The diagnostic performance of deep-learning-based CT severity score to identify COVID-19 pneumonia

Objective: To determine the diagnostic accuracy of a deep-learning (DL)-based algorithm using chest computed tomography (CT) scans for the rapid diagnosis of coronavirus disease 2019 (COVID-19), as compared to the reference standard reverse-transcription polymerase chain reaction (RT-PCR) test. Methods: In this retrospective analysis, data of COVID-19 suspected patients who underwent RT-PCR and chest CT examination for the diagnosis of COVID-19 were assessed. By quantifying the affected area of the lung parenchyma, severity score was evaluated for each lobe of the lung with the DL-based algorithm. The diagnosis was based on the total lung severity score ranging from 0 to 25. The data were randomly split into a 40% training set and a 60% test set. Optimal cut-off value was determined using Youden-index method on the training cohort. Results: A total of 1259 patients were enrolled in this study. The prevalence of RT-PCR positivity in the overall investigated period was 51.5%. As compared to RT-PCR, sensitivity, specificity, positive predictive value, negative predictive value and accuracy on the test cohort were 39.0%, 80.2%, 68.0%, 55.0% and 58.9%, respectively. Regarding the whole data set, when adding those with positive RT-PCR test at any time during hospital stay or “COVID-19 without virus detection”, as final diagnosis to the true positive cases, specificity increased from 80.3% to 88.1% and the positive predictive value increased from 68.4% to 81.7%. Conclusion: DL-based CT severity score was found to have a good specificity and positive predictive value, as compared to RT-PCR. This standardized scoring system can aid rapid diagnosis and clinical decision making. Advances in knowledge: DL-based CT severity score can detect COVID-19-related lung alterations even at early stages, when RT-PCR is not yet positive.

Download Full-text

Machine and Deep Learning Applied to Predict Metabolic Syndrome Without a Blood Screening

Applied Sciences ◽

10.3390/app11104334 ◽

2021 ◽

Vol 11 (10) ◽

pp. 4334

Author(s):

Guadalupe O. Gutiérrez-Esparza ◽

Tania A. Ramírez-delReal ◽

Mireya Martínez-García ◽

Oscar Infante Infante Vázquez ◽

Maite Vallejo ◽

...

Keyword(s):

Metabolic Syndrome ◽

Deep Learning ◽

Positive Predictive Value ◽

Predictive Value ◽

Dietary Habits ◽

Smoking Habit ◽

Performance Model ◽

Whole Milk ◽

Corn Tortilla ◽

Blood Screening

The exponential increase of metabolic syndrome and its association with the risk impact of morbidity and mortality has propitiated the development of tools to diagnose this syndrome early. This work presents a model that is based on prognostic variables to classify Mexicans with metabolic syndrome without blood screening applying machine and deep learning. The data that were used in this study contain health parameters related to anthropometric measurements, dietary information, smoking habit, alcohol consumption, quality of sleep, and physical activity from 2289 participants of the Mexico City Tlalpan 2020 cohort. We use accuracy, balanced accuracy, positive predictive value, and negative predictive value criteria to evaluate the performance and validate different models. The models were separated by gender due to the shared features and different habits. Finally, the highest performance model in women found that the most relevant features were: waist circumference, age, body mass index, waist to height ratio, height, sleepy manner that is associated with snoring, dietary habits related with coffee, cola soda, whole milk, and Oaxaca cheese and diastolic and systolic blood pressure. Men’s features were similar to women’s; the variations were in dietary habits, especially in relation to coffee, cola soda, flavored sweetened water, and corn tortilla consumption. The positive predictive value obtained was 84.7% for women and 92.29% for men. With these models, we offer a tool that supports Mexicans to prevent metabolic syndrome by gender; it also lays the foundation for monitoring the patient and recommending change habits.

