scholarly journals Lightweight Model For The Prediction of COVID-19 Through The Detection And Segmentation of Lesions in Chest CT Scans

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
Single Layer ◽  
Ground Glass Opacity ◽  
Chest Ct ◽  
Training Data ◽  
Classification Model ◽  
Ct Scans ◽  
True Negative ◽  
Training Time ◽  
Regional Features ◽  
Similar Solutions

AbstractWe introduce a lightweight Mask R-CNN model that segments areas with the Ground Glass Opacity and Consolidation in chest CT scans. The model uses truncated ResNet18 and ResNet34 nets with a single layer of Feature Pyramid Network as a backbone net, thus substantially reducing the number of the parameters and the training time compared to similar solutions using deeper networks. Without any data balancing and manipulations, and using only a small fraction of the training data, COVID-CT-Mask-Net classification model with 6.12M total and 600K trainable parameters derived from Mask R-CNN, achieves 91.35% COVID-19 sensitivity, 91.63% Common Pneumonia sensitivity, 96.98% true negative rate and 93.95% overall accuracy on COVIDx-CT dataset (21191 images). We also present a thorough analysis of the regional features critical to the correct classification of the image. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

scholarly journals Lightweight Model For The Prediction of COVID-19 Through The Detection And Segmentation of Lesions in Chest CT Scans

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
Single Layer ◽  
Ground Glass Opacity ◽  
Chest Ct ◽  
Training Data ◽  
Classification Model ◽  
Ct Scans ◽  
True Negative ◽  
Training Time ◽  
Regional Features ◽  
Similar Solutions

Abstract We introduce a lightweight Mask R-CNN model that segments areas with the Ground Glass Opacity and Consolidation in chest CT scans. The model uses truncated ResNet18 and ResNet34 nets with a single layer of Feature Pyramid Network as a backbone net, thus substantially reducing the number of the parameters and the training time compared to similar solutions using deeper networks. Without any data balancing and manipulations, and using only a small fraction of the training data, COVID-CT-Mask-Net classification model with 6.12M total and 600K trainable parameters derived from Mask R-CNN, achieves 91.35% COVID-19 sensitivity, 91.63% Common Pneumonia sensitivity, 96.98% true negative rate and 93.95% overall accuracy on COVIDx-CT dataset (21191 images). We also present a thorough analysis of the regional features critical to the correct classification of the image. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

scholarly journals Lightweight Model For The Prediction of COVID-19 Through The Detection And Segmentation of Lesions in Chest CT Scans

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
Single Layer ◽  
Ground Glass Opacity ◽  
Chest Ct ◽  
Training Data ◽  
Classification Model ◽  
Ct Scans ◽  
True Negative ◽  
Training Time ◽  
Regional Features ◽  
Similar Solutions

Abstract We introduce a lightweight Mask R-CNN model that segments areas with the Ground Glass Opacity and Consolidation in chest CT scans. The model uses truncated ResNet18 and ResNet34 nets with a single layer of Feature Pyramid Network as a backbone net, thus substantially reducing the number of the parameters and the training time compared to similar solutions using deeper networks. Without any data balancing and manipulations, and using only a small fraction of the training data, COVID-CT-Mask-Net classification model with 6.12M total and 600K trainable parameters derived from Mask R-CNN, achieves 91.35% COVID-19 sensitivity, 91.63% Common Pneumonia sensitivity, 96.98% true negative rate and 93.95% overall accuracy on COVIDx-CT dataset (21191 images). We also present a thorough analysis of the regional features critical to the correct classification of the image. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

scholarly journals Detection and Segmentation of Lesion Areas in Chest CT Scans For The Prediction of COVID-19

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
Ground Glass Opacity ◽  
Chest Ct ◽  
Training Data ◽  
Classification Model ◽  
Ct Scans ◽  
Average Precision ◽  
True Negative ◽  
True Negative Rate ◽  
Ct Test ◽  
The Mean

In this paper we compare the models for the detection and segmentation of Ground Glass Opacity and Consolidation in chest CT scans. These lesion areas are often associated both with common pneumonia and COVID-19. We train a Mask R-CNN model to segment these areas with high accuracy using three approaches: merging masks for these lesions into one, deleting the mask for Consolidation, and using both masks separately. The best model achieves the mean average precision of 44.68% using MS COCO criterion for instance segmentation across all accuracy thresholds. The classification model, COVID-CT-Mask-Net, which learns to predict the presence of COVID-19 vs common pneumonia vs control, achieves the 93.88% COVID-19 sensitivity, 95.64% overall accuracy, 95.06% common pneumonia sensitivity and 96.91% true negative rate on the COVIDx-CT test split (21192 CT scans) using a small fraction of the training data. We also analyze the effect of Non-Maximum Suppression of overlapping object predictions, both on the segmentation and classification accuracy. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

scholarly journals Detection and Segmentation of Lesion Areas in Chest CT Scans For The Prediction of COVID-19

