Diagnostic classification of coronavirus disease 2019 (COVID-19) and other pneumonias using radiomics features in CT chest images

AbstractWe propose a classification method using the radiomics features of CT chest images to identify patients with coronavirus disease 2019 (COVID-19) and other pneumonias. The chest CT images of two groups of participants (90 COVID-19 patients who were confirmed as positive by nucleic acid test of RT-PCR and 90 other pneumonias patients) were collected, and the two groups of data were manually drawn to outline the region of interest (ROI) of pneumonias. The radiomics method was used to extract textural features and histogram features of the ROI and obtain a radiomics features vector from each sample. Then, we divided the data into two independent radiomic cohorts for training (70 COVID-19 patients and 70 other pneumonias patients), and validation (20 COVID-19 patients and 20 other pneumonias patients) by using support vector machine (SVM). This model used 20 rounds of tenfold cross-validation for training. Finally, single-shot testing of the final model was performed on the independent validation cohort. In the COVID-19 patients, correlation analysis (multiple comparison correction—Bonferroni correction, P < 0.05/7) was also conducted to determine whether the textural and histogram features were correlated with the laboratory test index of blood, i.e., blood oxygen, white blood cell, lymphocytes, neutrophils, C-reactive protein, hypersensitive C-reactive protein, and erythrocyte sedimentation rate. The final model showed good discrimination on the independent validation cohort, with an accuracy of 89.83%, sensitivity of 94.22%, specificity of 85.44%, and AUC of 0.940. This proved that the radiomics features were highly distinguishable, and this SVM model can effectively identify and diagnose patients with COVID-19 and other pneumonias. The correlation analysis results showed that some textural features were positively correlated with WBC, and NE, and also negatively related to SPO2H and NE. Our results showed that radiomic features can classify COVID-19 patients and other pneumonias patients. The SVM model can achieve an excellent diagnosis of COVID-19.

Download Full-text

Diagnostic classification of coronavirus disease 2019 (COVID-19) and other pneumonias using radiomics features in CT chest images

10.21203/rs.3.rs-34648/v1 ◽

2020 ◽

Author(s):

Ning Yang ◽

Faming Liu ◽

Chunlong Li ◽

Wenqing Xiao ◽

Shuangcong Xie ◽

...

Keyword(s):

Correlation Analysis ◽

Area Under The Curve ◽

Region Of Interest ◽

Texture Features ◽

Support Vector ◽

C Reactive Protein ◽

Reactive Protein ◽

Svm Model ◽

Multiple Comparison Correction

Abstract We propose a classification method using the radiomics features of CT chest images to identify patients with coronavirus disease 2019 (COVID-19) and other pneumonias. The chest CT images of two groups of participants (90 COVID-19 patients and 90 other pneumonias patients) were collected, and the two groups of data were manually drawn to outline the region of interest (ROI) of pneumonias. The radiomics method was used to extract textural features and histogram features of the ROI and obtain a radiomics features vector from each sample. Finally, using the radiomics features as an input, a support vector machine (SVM) model was constructed to classify patients with COVID-19 and patients with other pneumonias. This model used 20 rounds of 10-fold cross-validation for training and testing. In the COVID-19 patients, correlation analysis (multiple comparison correction—Bonferroni correction, p<0.05/7) was also conducted to determine whether the textural and histogram features were correlated with the laboratory test index of blood, i.e., blood oxygen, white blood cell, lymphocytes, neutrophils, C-reactive protein, hypersensitive C-reactive protein, and erythrocyte sedimentation rate. The results showed that the proposed method had a classification accuracy as high as 88.33%, sensitivity of 83.56%, specificity of 93.11%, and an area under the curve of 0.947. This proved that the radiomics features were highly distinguishable, and this SVM model can effectively identify and diagnose patients with COVID-19 and other pneumonias. The correlation analysis results showed that some texture features were positively correlated with WBC, NE, and CRP and also negatively related to SPO2H and NE.