Download Full-text

The role of CT imaging for management of COVID-19 in epidemic area: early experience from a University Hospital

Insights into Imaging ◽

10.1186/s13244-020-00957-5 ◽

2021 ◽

Vol 12 (1) ◽

Author(s):

Vikram rao Bollineni ◽

Koenraad Hans Nieboer ◽

Seema Döring ◽

Nico Buls ◽

Johan de Mey

Keyword(s):

Ct Scan ◽

Predictive Value ◽

Ct Imaging ◽

Chest Ct ◽

University Hospital ◽

Rt Pcr ◽

Ct Findings ◽

Epidemic Area ◽

Chest Ct Scan ◽

Pcr Test

Abstract Background To evaluate the clinical value of the chest CT scan compared to the reference standard real-time polymerase chain reaction (RT-PCR) in COVID-19 patients. Methods From March 29th to April 15th of 2020, a total of 240 patients with respiratory distress underwent both a low-dose chest CT scan and RT-PCR tests. The performance of chest CT in diagnosing COVID-19 was assessed with reference to the RT-PCR result. Two board-certified radiologists (mean 24 years of experience chest CT), blinded for the RT-PCR result, reviewed all scans and decided positive or negative chest CT findings by consensus. Results Out of 240 patients, 60% (144/240) had positive RT-PCR results and 89% (213/240) had a positive chest CT scans. The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of chest CT in suggesting COVID-19 were 100% (95% CI: 97–100%, 144/240), 28% (95% CI: 19–38%, 27/240), 68% (95% CI: 65–70%) and 100%, respectively. The diagnostic accuracy of the chest CT suggesting COVID-19 was 71% (95% CI: 65–77%). Thirty-three patients with positive chest CT scan and negative RT-PCR test at baseline underwent repeat RT-PCR assay. In this subgroup, 21.2% (7/33) cases became RT-PCR positive. Conclusion Chest CT imaging has high sensitivity and high NPV for diagnosing COVID-19 and can be considered as an alternative primary screening tool for COVID-19 in epidemic areas. In addition, a negative RT-PCR test, but positive CT findings can still be suggestive of COVID-19 infection.

Download Full-text

Accuracy of Computed Tomography for Detection of COVID-19 among Suspected Patients

10.53350/pjmhs211592474 ◽

2021 ◽

Vol 15 (9) ◽

pp. 2474-2476

Author(s):

Maham Munir Awan ◽

Afshan Noreen ◽

Farah Kalsoom ◽

Muhammad Tahir ◽

Umaima Majeed ◽

...

Keyword(s):

Computed Tomography ◽

Predictive Value ◽

Gold Standard ◽

Radiology Department ◽

True Positive ◽

Test Results ◽

Rt Pcr ◽

Radiological Society ◽

Medical University ◽

Pcr Test

Objective: To determine the accuracy of CT chest in diagnosis of COVID-19 taking RT-PCR-testing as gold standard. Materials and Methods: A total of 150 patients of suspicion of COVID-19 who were referred for CT Chest in Radiology Department of Nishtar Medical University Multan from June-2020 to May-2021 were included. In all patients, two RT-PCR test results were obtained with 7 days of admission in hospital. Presence of any of these positive was labelled as COVID-19 infection. CT chest was performed in all patients within 2 days of admission in hospital using 128 slices CT scan machine. The diagnosis of COVID-19 infection was made according to the recommendations by Radiological Society of North America (RSNA) protocol. Results: Mean age was 51.3±14.7 years. 78 (52%) patients were male and 72 (48%) patients were female. RTPCR test was positive in 89 (59.3%) patients. While the CT chest findings were suggestive of COVID-19 infection in 130 (86.7%) patients. The sensitivity of CT chest was 95.5%, specificity 26.2%, PPV wad 65.4% and NPV was 80.0%. Conclusion: CT chest has a very good sensitivity for detection of COVID-19, it can be used as a rapid diagnostic tool especially in areas of pandemic. However, the specificity of CT chest is low, that can limit its use in low COVID-19 affected areas. Keywords: COVID-19, Computed tomography, False Positive, True Positive, Positive Predictive Value, Negative Predictive Value.

Download Full-text

Evaluation of a deep learning image assessment system for detecting severe retinopathy of prematurity

British Journal of Ophthalmology ◽

10.1136/bjophthalmol-2018-313156 ◽

2018 ◽

Vol 103 (5) ◽

pp. 580-584 ◽

Cited By ~ 20

Author(s):

Travis K Redd ◽

John Peter Campbell ◽

James M Brown ◽

Sang Jin Kim ◽

Susan Ostmo ◽

...

