic features and BRAF V600E gene mutation abundance in papillary thyroid carcinoma. Methods:The clinicopathologic data of 123 PTC patients admitted from Jan 2018 to Oct 2020 in Peking University People's Hospital were analyzed retrospectively. All the PTC cases were examined by next generation sequencing technology. BRAF V600E gene mutation abundances were detected and analyzed with different clinicopathologic features, and differences between groups were compared. Results:There were statistical differences compared to control group in BRAF V600E gene mutation abundances in lymph node metastasized group and extra thyroidal extended group(all P<0.05). BRAF V600E gene mutation abundances were positively related with severity of extra thyroidal extension( P<0.05). Conclusion:BRAF V600E gene mutation abundance was related with cervical lymph node metastasis and extra thyroidal extension in PTC patients.
4.Ultrasonic classification and evolution of thyroid shrinking nodules
Yifan ZHAO ; Penglin ZOU ; Yuchen TAO ; Chao JIA ; Gang LI ; Yunhua LI ; Feng GAO ; Yubiao JIN ; Lianfang DU ; Qiusheng SHI
Chinese Journal of Ultrasonography 2021;30(12):1046-1051
Objective:To investigate the ultrasonographic features and evolution of thyroid shrinking nodules in order to improve the differential diagnosis and management strategy and avoid unnecessary biopsy.Methods:A total of 245 patients with old bleeding of benign thyroid nodules diagnosed via fine needle aspiration cytology (FNAC) from May 2015 to July 2020 in the southern part of the Shanghai General Hospital Affiliated to the Medical College of Shanghai Jiaotong University, including 263 nodules. The sonographic parameters such as size, shape, aspect ratio, echo, edge, boundary, periphery, halo, calcification, posterior echo enhancement and posterior acoustic attenuation of nodules were analyzed retrospectively, and the ultrasonic appearances of nodules were classified; 41 nodules from 40 patients, who did ultrasound examinations more than twice and had both complete ultrasonographic data, were compared and then classified in order to explore the evolution rules of the nodules.Results:Two hundred and sixty-three thyroid shrinking nodules in 245 cases were classified into 4 types: "cystic wall shrinkage sign" type(71.48%, 188/263), "carcinoma-like" type (22.05%, 58/263), "inflammation-like" type(3.04%, 8/263) and "undefined atypical solid nodules" type(3.42%, 9/263). Of the 41 consecutive follow-up nodules, 37 cases showed obvious cystic wall shrinkage sign, and 7 of them developed into "carcinoma-like" type and 6 cases into "inflammation-like" type.Conclusions:The ultrasonographic appearance of thyroid shrinking nodules is a dynamic process, which can be divided into 4 types: "cystic wall shrinkage sign" , "carcinoma-like" , "inflammation-like" and "undefined atypical solid nodules" types. The "cystic wall shrinkage sign" type is typical and common. The "cystic wall shrinkage sign" type can develope into the "carcinoma-like" type or the "inflammation-like" type.
