Journal of Shandong University (Health Sciences)

Research progress in the application of artificial intelligence in myocardial imaging

Pei NIE,Ximing WANG

Journal of Shandong University (Health Sciences). 2023, 61(12): 1-6. doi:10.6040/j.issn.1671-7554.0.2023.0773

Abstract ( 371 )

HTML ( 30 )

PDF (1451KB) ( 440 )

Save

References | Related Articles | Metrics

Recently, artificial intelligence (AI) has shown great potential in myocardial imaging. AI algorithms achieve automatic segmentation and measurement of myocardial images thus optimizing the workflow. The quantitative features which characterized the pathological changes of myocardium were extracted through radiomics and deep learning techniques. These features may facilitate precise diagnosis and outcome prediction of ischemic and non-ischemic cardiomyopathies. In this review, we will introduce the research progress of AI in myocardial imaging from several aspects: AI-assisted image analysis, diagnosis and outcome evaluation of cardiomyopathies. The limitations of AI in myocardial imaging will also be discussed. We hope this review may provide references for further clinical application research of AI in myocardial imaging.

Innovation and challenge of imaging artificial intelligence in medical field

Ziliang XU,Minwen ZHENG

Journal of Shandong University (Health Sciences). 2023, 61(12): 7-12, 20. doi:10.6040/j.issn.1671-7554.0.2023.0705

Abstract ( 668 )

HTML ( 58 )

PDF (1306KB) ( 469 )

Save

References | Related Articles | Metrics

With the development of science and technology, artificial intelligence (AI) has been applied in the medical imaging field gradually. However, the AI still faces many challenges. In this paper, the imaging application progress of AI in medical field will be reviewed from the aspect of tissue segmentation, auxiliary diagnosis of disease and clinical research, respectively, and the problems in them will also be pointed out. Finally, the challenges of imaging AI in medical field will be discussed.

Research advances of artificial intelligence-based medical imaging in the screening, diagnosis and prediction of pneumonia

Xiao LI,Zhiyuan SUN,Longjiang ZHANG

Journal of Shandong University (Health Sciences). 2023, 61(12): 13-20. doi:10.6040/j.issn.1671-7554.0.2023.0803

Abstract ( 422 )

HTML ( 7 )

PDF (1839KB) ( 527 )

Save

References | Related Articles | Metrics

Pneumonia has become the third leading cause of death in the world after ischemic heart disease and cerebrovascular disease, and is a major public health problem that seriously threatens human health. Early, rapid and accurate etiological diagnosis and risk prediction are the primary tasks in the diagnosis, treatment and prevention of pneumonia. However, due to the heavy workload of radiologists and overlapping image manifestations of different types of pneumonia, timely, rapid, and accurate diagnosis and prediction is rather challenging. The rapid development of artificial intelligence (AI) in the imaging field offers hope for solving these clinical challenges. This paper reviews the latest research results of AI in the diagnosis of pneumonia, aiming to discuss the latest progress of AI system in the field of screening, diagnosis and prediction of pneumonia, and provide prospects in the field of pneumonia, so as to provide references for promoting reasonable optimization of clinical management of pneumonia patients in China and improving the level of intelligent diagnosis and treatment of pneumonia.

Research status and development prospect of deep learning in medical imaging

Bingjie LIN,Meiyun WANG

Journal of Shandong University (Health Sciences). 2023, 61(12): 21-29. doi:10.6040/j.issn.1671-7554.0.2023.0774

Abstract ( 1377 )

HTML ( 66 )

PDF (1607KB) ( 750 )

Save

References | Related Articles | Metrics

Precision medicine, imaging first; precision imaging, technology first. In recent years, with the rapid development of artificial intelligence, deep learning, as an important branch, has been widely used in many fields such as signal processing, computer vision and natural language processing, etc., among which medical image data segmentation, disease detection and prognosis prediction based on deep learning have become the hot spots of many scholars' research. In this paper, we will briefly outline the current status of deep learning application in the main technical fields of medical imaging, and analyze the challenges and development prospects of its clinical application in medical imaging, aiming to provide reference for the transformation of deep learning algorithms in the clinic.