Keyword(s):

Deep Learning ◽

Retinopathy Of Prematurity ◽

Predictive Value ◽

Severity Score ◽

Receiver Operating Curve ◽

Fundus Photography ◽

Image Analysis System ◽

Reference Standard ◽

Type 1 Rop

BackgroundPrior work has demonstrated the near-perfect accuracy of a deep learning retinal image analysis system for diagnosing plus disease in retinopathy of prematurity (ROP). Here we assess the screening potential of this scoring system by determining its ability to detect all components of ROP diagnosis.MethodsClinical examination and fundus photography were performed at seven participating centres. A deep learning system was trained to detect plus disease, generating a quantitative assessment of retinal vascular abnormality (the i-ROP plus score) on a 1–9 scale. Overall ROP disease category was established using a consensus reference standard diagnosis combining clinical and image-based diagnosis. Experts then ranked ordered a second data set of 100 posterior images according to overall ROP severity.Results4861 examinations from 870 infants were analysed. 155 examinations (3%) had a reference standard diagnosis of type 1 ROP. The i-ROP deep learning (DL) vascular severity score had an area under the receiver operating curve of 0.960 for detecting type 1 ROP. Establishing a threshold i-ROP DL score of 3 conferred 94% sensitivity, 79% specificity, 13% positive predictive value and 99.7% negative predictive value for type 1 ROP. There was strong correlation between expert rank ordering of overall ROP severity and the i-ROP DL vascular severity score (Spearman correlation coefficient=0.93; p<0.0001).ConclusionThe i-ROP DL system accurately identifies diagnostic categories and overall disease severity in an automated fashion, after being trained only on posterior pole vascular morphology. These data provide proof of concept that a deep learning screening platform could improve objectivity of ROP diagnosis and accessibility of screening.

Download Full-text

Diagnosis and Etiological Analysis of Gastroesophageal Reflux Disease by Gastric Filling Ultrasound and GerdQ Scale

Journal of Healthcare Engineering ◽

10.1155/2021/5629067 ◽

2021 ◽

Vol 2021 ◽

pp. 1-6

Author(s):

Bo Wang ◽

Qian Sun ◽

Yonghong Du ◽

Kexiao Mu ◽

Jingxia Jiao

Keyword(s):

Gastroesophageal Reflux ◽

Positive Predictive Value ◽

Negative Predictive Value ◽

Predictive Value ◽

Structure And Function ◽

Youden Index ◽

Abnormal Structure ◽

And Function ◽

Antireflux Barrier ◽

Ultrasound Parameters

Objective. To investigate the diagnosis and etiological analysis of GERD by gastric filling ultrasound and GerdQ scale. Methods. The clinical data of 100 suspected GERD patients were selected for retrospective analysis. The selection time was from June 2016 to June 2019. According to the gold standard (endoscopy) results, they were divided into the gastroesophageal reflux group (positive, n = 62) and the nongastroesophageal reflux group (negative, n = 38); both gastric filling ultrasound and GerdQ scale examination were performed to compare the positive predictive value and negative predictive value, evaluate the abdominal esophageal length, His angle, and GerdQ scale score, and analyze the AUC value, sensitivity, specificity, and Youden index of His angle, length of abdominal esophagus, combined ultrasound parameters, and GerdQ scale in the diagnosis of GERD. Results. 100 patients with suspected GERD were diagnosed as GERD by endoscopy; in a total of 62 cases, the percentage was 62.00%. Among them, 28 cases were caused by the abnormal structure and function of the antireflux barrier, accounting for 45.16%, 18 cases were caused by the reduction of acid clearance of the esophagus, accounting for 29.03%, and 16 cases were caused by the weakening of the esophageal mucosal barrier, accounting for 25.81%. After ultrasound detection, the positive predictive value was 88.71% and the negative predictive value was 81.58%; after the GerdQ scale was tested, the positive predictive value was 71.43% and the negative predictive value was 54.05%. The length of the abdominal esophagus in the gastroesophageal reflux group was lower than that of the nongastroesophageal reflux group, while the scores of His angle and GerdQ scale were higher than those in the gastroesophageal reflux group ( P < 0.05 ). ROC curve analysis showed that the AUC values of His angle, length of abdominal esophagus, combined ultrasound parameters, and GerdQ scale to diagnose GERD were 0.957, 0.861, 0.996, and 0.931 ( P < 0.05 ), their sensitivity was 93.5%, 98.40%, 98.40%, and 90.30%, and the specificity was 92.10%, 63.20%, 100.00%, and 92.10%, respectively. Conclusion. Both gastric filling ultrasound and GerdQ scale have a certain application value in the diagnosis of GERD, but the former has a higher accuracy rate, and it is more common for gastroesophageal reflux caused by abnormal structure and function of antireflux barrier in etiological analysis.