Download Full-text

Exclusion of bacterial co-infection in COVID-19 using baseline inflammatory markers and their response to antibiotics

10.1101/2020.10.09.20199778 ◽

2020 ◽

Author(s):

Claire Y Mason ◽

Tanmay Kanitkar ◽

Charlotte J Richardson ◽

Marisa Lanzman ◽

Zak Stone ◽

...

Keyword(s):

Inflammatory Markers ◽

Pulmonary Infection ◽

Validation Cohort ◽

White Cell Count ◽

Community Acquired Pneumonia ◽

C Reactive Protein ◽

Blood Samples ◽

Reactive Protein ◽

Independent Validation ◽

Laboratory Investigations

AbstractBackgroundCOVID-19 is infrequently complicated by secondary bacterial infection, but nevertheless antibiotic prescriptions are common. We used community-acquired pneumonia (CAP) as a benchmark to define the processes that occur in a bacterial pulmonary infection, and tested the hypothesis that baseline inflammatory markers and their response to antibiotic therapy could distinguish CAP from COVID-19.MethodsIn patients admitted to Royal Free Hospital (RFH) and Barnet Hospital (BH) we defined CAP by lobar consolidation on chest radiograph, and COVID-19 by SARS-CoV-2 detection by PCR. Data were derived from routine laboratory investigations.ResultsOn admission all CAP and >90% COVID-19 patients received antibiotics. We identified 106 CAP and 619 COVID-19 patients at RFH. CAP was characterised by elevated white cell count (WCC) and C-reactive protein (CRP) compared to COVID-19 (median WCC 12.48 (IQR 8.2-15.3) vs 6.78 (IQR 5.2-9.5) x106 cells/ml and median CRP CRP 133.5 (IQR 65-221) vs 86 (IQR 42-160) mg/L). Blood samples collected 48-72 hours into admission revealed decreasing CRP in CAP but not COVID-19 (CRP difference −33 (IQR −112 to +3.5) vs +15 (IQR −15 to +70) mg/L respectively). In the independent validation cohort (BH) consisting of 169 CAP and 181 COVID-19 patients, admission WCC >8.2×106 cells/ml or falling CRP during admission identified 95% of CAP cases, and predicted the absence of bacterial co-infection in 45% of COVID-19 patients.ConclusionsWe propose that in COVID-19 the absence of both elevated baseline WCC and antibiotic-related decrease in CRP can exclude bacterial co-infection and facilitate antibiotic stewardship efforts.

Download Full-text

A novel procalcitonin-based score for detecting sepsis among critically ill patients

PLoS ONE ◽

10.1371/journal.pone.0245748 ◽

2021 ◽

Vol 16 (1) ◽

pp. e0245748

Author(s):

Tung-Lin Tsui ◽

Ya-Ting Huang ◽

Wei-Chih Kan ◽

Mao-Sheng Huang ◽

Min-Yu Lai ◽

...

Keyword(s):