5.The contrast enhanced ultrasound qualitative diagnosis of benign and malignant in BI-RADS 4 breast nodules with different sizes
Chao JIA ; Lianfang DU ; Qiusheng SHI ; Fan LI ; Lifang JIN ; Feng GAO
Chinese Journal of Ultrasonography 2020;29(4):343-348
Objective:To evaluate the qualitative diagnosis of contrast enhanced ultrasound (CEUS) in different sizes′ benign and malignant breast nodules (BNs) of BI-RADS 4.Methods:A total of 506 BNs in 467 patients from Shanghai General Hospital between October 2017 and May 2019 classified as BI-RADS 4 were divided into large size group (>20 mm, n=198) and small size group (≤20 mm, n=308) according to their largest diameter lines. The CEUS characteristics of benign and malignant BNs in two groups were analyzed by Chi-square test. The CEUS characteristics of BNs were assigned scores for diagnosis of benign and malignant BNs with different sizes, the diagnostic efficacy of these scores was evaluated by ROC curve. Results:There were statistic differences in the degree, size change of the enhanced lesions, enhancement time of lesions compared with surrounding tissues, radial enhancement or not, homogeneity of enhancement, with or without penetrating of perfusion between benign and malignant BNs in both groups (all P<0.001). In small size group, with the critical value of 2.5 points, area under curve(AUC) of ROC curve was 0.793, and the sensitivity, specificity, accuracy of CEUS was 82.67%, 60.22% and 69.48%, respectively. In large size group, with the critical value of 3.5 points, AUC of ROC curve was 0.901, and the sensitivity, specificity, accuracy of CEUS was 85.42%, 80.00% and 82.65%, respectively. Conclusions:CEUS has great significance in qualitative diagnosis of benign and malignant in BI-RADS 4 BNs with different sizes. The biopsy could be avoided in BNs greater than 20 mm, while biopsy is still needed in BNs less than or equal to 20 mm.
6.Analysis of cancer incidence and mortality in registration areas of Shanxi Province in 2014
Zhaohui MA ; Qiusheng GAO ; Ling CAO ; Xinzheng WANG ; Xuerong GUO ; Xinchen WANG ; Fang SU ; Nan QIAO ; Yuan WANG ; Ruifeng ZHANG ; Yongzhen ZHANG
Cancer Research and Clinic 2020;32(3):186-191
Objective:To explore the cancer incidence and mortality in registration areas of Shanxi Province in 2014.Methods:The data of 12 cancer registration areas of Shanxi Province in 2014 were taken to analyze the characteristics of cancer incidence and mortality for patients with different age and gender in different areas. And then the results were compared with the malignant cancer incidence and mortality in the nationwide.Results:There were 11 703 new cases, including 6 559 males and 5 144 females in registration areas of Shanxi Province in 2014, and the incidence rate was 221.21/10 5, while the age-standardized incidence rate of Chinese population and world population was 163.91/10 5 and 163.25/10 5, respectively. The cancer incidence rate in urban areas was 247.02/10 5 and the age-standardized incidence rate of Chinese population was 171.35/10 5. In rural areas, the cancer incidence rate was 205.98/10 5 and the age-standardized incidence rate of Chinese population was 159.03/10 5. The common cancer sites were stomach, lung, esophagus, liver and colorectum for males. And breast, cervix, lung, esophagus, stomach were the common cancer sites for females. There were 7 283 malignant death cases, including 4 548 males and 2 735 females. The crude cancer mortality rate was 137.66/10 5, and the age-standardized mortality rate of Chinese population was 99.67/10 5 and world population was 100.11/10 5. The crude cancer mortality rate in urban areas was 141.03/10 5 and the age-standardized incidence rate of Chinese population was 92.84/10 5. In rural areas, the crude cancer mortality rate was 135.68/10 5 and the age-standardized mortality rate of Chinese population was 103.69/10 5. Male common malignant tumor deaths included lung cancer, gastric cancer, liver cancer, esophageal cancer and colorectal cancer, while lung, stomach, liver, esophagus and cervix were the common cancer death sites for females. Conclusions:The incidence and mortality of malignant tumors in registration areas in Shanxi Province are mainly lung cancer, upper gastrointestinal cancer and cervix uteri cancer. The incidence rates of stomach cancer and cervical cancer are high.
7.Clinical practice and development in minimally invasive functional surgery of digestive tract
Qiusheng WANG ; Zhuomiaoyu CHEN ; Bo GAO ; Yuan WANG ; Xueguang ZHU
Chinese Journal of Digestive Surgery 2020;19(5):486-490
Digestival functional surgery refers to surgical treatment for patients with functional diseases of the digestive tract. With the continuous development of minimally invasive surgery, most of surgical procedures for digestive functional diseases have been conducted by minimal invasive approach. Based on decades of clinical practice, the authors focus on minimally invasive surgeries for a group of digestive functional diseases such as achalasia of cardia, gastroesophageal reflux disease, volvulus of stomach, gastroptosis, dolichocolon, refractory constipation, rectocele and rectal prolapse, and make ageneral summary on current status and prospects of its clinical application combined with literatures.