Advances in the application of artificial intelligence in coronary computed tomography angiography

Guyue ZHAO,Jin SHANG,Yang HOU

Journal of Shandong University (Health Sciences). 2023, 61(12): 30-35. doi:10.6040/j.issn.1671-7554.0.2023.0795

Abstract ( 539 )

HTML ( 18 )

PDF (1479KB) ( 807 )

Save

References | Related Articles | Metrics

With the increasingly widespread application of artificial intelligence in the field of medical imaging, its application in coronary artery CT angiography has shown great potential, which helps to improve image quality, optimize post-processing processes, assist disease detection, evaluate functional status, analyse prognosis, and other aspects. Meanwhile, there arise some problems, and the full inspection process should be further optimized to enhance its practicality and efficiency. This article reviews the research progress, existing problems, and future development of artificial intelligence in coronary artery CT angiography.

Identification of carotid high-risk plaques by non-alcoholic fatty liver disease based on CTA

XU Tianqi, CHANG Na, ZHANG Shuai, LI Sha, JIAO Bingxuan, YU Xinxin, WANG Ximing

Journal of Shandong University (Health Sciences). 2023, 61(12): 36-43. doi:10.6040/j.issn.1671-7554.0.2023.0389

Abstract ( 310 )

PDF (3117KB) ( 114 )

Save

References | Related Articles | Metrics

Objective To investigate the correlation between non-alcoholic fatty liver disease(NAFLD)and carotid artery high-risk plaques by computed tomographic angiography(CTA). Methods A total of 203 patients(119 males and 84 females)aged 61.4±12.5 years who underwent carotid CTA for suspected cardiovascular diseases during Jan. and Oct. 2022 were involved. NAFLD was defined as hepatic steatosis without clinical liver disease, cirrhosis or alcohol abuse as demonstrated by non-enhanced CT(hepatosplenic CT value ratio <1). The patients were divided into NAFLD(case group)and non-NAFLD group(control group)according to the abdominal CT findings. The general clinical data, characteristics of high-risk plaques(positive remodeling, ulceration, lipid plaques, "napkin ring" sign), nature of plaques, and degree of carotid stenosis were compared between the two groups. The correlation between NAFLD and presence of high-risk plaques in carotid arteries was analyzed with binary Logistic regression. Results Patients in the NAFLD group had a significantly higher prevalence of diabetes mellitus(P=0.047), levels of alanine aminotransferase(P=0.015), glutamyl transpeptidase(P=0.029), triglycerides(P=0.004), and high-density lipoprotein(P<0.001)than the non-NAFLD group; the presence of high-risk plaques(P=0.001), non-calcified plaques(P=0.015)and incidence of partially calcified plaques(P=0.008)in the common carotid artery and extracranial segment of the internal carotid artery, and non-calcified plaques(P=0.022)and incidence of partially calcified plaques(P=0.002)in the intracranial segment of the internal carotid artery were also higher than the non-NAFLD group. In multivariate Logistic analysis, NAFLD remained independently associated with high-risk plaques after adjusments of age, sex, history of diabetes, hypertension, GGT, uric acid, total carotid artery, calcified plaques in the internal carotid artery, noncalcified plaques, and partially calcified plaques and degree of stenosis(OR: 4.85; 95%CI: 1.77-13.28; P=0.002). Conclusion NAFLD is independently associated with carotid high-risk plaques detected by CTA.

Prediction of isocitrate dehydrogenase mutation in glioma with different radiomic models based on susceptibility-weighted imaging

ZHU Zhengyang, SHEN Jingfei, CHEN Sixuan, YE Meiping, YANG Huiquan, ZHOU Jianan, LIANG Xue, ZHANG Xin, ZHANG Bing

Journal of Shandong University (Health Sciences). 2023, 61(12): 44-50. doi:10.6040/j.issn.1671-7554.0.2023.0770

Abstract ( 332 )

PDF (3739KB) ( 191 )