Logistic Regression ◽

Regression Analysis ◽

Critically Ill ◽

Critically Ill Patients ◽

Validation Cohort ◽

Multivariate Logistic Regression Analysis ◽

Sepsis Group ◽

C Reactive Protein ◽

Multivariate Logistic Regression ◽

Reactive Protein

Background Procalcitonin (PCT) has been widely investigated as an infection biomarker. The study aimed to prove that serum PCT, combining with other relevant variables, has an even better sepsis-detecting ability in critically ill patients. Methods We conducted a retrospective cohort study in a regional teaching hospital enrolling eligible patients admitted to intensive care units (ICU) between July 1, 2016, and December 31, 2016, and followed them until March 31, 2017. The primary outcome measurement was the occurrence of sepsis. We used multivariate logistic regression analysis to determine the independent factors for sepsis and constructed a novel PCT-based score containing these factors. The area under the receiver operating characteristics curve (AUROC) was applied to evaluate sepsis-detecting abilities. Finally, we validated the score using a validation cohort. Results A total of 258 critically ill patients (70.9±16.3 years; 55.4% man) were enrolled in the derivation cohort and further subgrouped into the sepsis group (n = 115) and the non-sepsis group (n = 143). By using the multivariate logistic regression analysis, we disclosed five independent factors for detecting sepsis, namely, “serum PCT level,” “albumin level” and “neutrophil-lymphocyte ratio” at ICU admission, along with “diabetes mellitus,” and “with vasopressor.” We subsequently constructed a PCT-based score containing the five weighted factors. The PCT-based score performed well in detecting sepsis with the cut-points of 8 points (AUROC 0.80; 95% confidence interval (CI) 0.74–0.85; sensitivity 0.70; specificity 0.76), which was better than PCT alone, C-reactive protein and infection probability score. The findings were confirmed using an independent validation cohort (n = 72, 69.2±16.7 years, 62.5% men) (cut-point: 8 points; AUROC, 0.79; 95% CI 0.69–0.90; sensitivity 0.64; specificity 0.87). Conclusions We proposed a novel PCT-based score that performs better in detecting sepsis than serum PCT levels alone, C-reactive protein, and infection probability score.

Download Full-text

Pattern recognition of enrichment levels of SELEX-based candidate aptamers for human C-reactive protein

Biomedical Engineering / Biomedizinische Technik ◽

10.1515/bmt-2015-0230 ◽

2017 ◽

Vol 62 (3) ◽

Author(s):

Xinliang Yu ◽

Ruqin Yu ◽

Xiaohai Yang

Keyword(s):

Pattern Recognition ◽

Latent Variables ◽

Molecular Descriptors ◽

Principal Component ◽

Support Vector ◽

C Reactive Protein ◽

Reactive Protein ◽

Systematic Evolution ◽

Model C ◽

Exponential Enrichment

AbstractSelecting aptamers for human C-reactive protein (CRP) would be of critical importance in predicting the risk for cardiovascular disease. The enrichment level of DNA aptamers is an important parameter for selecting candidate aptamers for further affinity and specificity determination. This paper is the first report on pattern recognition used for CRP aptamer enrichment levels in the systematic evolution of ligands by exponential enrichment (SELEX) process, by applying structure-activity relationship models. After generating 10 rounds of graphene oxide (GO)-SELEX and 1670 molecular descriptors, eight molecular descriptors were selected and five latent variables were then obtained with principal component analysis (PCA), to develop a support vector classification (SVC) model. The SVC model (C=8.1728 and

Download Full-text

Cascaded Machine-Learning Technique for Debris Classification in Floor-Cleaning Robot Application

Applied Sciences ◽

10.3390/app8122649 ◽

2018 ◽

Vol 8 (12) ◽

pp. 2649 ◽

Cited By ~ 14

Author(s):

Balakrishnan Ramalingam ◽

Anirudh Lakshmanan ◽

Muhammad Ilyas ◽

Anh Le ◽

Mohan Elara

Keyword(s):

Binary Classification ◽

Support Vector ◽

Single Shot ◽

Machine Learning Technique ◽

Cleaning Robot ◽

Svm Model ◽

Learning Technique ◽

Essential Function ◽

Cleaning Robots

Debris detection and classification is an essential function for autonomous floor-cleaning robots. It enables floor-cleaning robots to identify and avoid hard-to-clean debris, specifically large liquid spillage debris. This paper proposes a debris-detection and classification scheme for an autonomous floor-cleaning robot using a deep Convolutional Neural Network (CNN) and Support Vector Machine (SVM) cascaded technique. The SSD (Single-Shot MultiBox Detector) MobileNet CNN architecture is used for classifying the solid and liquid spill debris on the floor through the captured image. Then, the SVM model is employed for binary classification of liquid spillage regions based on size, which helps floor-cleaning devices to identify the larger liquid spillage debris regions, considered as hard-to-clean debris in this work. The experimental results prove that the proposed technique can efficiently detect and classify the debris on the floor and achieves 95.5% percent classification accuracy. The cascaded approach takes approximately 71 milliseconds for the entire process of debris detection and classification, which implies that the proposed technique is suitable for deploying in real-time selective floor-cleaning applications.