Download Full-text

Diagnostic assessment of a deep learning system for detecting atrial fibrillation in pulse waveforms

Heart ◽

10.1136/heartjnl-2018-313147 ◽

2018 ◽

Vol 104 (23) ◽

pp. 1921-1928 ◽

Cited By ~ 36

Author(s):

Ming-Zher Poh ◽

Yukkee Cheung Poh ◽

Pak-Hei Chan ◽

Chun-Ka Wong ◽

Louise Pun ◽

...

Keyword(s):

Atrial Fibrillation ◽

Deep Learning ◽

Test Data ◽

Predictive Value ◽

Characteristic Curve ◽

Performance Comparison ◽

Learning System ◽

Training Data ◽

Validation Data ◽

Data Set

ObjectiveTo evaluate the diagnostic performance of a deep learning system for automated detection of atrial fibrillation (AF) in photoplethysmographic (PPG) pulse waveforms.MethodsWe trained a deep convolutional neural network (DCNN) to detect AF in 17 s PPG waveforms using a training data set of 149 048 PPG waveforms constructed from several publicly available PPG databases. The DCNN was validated using an independent test data set of 3039 smartphone-acquired PPG waveforms from adults at high risk of AF at a general outpatient clinic against ECG tracings reviewed by two cardiologists. Six established AF detectors based on handcrafted features were evaluated on the same test data set for performance comparison.ResultsIn the validation data set (3039 PPG waveforms) consisting of three sequential PPG waveforms from 1013 participants (mean (SD) age, 68.4 (12.2) years; 46.8% men), the prevalence of AF was 2.8%. The area under the receiver operating characteristic curve (AUC) of the DCNN for AF detection was 0.997 (95% CI 0.996 to 0.999) and was significantly higher than all the other AF detectors (AUC range: 0.924–0.985). The sensitivity of the DCNN was 95.2% (95% CI 88.3% to 98.7%), specificity was 99.0% (95% CI 98.6% to 99.3%), positive predictive value (PPV) was 72.7% (95% CI 65.1% to 79.3%) and negative predictive value (NPV) was 99.9% (95% CI 99.7% to 100%) using a single 17 s PPG waveform. Using the three sequential PPG waveforms in combination (<1 min in total), the sensitivity was 100.0% (95% CI 87.7% to 100%), specificity was 99.6% (95% CI 99.0% to 99.9%), PPV was 87.5% (95% CI 72.5% to 94.9%) and NPV was 100% (95% CI 99.4% to 100%).ConclusionsIn this evaluation of PPG waveforms from adults screened for AF in a real-world primary care setting, the DCNN had high sensitivity, specificity, PPV and NPV for detecting AF, outperforming other state-of-the-art methods based on handcrafted features.

Download Full-text

Impact of COVID-19 pre-test probability on positive predictive value of high cycle threshold SARS-CoV-2 real-time reverse transcription PCR test results

Infection Control and Hospital Epidemiology ◽

10.1017/ice.2021.369 ◽

2021 ◽

pp. 1-18

Author(s):

Jonathan B. Gubbay ◽

Heather Rilkoff ◽

Heather L. Kristjanson ◽

Jessica D. Forbes ◽

Michelle Murti ◽

...