8.Clinical comparative analysis of domestic 16-row and imported 8-row mobile CT head scans
Zhiqiang ZHANG ; Quanle ZHENG ; Haifeng WANG ; Lei YANG ; Fei LI ; Boyun DING ; Li ZHANG ; Shunyi ZHOU ; Yaxin JING ; Zhenfang WANG ; Fei GAO ; Qiusheng DAI ; Ruxiang XU
Chinese Journal of Neuromedicine 2020;19(4):376-380
Objective:To compare the efficacy and safety of domestic 16-row and imported 8-row mobile CT in clinics.Methods:A total of 1469 patients accepted domestic 16-row mobile CT head scans (1604 times) from March 2017 to August 2018 in Bayi Brain Hospital Affiliated to 7 th Medical Center of General Hospital of People's Liberation Army and Langfang Aidebao Hospital; and 15510 patients accepted imported 8-row mobile CT head scans (24994 times) from January 2016 to August 2018 in Bayi Brain Hospital Affiliated to 7 th Medical Center of General Hospital of People's Liberation Army. All patients underwent horizontal plain and enhanced head scans, cerebral CT angiography (CTA), and helical 3D imaging; and the imaging quality, operating power consumption, computed tomography dose index volume (CTDIvol) and stability within scanning volume ranges under different scanning modes of the two CT scans were compared. Results:(1) Imaging quality: the horizontal scanning of domestic 16-row mobile CT could clearly display low-density tissues such as the eyeball, optic nerve, brain stem, sulcus and cerebral gyrus; the imaging quality of both CT scans in patients with traumatic subdural hematoma and ischemic stroke completely met the clinical diagnosis and treatment standards. (2) Operating power consumption: the per-hour operating power consumption of domestic 16-row mobile CT ([0.286±0.018] kW·h) was obviously lower than that of imported 8-row mobile CT ([0.485±0.028] kW·h). (3) Radiological hazard: the CTDIvol of the horizontal scanning volume range in domestic 16-row mobile CT ([36.270±0.281] mGy) was significantly lower than that in the imported 8-row mobile CT ([82.520±0.441] mGy, P<0.05); the CTDIvol of enhanced axis scan volume range in the domestic 16-row mobile CT ([36.270±0.335] mGy) was significantly lower than that in the imported 8-row mobile CT ([70.728±0.424] mGy, P<0.05); the CTDIvol in the volume of CTA imaging of domestic 16-row mobile CT ([20.600±0.087] mGy) was significantly lower than that in the imported 8-row mobile CT ([29.300±0.335] mGy, P<0.05). The domestic 16-row mobile CT was designed with shock absorbers and guides; domestic 16-row mobile CT had small load, a low center of gravity, and good stability as compared with imported 8-row mobile CT. Conclusion:In terms of head scanning applications, the imaging quality of domestic 16-row mobile CT and imported 8-row mobile CT is in full compliance with clinical diagnostic standards, but the energy consumption and radiation risk of domestic 16-row mobile CT is significantly lower than imported 8-row mobile CT, enjoying good stability as compared with imported 8-row mobile CT.