Download Full-text

C-Reactive Protein (CRP) Levels in Immune Checkpoint Inhibitor Response and Progression in Advanced Non-Small Cell Lung Cancer: A Bi-Center Study

Cancers ◽

10.3390/cancers12082319 ◽

2020 ◽

Vol 12 (8) ◽

pp. 2319

Author(s):

Jakob M. Riedl ◽

Dominik A. Barth ◽

Wolfgang M. Brueckl ◽

Gloria Zeitler ◽

Vasile Foris ◽

...

Keyword(s):

Lung Cancer ◽

Validation Cohort ◽

External Validation ◽

Small Cell ◽

C Reactive Protein ◽

Small Cell Lung ◽

Discovery Cohort ◽

Reactive Protein ◽

Progression Risk ◽

Center Study

Background: Biomarkers for predicting response to immune checkpoint inhibitors (ICI) are scarce and often lack external validation. This study provides a comprehensive investigation of pretreatment C-reactive protein (CRP) levels as well as its longitudinal trajectories as a marker of treatment response and disease outcome in patients with advanced non-small cell lung cancer (NSCLC) undergoing immunotherapy with anti PD-1 or anti PD-L1 agents. Methods: We performed a retrospective bi-center study to assess the association between baseline CRP levels and anti PD-(L)1 treatment outcomes in the discovery cohort (n = 90), confirm these findings in an external validation cohort (n = 101) and explore the longitudinal evolution of CRP during anti PD-(L)1 treatment and the potential impact of dynamic CRP changes on treatment response and disease outcome in the discovery cohort. Joint models were implemented to evaluate the association of longitudinal CRP trajectories and progression risk. Primary treatment outcomes were progression-free survival (PFS) and overall survival (OS), while the objective response rate (ORR) was a secondary outcome, respectively. Results: In the discovery cohort, elevated pretreatment CRP levels emerged as independent predictors of worse PFS (HR per doubling of baseline CRP = 1.37, 95% CI: 1.16–1.63, p < 0.0001), worse OS (HR per doubling of baseline CRP = 1.42, 95% CI: 1.18–1.71, p < 0.0001) and a lower ORR ((odds ratio (OR) of ORR per doubling of baseline CRP = 0.68, 95% CI: 0.51–0.92, p = 0.013)). In the validation cohort, pretreatment CRP could be fully confirmed as a predictor of PFS and OS, but not ORR. Elevated trajectories of CRP during anti PD-(L)1 treatment (adjusted HR per 10 mg/L increase in CRP = 1.22, 95% CI: 1.15–1.30, p < 0.0001), as well as a faster increases of CRP over time (HR per 10 mg/L/month faster increase in CRP levels = 13.26, 95% CI: 1.14–154.54, p = 0.039) were strong predictors of an elevated progression risk, whereas an early decline of CRP was significantly associated with a reduction in PFS risk (HR = 0.91, 95% CI: 0.83–0.99, p = 0.036), respectively. Conclusion: These findings support the concept that CRP should be further explored by future prospective studies as a simple non-invasive biomarker for assessing treatment benefit during anti PD-(L)1 treatment in advanced NSCLC.

Download Full-text

Forecasting Dry Bulk Freight Index with Improved SVM

Mathematical Problems in Engineering ◽

10.1155/2014/460684 ◽

2014 ◽

Vol 2014 ◽

pp. 1-12 ◽

Cited By ~ 1

Author(s):

Qianqian Han ◽

Bo Yan ◽

Guobao Ning ◽

B. Yu

Keyword(s):