Save

References | Related Articles | Metrics

Objective To explore the efficacy of different radiomic models based on susceptibility-weighted imaging(SWI)sequences in predicting the isocitrate dehydrogenase(IDH)mutation status of glioma before operation. Methods A retrospective analysis was conducted on the imaging data of 493 adult patients with confirmed diffuse glioma from UCSF-PDGM, including 393 wild-type cases and 100 mutation cases. The patients were divided into training(n=395)and testing(n=98)sets in an 8∶2 ratio. Radiomic features were extracted from SWI sequences based on two regions of interests(ROIs): tumor core(TC)and whole tumor(WT). A total of 1 316 radiomic features were standardized using the z-score method, and dimensionality reduction was performed using principal component analysis(PCA). Feature selection was conducted via variance analysis. Support vector machine(SVM), linear discriminant analysis(LDA), auto-encoder(AE), Logistic regression(LR), Logistic regression via Lasso(LR-Lasso), and native bayes(NB)models were constructed. Receiver operating characteristic(ROC)curves were drawn to assess the accuracy, sensitivity, and specificity. The models performance was evaluated using the area under the curve(AUC)on the testing set. Results A total of 20 radiomic features were selected to establish the radiomic model. The AUC and accuracy of the SVM model were 0.841 and 0.755; the AUC and accuracy of the LDA model were 0.800 and 0.735; the AUC and accuracy of the AE model were 0.743 and 0.745; the AUC and accuracy of the LR model were 0.842 and 0.725; the AUC and accuracy of the LR-LASSO model were 0.880 and 0.857; the AUC and accuracy of the NB model were 0.806 and 0.725. Conclusion SWI-based radiomic features hold certain value in predicting IDH gene mutations in glioma. The LR-Lasso model demonstrates the best predictive performance among the models.

Differentiation of benign and malignant skin focal protuberant nodules using microscope high-resolution MRI

JIN Xinjuan, CAI Daxing, FAN Jinlei, DENG Zhanhao, LI Nan, YU Dexin, LI Anning

Journal of Shandong University (Health Sciences). 2023, 61(12): 51-61. doi:10.6040/j.issn.1671-7554.0.2023.0654

Abstract ( 314 )

PDF (5782KB) ( 159 )

Save

References | Related Articles | Metrics

Objective To explore the value of microscope high-resolution magnetic resonance imaging(MHR-MRI)in the differentiation of benign and malignant skin focal protuberant nodules. Methods A total of 45 patients with skin focal protuberant nodules admitted to Qilu Hospital of Shandong University during April 2020 and April 2023 and examined with preoperative MHR-MRI were prospectively analyzed. Based on pathological results, they were divided into benign group(n=19)and malignant group(n=26). The MHR-MRI characteristics of the lesions in the two groups were assessed. Differences between the benign and malignant groups, and among basal-cell carcinoma, protuberant skin fibrosarcoma and squamous cell carcinoma were compared. The sensitivity, specificity, Youden index and the area under the operating characteristic curve(AUC)and diagnostic efficiency of MHR-MRI were calculated. Results Lesions in the malignant group were more commonly characterized by sharp-angle-wedge sign(P<0.001), involvement of the epidermis and dermis(P<0.001), unclear boundaries(P=0.003), and the presence of surrounding vessels(P=0.003). The Youden index for a combined imaging parameters(sharp-angle-wedge sign, invasion of the epidermis and dermis, poorly defined borders, high signal intensity on T₂WI, and combination of peripheral vessels)was 0.909(sensitivity 0.962, specificity 0.947)with an AUC of 0.995. The basal-cell carcinoma and protuberant skin fibrosarcoma were more often characterized by a wide base of the focus than the squamous cell carcinoma on MHR-MRI(P=0.021). The protuberant skin fibrosarcoma and squamous cell carcinoma were more often associated with sharp-angle-wedge sign than basal-cell carcinoma(P=0.001). Conclusion MHR-MRI is helpful in the differentiation of benign and malignant skin focal protuberant nodules and common malignant tumors.