Keyword(s):

Positive Predictive Value ◽

Real Time ◽

Reverse Transcription ◽

Predictive Value ◽

Large Scale ◽

Rdrp Gene ◽

Nucleic Acid Detection ◽

Rt Pcr ◽

Pcr Assay ◽

Test Probability

Abstract Objectives Performance characteristics of SARS-CoV-2 nucleic acid detection assays are understudied within contexts of low pre-test probability, including screening asymptomatic persons without epidemiological links to confirmed cases, or asymptomatic surveillance testing. SARS-CoV-2 detection without symptoms may represent presymptomatic or asymptomatic infection, resolved infection with persistent RNA shedding, or a false positive test. This study assessed positive predictive value of SARS-CoV-2 real-time reverse transcription polymerase chain reaction (rRT-PCR) assays by retesting positive specimens from five pre-test probability groups ranging from high to low with an alternate assay. Methods A total of 122 rRT-PCR positive specimens collected from unique patients between March and July 2020 were retested using a laboratory-developed nested RT-PCR assay targeting the RNA-dependent RNA polymerase (RdRp) gene followed by Sanger sequencing. Results Significantly fewer (15.6%) positive results in the lowest pre-test probability group (facilities with institution-wide screening having ≤ 3 positive asymptomatic cases) were reproduced with the nested RdRp gene RT-PCR assay than in each of the four groups with higher pre-test probability (individual group range 50·0% to 85·0%). Conclusions Large-scale SARS-CoV-2 screening testing initiatives among low pre-test probability populations should be evaluated thoroughly prior to implementation given the risk of false positives and consequent potential for harm at the individual and population level.

Download Full-text

Is the Portable Chest Radiographic More Reliable to Reveal Covid19 in Highly Suspicion Patient before Real-Time Reverse Transcription Polymerase Chain Reaction (RT-PCR) Test?

International Journal of Innovative Research in Medical Science ◽

10.23958/ijirms/vol06-i03/1085 ◽

2021 ◽

Vol 6 (03) ◽

pp. 228-236

Author(s):

Bushra A. A. Albazi ◽

Dr Noof. Albaz ◽

Dr Nayef. Alqahtani ◽

Dr. Angham Salih ◽

Dr Rafat Mohtasab

Keyword(s):

Polymerase Chain Reaction ◽

Retrospective Study ◽

Predictive Value ◽

Rt Pcr ◽

Chain Reaction ◽

X Ray ◽

Polymerase Chain ◽

Rate Type ◽

Test Result ◽

Pcr Test

A large number of patients with coronavirus disease 2019 (COVID-19) present at hospitals. There are a limited number of isolation rooms open, and patients must often wait a long time to get a reverse transcription-polymerase chain reaction (RT-PCR) test done. This necessitates the introduction of effective triage plans. A patient with suspicions is referred to an emergency room (ED) depending on their medical record for a simple physical assessment, blood test findings, and chest imaging.A retrospective study design was conduct at Prince Sultan Medical Military City (PSMMC). Ethical approval was obtained from the institutional board to wave the consent forms since it is a retrospective study. Only the primary investigator has had the data access to the patients’ medical records. The collected patient records were under specific categories, including symptoms score starts from 5 and above, RT-PCR test result done after CXRP imaging, the patient admitted to the emergency department (ED). Excluding all CXRP done after RT-PCR TEST, positive Covid 19 admitted to the intensive care unit (ICU), pediatric patients, and patients with score symptoms were less than five. Two experienced radiologists reviewed the images blindly, and the inter-observer reliability of observations noted by the radiologists was calculated. As for the relationship between the x-ray reading and the RT-PCR test result, our results showed a high correlation between the variables (chi-square χ² = 12.44, with df =1, and p<0.001). The sensitivity of x-ray diagnosing covid19 was 65.52 %, while the specificity was 54.51 %, and the accuracy of radiologists reading was 58.17 %. Furthermore, the positive predictive value (PPV) was 41.76 %, and the negative predictive value (NPV) was 76.05%. Finally, the false positive rate (type-i error (alpha) was 45.49%, and the false-negative rate (type-ii error (beta) was 34.48% Our research findings show that CXRP imaging can detect COVID-19 infection in symptomatic patients and can be a valuable addition to RT-PCR testing. In an inpatient ED environment where availability of test kits, laboratory equipment, and laboratory personnel is compromised and risks delaying patient treatment and hospital workflow, serial CXRP could theoretically be used as an adjunct diagnostic function and monitoring in patients suspected of having COVID-19.