9.Novel 16-slice mobile CT head scan in 391 patients
Zhiqiang ZHANG ; Qiusheng DAI ; Fei LI ; Fei GAO ; Yaxin JING ; Ruonan BI ; Zhenfang WANG ; Rongrong CHEN ; Lida XU ; Daiqiang YIN ; Zhonghua YU ; Qiang ZHANG ; Ruxiang XU
Chinese Journal of Neuromedicine 2018;17(2):184-188
Objective To discuss the application of self-developed novel 16-slice mobile CT head scan.Methods A total of 391 patients were performed 16-slice mobile CT scan:145 were scanned in the emergency department,156 in the neurosurgical ICU,55 in the operated room,and 35 in the ambulance vehicle.Sixty-eight patients were with brain injury,122 were with cerebral hemorrhage,120 were with cerebral infarction,59 were with brain tumors,and 22 were with hemifacial spasm.Thirty-five patients were randomly selected from 391 patients and 8-slice mobile CT head scan was performed on them,which included 12 with brain injury,6 with cerebral hemorrhage,12 with cerebral infarction,3 with brain tumors and 2 with hemifacial spasm.The resolution,imaging quality,radiation doses,power consumption and performance stability of novel 16-slice mobile CT and 8-slice mobile CT head scan were compared.Results The resolution line pairs of brain tissues were 91 p/cm by 16-slice mobile CT and 71 p/cm by 8-slice mobile CT,respectively.The imaging quality of the two kinds of mobile CT head scans was high level to the clinic diagnostic criteria.The radiation dose of 16-slice mobile CT were 40.43 mGy,which decreased by 51.01% as compared with that of 8-slice mobile CT (82.52 mGy).The personal power consumption of 16-silce mobile CT (0.29 kW· h) decreased by 38.30% as compared with those of 8-layer mobile CT (0.47 kW· h).The 16-slice mobile CT kept regularly,while 8-slicer mobile CT stopped to work twice during clinical trial.Conclusion The 16-slice mobile CT scan has high resolution,fine imaging quality,low radiation dose,small power consumption and stable working performance.
10.Locally advanced pancreatic cancer:treatment with Nanoknife percutaneously under guidance of ultrasound and effectiveness evaluated by contrast enhanced ultrasonography
Qiusheng SHI ; Jinglong ZHAO ; Lianfang DU ; Zhiyong WAN ; Xin LI ; Long LIU ; Gang LI ; Kang GAO ; Yanfeng ZHANG
Chinese Journal of Ultrasonography 2017;26(10):877-881
Objective To explore the treatment of locally advanced pancreatic cancer(LAPC)by Nanoknife,and the demonstrations on contrast enhanced ultrasonography(CEUS)after the procedures. Methods Four patients were enrolled.Each of them was percutaneously inserted 3-5 Nanoknife electrodes to ablate the lesions with 2-3 sessions under the guidance of ultrasound.After the procedures,CEUS, contrast enhanced CT,contrast enhanced MRI or computed tomography angiography (CTA) were performed to assess the efficacy.Results All patients uneventfully were able to tolerate the high-voltage Nanoknife procedures.However,the endurable pain or discomfort in abdomen was complained.For one patient with obstructive jaundice and mild liver dysfunction,pancreatic amylase in serum increased slightly at the first day after treatment.Furthermore,C response protein in serum elevated,and biliary tract infection happened at the third day.Moreover,thromboses in portal and splenic vein were found at the seventh day.The survival time for all cases were 3-8 months.According to the findings of CEUS,a well-defined no contrast enhancement was detected at ablative zone for all the patients within one week.Yet, contrast enhancement was sporadically detected at the periphery in 3 patients 1-2 months late.Meanwhile, on contrast enhanced CT or MRI,an area of absent enhancement was also depicted at the ablation zone with heterogeneously decreased enhancement at periphery of the lesion.Base on CTA findings,local stenosis of splenic vein and superior mesenteric vein with varices of collateral vascularity were found in one patient at the fiftieth day,and superior mesenteric vein obstruction accompanied by varices of collateral vascularity was also found in another patient at the seventieth day.Conclusions For patients with LAPC,Nanoknife treatment is a safe and convenient procedure that can be percutaneously completed under the guidance of ultrasound.Along with enhanced CT,MRI or CTA,CEUS is a promising way to evaluate the efficacy of this procedure.

Result Analysis
Print
Save
E-mail