Radiomics of enhanced CT in predicting the response to platinum-based chemotherapy for ovarian cancer

JIAO Guangli, SHI Zixin, CHEN Rong, SONG Yabo, YANG Fei, CUI Shujun

Journal of Shandong University (Health Sciences). 2023, 61(12): 62-69. doi:10.6040/j.issn.1671-7554.0.2023.0738

Abstract ( 279 )

PDF (5627KB) ( 116 )

Save

References | Related Articles | Metrics

Objective To construct and compare the performance of clinical model, radiomics model and combined model in predicting the patients response to platinum-based chemotherapy for ovarian cancer(OC). Methods The complete data of 94 OC patients confirmed by pathology were retrospectively analyzed. The 2D-ROI and 3D-ROI of tumors were sketched along the tumor contour on the venous phase images, and then radiomics features were extracted and dimensionality reduction filtering was performed. After the radscore was calculated, 2D and 3D radiomics models were constructed, and their prediction efficiency was compared. Meaningful clinical features were retained to build a clinical model, and then radscore was added to construct a combined model. The predictive efficacy of clinical, radiomics and combined models was evaluated with the receiver operating characteristic(ROC)curve, calibration curve, and decision analysis curve(DCA). Results The 3D model showed higher predictive efficiency than 2D model, with the area under ROC curve(AUC)being 0.766 and 0.677, respectively. Compared with the clinical model(age, RT)and radiomics model, the combined model showed the best predictive efficacy, with the AUC being 0.708, 0.766, and 0.827, respectively(P=0.010). Conclusion Modeling based on enhanced CT radiomics and clinical features can predict OC patients response to platinum-based chemotherapy, and the combined model has high predictive efficacy.

Diagnostic value of CT radiomics in lung cancer associated with cystic airspaces

AI Jiangshan, GAO Huijiang, AI Shiwen, LI Hengyan, SHI Guodong, WEI Yucheng

Journal of Shandong University (Health Sciences). 2023, 61(12): 70-77. doi:10.6040/j.issn.1671-7554.0.2023.0748

Abstract ( 283 )

PDF (3377KB) ( 168 )

Save

References | Related Articles | Metrics

Objective To develop a prediction model based on clinical features and preoperative computed tomography(CT), which would serve as a non-invasive method for diagnosing lung cancers associated with cystic airspaces(LCCA). Methods Clinical data of patients undergoing surgery during Jan. 2020 and Dec. 2021 were retrospectively reviewed. A total of 296 patients were enrolled and divided into LCCA group and benign group. Radiomics features were extracted from lesions on preoperative CT images, and clinicopathological characteristics were carefully analyzed. The aforementioned features were used to construct a diagnostic model with random forest algorithm following feature reduction. The 214 patients from the Affiliated Hospital of Qingdao University underwent training and internal verification(D1). The remaining 82 patients from the Affiliated Hospital of Jining Medical College served as an independent external validation cohort(D2). The classification ability of the model was evaluated with receiver operating characteristic(ROC)curve, calibration curve and decision curve analysis(DCA). Results There were no other significant differences between the two groups in clinical features except that the LCCA group averaged older than the benign group(60.12±9.95 vs. 50.86±13.66, P<0.001). After radiomic features were screened, the following features were obtained, including shape flatness, robust mean absolute deviation and difference variance. The model demonstrated superior performance in discriminating LCCA in both D1(AUC 0.97)and D2(AUC 0.74), with accuracy, sensitivity and specificity on D1 being 0.92, 0.85 and 0.99, respectively. Conclusion We have developed and externally validated a practical model that can accurately differentiate LCCA from benign lesions. This innovative approach may serve as a new non-invasive diagnostic tool for LCCA.

Texture analysis based on CT to predict the short-term outcomes of acute pulmonary embolism

CHEN Rong, YANG Yue, YANG Zhixiang, SU Yaying, PANG Zhiying, WANG Dawei, CUI Shujun, YANG Fei

Journal of Shandong University (Health Sciences). 2023, 61(12): 78-85. doi:10.6040/j.issn.1671-7554.0.2023.0659

Abstract ( 279 )

PDF (4518KB) ( 172 )