Download Full-text

Coronavirus Disease 2019 (COVID-19) in Italy: Double reading of chest CT examination

10.21203/rs.3.rs-36750/v1 ◽

2020 ◽

Author(s):

Reginelli Alfonso ◽

Grassi Roberto ◽

Feragalli Beatrice ◽

Belfiore Maria Paola ◽

Montanelli Alessandro ◽

...

Keyword(s):

Predictive Value ◽

False Negative ◽

Viral Pneumonia ◽

Ct Diagnosis ◽

Rt Pcr ◽

Ct Examination ◽

Polymerase Chain ◽

Sensitivity Specificity ◽

Age Range ◽

Pcr Test

Abstract OBJECTIVETo assess the performance of the second reading of chest Compute Tomography (CT) examinations by expert radiologists in patients with discordance between the reverse transcription real-time fluorescence polymerase chain reaction (RT-PCR) test for COVID-19 viral pneumonia and the first CT report.MATERIALS AND METHODS.Three hundred seventy-heigth patients were included in this retrospective study (121 women and 257 men; 71 years of median age - range, 29–93 years) subjected to RT-PCR test for suspicious COVID-19 infection. All patients were subjected to CT examination in order to evaluate the pulmonary disease involvement by COVID-19. CT images were reviewed first by two radiologists who identified COVID-19 typical CT patterns and then reanalyzed by anoter two radiologists using a CT structured report for COVID-19 diagnosis.RESULTS.The median temporal window between RT-PCRs execution and CT scan was 0 days with a range of [-9, 11] days. RT-PCR test was resulted positive in 328/378 (86.8%). Discordance between RT-PCR and CT findings for viral pneumonia was revealed in 60 cases. The second reading changed the CT diagnosis in 16/60 (26.7%) cases contributing to increase the concordance with the RT-PCR. Among these 60 cases, 8 were false negative with positive RT-PCR, and 36 were false positive with negative RT-PCR. Sensitivity, specificity, positive predictive value and negative predictive value of CT were respectively of 97.3%, 53.8%, 89.0%, and 88.4%.CONCLUSION.Double reading of CT could increase the diagnostic confidence of radiological interpretation in COVID-19 patients. Using expert second readers could reduce the rate of discrepant cases between RT-PCR results and CT diagnosis for COVID-19 viral pneumonia.

Download Full-text

A Method for Optimizing and Validating Institution-Specific Flagging Criteria for Automated Cell Counters

Archives of Pathology & Laboratory Medicine ◽

10.5858/2009-0531-oa.1 ◽

2010 ◽

Vol 134 (10) ◽

pp. 1528-1533

Author(s):

Anthony Sireci ◽

Robert Schlaberg ◽

Alexander Kratz

Keyword(s):

Positive Predictive Value ◽

Blood Cell ◽

White Blood Cell ◽

Predictive Value ◽

Youden Index ◽

Design Data ◽

Cell Counts ◽

Blood Cell Counts ◽

White Blood Cell Counts ◽

Differential White Blood Cell

Abstract Context.—Automated cell counters use alerts (flags) to indicate which differential white blood cell counts can be released directly from the instrument and which samples require labor-intensive slide reviews. The thresholds at which many of these flags are triggered can be adjusted by individual laboratories. Many users, however, use factory-default settings or adjust the thresholds through a process of trial and error. Objective.—To develop a systematic method, combining statistical analysis and clinical judgment, to optimize the flagging thresholds on automated cell counters. Design.—Data from 502 samples flagged by Sysmex XE-2100/5000 (Sysmex, Kobe, Japan) instruments, with at least 1 of 5 user-adjustable, white blood cell count flags, were used to change the flagging thresholds for maximal diagnostic effectiveness by optimizing the Youden index for each flag (the optimization set). The optimized thresholds were then validated with a second set of 378 samples (the validation set). Results.—Use of the new thresholds reduced the review rate caused by the 5 flags from 6.5% to 2.9% and improved the positive predictive value of the flagging system for any abnormality from 27% to 37%. Conclusions.—This method can be used to optimize thresholds for flag alerts on automated cell counters of any type and to improve the overall positive predictive value of the flagging system at the expense of a reduction in the negative predictive value. A reduced manual review rate helps to focus resources on differential white blood cell counts that are of clinical significance and may improve turnaround time.

Download Full-text