Wavelet Transform ◽

Support Vector ◽

Data Series ◽

Shipping Industry ◽

Combined Model ◽

Final Model ◽

Proposed Model ◽

Svm Model ◽

The Impact ◽

Short Term Trend

An improved SVM model is presented to forecast dry bulk freight index (BDI) in this paper, which is a powerful tool for operators and investors to manage the market trend and avoid price risking shipping industry. The BDI is influenced by many factors, especially the random incidents in dry bulk market, inducing the difficulty in forecasting of BDI. Therefore, to eliminate the impact of random incidents in dry bulk market, wavelet transform is adopted to denoise the BDI data series. Hence, the combined model of wavelet transform and support vector machine is developed to forecast BDI in this paper. Lastly, the BDI data in 2005 to 2012 are presented to test the proposed model. The 84 prior consecutive monthly BDI data are the inputs of the model, and the last 12 monthly BDI data are the outputs of model. The parameters of the model are optimized by genetic algorithm and the final model is conformed through SVM training. This paper compares the forecasting result of proposed method and three other forecasting methods. The result shows that the proposed method has higher accuracy and could be used to forecast the short-term trend of the BDI.

Download Full-text

Combination of lymphocyte count and albumin concentration as a new prognostic biomarker for rectal cancer

Scientific Reports ◽

10.1038/s41598-021-84475-4 ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Takehito Yamamoto ◽

Kenji Kawada ◽

Koya Hida ◽

Ryo Matsusue ◽

Yoshiro Itatani ◽

...

Keyword(s):

Rectal Cancer ◽

Curative Resection ◽

Validation Cohort ◽

Prognostic Biomarker ◽

Stage Ii ◽

Albumin Concentration ◽

C Reactive Protein ◽

Discovery Cohort ◽

Free Survival ◽

Reactive Protein

AbstractAlthough numerous studies have highlighted the prognostic values of various inflammation-related markers, clinical significance remains to be elucidated. The prognostic values of inflammation-related biomarkers for rectal cancer were investigated in this study. A total of 448 patients with stage II/III rectal cancer undergoing curative resection were enrolled from the discovery cohort (n = 240) and validation cohort (n = 208). We comprehensively compared the prognostic values of 11 inflammation-related markers-derived from neutrophil, lymphocyte, platelet, monocyte, albumin, and C-reactive protein for overall survival (OS) and recurrence-free survival (RFS). Among 11 inflammation-related markers, only “lymphocyte × albumin (LA)” was significantly associated with both OS and RFS in the discovery cohort (P = 0.007 and 0.015, respectively). Multivariate analysis indicated that low LA was significantly associated with poor OS (hazard ratio [HR] 2.19, 95% confidence interval [CI] 1.09–4.58, P = 0.025), and poor RFS (HR 1.61, 95% CI 1.01–2.80, P = 0.048). Furthermore, using the discovery cohort, we confirmed that low LA was significantly associated with poor OS (HR 2.89, 95% CI 1.42–6.00, P = 0.002), and poor RFS (HR 1.79, 95% CI 1.04–2.95, P = 0.034). LA can be a novel prognostic biomarker for stage II/III rectal cancer.

Download Full-text

Associations between intraoperative nociceptive and early postoperative C-reactive protein levels in patients undergoing non-cardiac surgery under general anesthesia

10.21203/rs.2.208/v1 ◽

2019 ◽

Author(s):

Shiroh Nakamoto ◽

Hiroki Ogata ◽

Ayano Saeki ◽

Ryusuke Ueki ◽

Nobutaka Kariya ◽

...

Keyword(s):