Save

References | Related Articles | Metrics

Objective To explore the value of quantitative prediction of the short-term outcomes in patients with acute pulmonary embolism(APE)based on CT texture analysis. Methods The CT pulmonary angiography(CTPA)images and clinical data of 79 APE patients were retrospectively collected. The patients were divided into good prognosis group(n=56)and poor prognosis group(n=23). After the ratio of the maximum diameter of the right ventricle to the left ventricle of the heart(RV/LV)and Qanadli index were recorded, and texture features were extracted, univariate and multivariate Logistic regression was conducted with R software. The receiver operating characteristic(ROC)curve, calibration curve and clinical decision curve were drawn to evaluate the efficacy of texture feature parameters, RV/LV ratio and Qanadli index in predicting poor prognosis of APE patients. Results RV/LV≥1.0, Qanadli index, and texture characterization parameters GLSZM-GLN were independent predictors of short-term poor prognosis in APE patients, with the area under the ROC curve(AUC)being 0.826(0.725-0.902), 0.922(0.839-0.970), and 0.867(0.772-0.933), respectively. The AUC of the combined curve was 0.958(0.887-0.990). The clinical decision curve showed the combined curve had the best efficacy in predicting poor prognosis of APE and the highest clinical application value. Conclusion Texture analysis based on CT thrombi, RV/LV≥1.0 and Qanadli index are able to assess the short-term poor prognosis of APE patients, which have certain clinical application value.

Construction of a risk prediction model for ventilator weaning in mechanically ventilated patients in ICU

WANG Jianhua, SUN Shuqing, ZHANG Xiaodong, YANG Xiaoxiao, WANG Youjian, LU Jinbao, LI Zanwu

Journal of Shandong University (Health Sciences). 2023, 61(12): 86-93. doi:10.6040/j.issn.1671-7554.0.2023.0796

Abstract ( 379 )

PDF (1498KB) ( 235 )

Save

References | Related Articles | Metrics

Objective To investigate the risk factors of weaning failure in mechanically ventilated patients in the intensive care unit(ICU), develop a risk prediction model, and perform internal validation to assess the predictive performance. Methods A total of 546 cases of mechanically ventilated patients were collected during Jan. 2020 and Jan. 2022 with convenience sampling method from the ICU of Weifang Peoples Hospital, including 358 collected from Jan. 2020 and Jan. 2021 in the modeling group, and 188 collected from Feb. 2021 to Jan. 2022 in the model testing group. The risk factors of weaning failure were analyzed, based on which a prediction model was developed. The goodness of fit was assessed with Hosmer-Lemeshow test. The prediction performance was evaluated with receiver operating characteristic(ROC)curve, and the area under the ROC curve(AUC). Results The risk factors included mechanical ventilation time(OR=0.993), diaphragm mobility(OR=3.886), diaphragm thickness variability(OR=65.917), shallow fast respiratory index(RSBI, OR=0.960), and inferior vena cava variability(OR=1.176). The model equation was as follows: Z=-0.007×mechanical ventilation time+1.357×diaphragm mobility +4.188×diaphragm thickness variability -0.041×shallow fast respiratory index+ 0.162×inferior vena cava variability -3.183. The AUC was 0.926, with a specificity of 0.961 and a sensitivity of 0.887. Conclusion The risk prediction model demonstrates good performance, which can provide reference for the development and implementation of future interventions.

Efficacy and safety of neuroendoscopy and borehole drainage in the treatment of massive chronic subdural hematoma

QIAN Ming, JIANG Lei, WANG Xuejian, ZHAO Wei, WANG Zhifeng, ZHANG Yi

Journal of Shandong University (Health Sciences). 2023, 61(12): 94-99. doi:10.6040/j.issn.1671-7554.0.2023.0490

Abstract ( 270 )

PDF (2575KB) ( 144 )

Save

References | Related Articles | Metrics

Objective To investigate the efficacy and safety of neuroendoscopy and borehole drainage in the treatment of chronic subdural hematoma(CSDH). Methods Clinical data of 117 CSDH patients hospitalized in our department during Jan. 2020 and Dec. 2022 were retrospectively analyzed. The patients were divided into two groups according to the surgical approaches, 52 cases in the neuroendoscopic group and 65 cases in the borehole drainage group. The demography, CT imaging, clinical efficacy, postoperative complications and recurrence rate were compared between the two groups. Results Compared with the borehole drainage group, the neuroendoscopic group had a shorter drainage tube retention time and hospital stay, and higher hematoma clearance rate(P<0.05), but longer operation time. There was no significant difference in the incidence of postoperative complications between the two groups. The neuroendoscopic group had a better prognosis at the 3-month follow-up. Conclusion Both neuroendoscopy and borehole drainage can effectively treat CSDH, while neuroendoscopy has a higher hematoma clearance rate, shorter retention of drainage tube and hospital stay, and better clinical efficacy.

Table of Content