Postoperative Complications ◽

Prediction Model ◽

General Anesthesia ◽

Multivariate Regression ◽

Validation Cohort ◽

Multivariate Regression Analysis ◽

C Reactive Protein ◽

Reactive Protein ◽

Major Complications ◽

Duration Of Surgery

Abstract Background: Early detection of postoperative increase in C-reactive protein (CRP) predicts complications after surgery. The preoperative and intraoperative factors associated with postoperative CRP changes are potentially significant in the prophylactic management of postoperative complications. Although ongoing nociception during surgery under general anesthesia is one of potential candidates for these factors, it has not been evaluated with the unavailability of valid nociception measures in clinical practice. Then we adopted averaged values of nociceptive response (NR) throughout surgery as intraoperative nociceptive levels, being examined the association between perioperative factors, early changes in postoperative CRP levels, and postoperative complications Material and Methods: Data from 174 adult patients undergoing elective non-cardiac surgery under general anesthesia on perioperative variables, including age, sex, BMI, American Society of Anesthesiologists-physical status (ASA-PS), duration of surgery, mean NR during surgery as intraoperative nociceptive level, CRP levels before and after surgery on postoperative day (POD) 1, and postoperative complications using the extended Clavien-Dindo classification were retrospectively obtained in a training cohort. Multivariate regression analysis was performed to determine the independent factor of CRP levels on POD1and to develop a prediction model. In two validation cohorts, both 75 patients undergoing mastectomy (validation cohort A) and 139 patients undergoing laparoscopic or open abdominal surgery (validation cohort B) were separately selected, and retrospectively utilized to evaluate the value of the prediction model. Results: CRP levels on POD1 in the training cohort significantly increased in the order of Clavien-Dindo grades. Multivariate regression analysis selected mean NR, BMI, and duration of surgery to set up the prediction model of CRP level on POD1, which showed significant correlation with the measured CRP in both two validation cohorts. To confirm associations between mean NR, postoperative CRP, and major complications (Clavien-Dindo grade ≥ IIIa), we performed a propensity score matching in the validation cohort B, using age, BMI, ASA-PS, and duration of operations, finding that both mean NR and CRP levels on POD1 were significantly higher in patients with major complications than those without major complications. Conclusion: Increases in the intraoperative nociceptive level likely associated with early increases in CRP level after surgery. Keywords: C-reactive protein, Postoperative complications, Nociception.

Download Full-text

A simple tool to help ruling-out Covid-19 in the emergency department: derivation and validation of the LDH-CRP-Lymphocyte (LCL) score

Emergency Care Journal ◽

10.4081/ecj.2020.9336 ◽

2020 ◽

Vol 16 (3) ◽

Author(s):

Jacopo Davide Giamello ◽

Giulia Paglietta ◽

Giulia Cavalot ◽

Attilio Allione ◽

Sara Abram ◽

...

Keyword(s):

Validation Cohort ◽

Negative Likelihood Ratio ◽

C Reactive Protein ◽

Derivation Cohort ◽

Reactive Protein ◽

Simple Tool ◽

Lactate Dehydrogenase C ◽

The Difference ◽

To Receive ◽

Rule Out

After the outbreak of the Covid-19 pandemic, cases of SARSCoV- 2 infections may gradually decrease in the next months. Given the reduced prevalence of the disease, Emergency Departments (ED) are starting to receive more and more non- Covid19 patients. Thus, a way to quickly discriminate ED patients with potential Covid-19 infection from non-Covid19 patients is needed in order to keep potentially contagious patients isolated while awaiting second-level testing. In this paper, we present the derivation and validation of a simple, practical, and cheap score that could be helpful to rule out Covid-19 among ED patients with suspicious symptoms (fever and/or dyspnoea). The LCL score was derived from a cohort of 335 patients coming to the ED of our hospital from March 16th to April 1st, 2020. It was then retrospectively validated in a similar cohort of 173 patients admitted to our ED during April. The score is based on blood values of lactate dehydrogenase, C-reactive protein, and lymphocyte count. The LCL score performed well both in the derivation and in the validation cohort, with an AUC respectively of 0.81 (95% CI: 0.77 – 0.86) and of 0.71 (95% CI: 0.63 – 0.78), given the difference in Covid- 19 prevalence between the two cohorts (57% vs 41% respectively). An LCL score equal to 0 had a negative predictive value of 0.92 in the derivation cohort and of 0.81 in the validation cohort, with a negative likelihood ratio respectively of 0.08 and 0.36 for Covid- 19 exclusion. This score could, therefore, constitute a useful tool to help physicians manage patients in the ED.

Download Full-text