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
Ground Glass Opacity ◽  
Chest Ct ◽  
Training Data ◽  
Classification Model ◽  
Ct Scans ◽  
Average Precision ◽  
True Negative ◽  
True Negative Rate ◽  
Ct Test ◽  
The Mean

Abstract In this paper we compare the models for the detection and segmentation of Ground Glass Opacity and Consolidation in chest CT scans. These lesion areas are often associated both with common pneumonia and COVID-19. We train a Mask R-CNN model to segment these areas with high accuracy using three approaches: merging masks for these lesions into one, deleting the mask for Consolidation, and using both masks separately. The best model achieves the mean average precision of 44.68% using MS COCO criterion for instance segmentation across all accuracy thresholds. The classification model, COVID-CT-Mask-Net, which learns to predict the presence of COVID-19 vs common pneumonia vs control, achieves the 93.88% COVID-19 sensitivity, 95.64% overall accuracy, 95.06% common pneumonia sensitivity and 96.91% true negative rate on the COVIDx-CT test split (21192 CT scans) using a small fraction of the training data. We also analyze the effect of Non-Maximum Suppression of overlapping object predictions, both on the segmentation and classification accuracy. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

scholarly journals Detection and Segmentation of Lesion Areas in Chest CT Scans For The Prediction of COVID-19

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
Ground Glass Opacity ◽  
Chest Ct ◽  
Training Data ◽  
Classification Model ◽  
Ct Scans ◽  
Average Precision ◽  
True Negative ◽  
True Negative Rate ◽  
Ct Test ◽  
The Mean

Abstract In this paper we compare the models for the detection and segmentation of Ground Glass Opacity and Consolidation in chest CT scans. These lesion areas are often associated both with common pneumonia and COVID-19. We train a Mask R-CNN model to segment these areas with high accuracy using three approaches: merging masks for these lesions into one, deleting the mask for Consolidation, and using both masks separately. The best model achieves the mean average precision of 44.68% using MS COCO criterion for instance segmentation across all accuracy thresholds. The classification model, COVID-CT-Mask-Net, which learns to predict the presence of COVID-19 vs common pneumonia vs control, achieves the 93.88% COVID-19 sensitivity, 95.64% overall accuracy, 95.06% common pneumonia sensitivity and 96.91% true negative rate on the COVIDx-CT test split (21192 CT scans) using a small fraction of the training data. We also analyze the effect of Non-Maximum Suppression of overlapping object predictions, both on the segmentation and classification accuracy. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

scholarly journals One Shot Model For The Prediction of COVID-19 and Lesions Segmentation In Chest CT Scans Through The Affinity Among Lesion Mask Features

2021 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
Source Code ◽  
Chest Ct ◽  
Training Data ◽  
Ct Scans ◽  
Affinity Matrix ◽  
True Negative ◽  
Data Manipulation ◽  
True Negative Rate ◽  
Affinity Model ◽  
The Relationship

AbstractWe introduce a model that segments lesions and predicts COVID-19 from chest CT scans through the derivation of an affinity matrix between lesion masks. The novelty of the methodology is based on the computation of the affinity between the lesion masks’ features extracted from the image. First, a batch of vectorized lesion masks is constructed. Then, the model learns the parameters of the affinity matrix that captures the relationship between features in each vector. Finally, the affinity is expressed as a single vector of pre-defined length. Without any complicated data manipulation, class balancing tricks, and using only a fraction of the training data, we achieve a 91.74% COVID-19 sensitivity, 85.35% common pneumonia sensitivity, 97.26% true negative rate and 91.94% F1-score. Ablation studies show that the method can quickly generalize to new datasets. All source code, models and results are publicly available on https://github.com/AlexTS1980/COVID-Affinity-Model.

scholarly journals The halo sign: HRCT findings in 85 patients

2016 ◽  
Vol 42 (6) ◽  
pp. 435-439 ◽  
Author(s):  
Giordano Rafael Tronco Alves ◽  
◽  
Edson Marchiori ◽  
Klaus Irion ◽  
Carlos Schuler Nin ◽  
...  
Keyword(s):  
Immune Status ◽  
Ground Glass Opacity ◽  
Diagnostic Value ◽  
Chest Ct ◽  
Ct Scans ◽  
Pulmonary Lesions ◽  
Halo Sign ◽  
Immunosuppressed Patients ◽  
Primary Neoplasm ◽  
Immunocompetent Patients

ABSTRACT Objective: The halo sign consists of an area of ground-glass opacity surrounding pulmonary lesions on chest CT scans. We compared immunocompetent and immunosuppressed patients in terms of halo sign features and sought to identify those of greatest diagnostic value. Methods: This was a retrospective study of CT scans performed at any of seven centers between January of 2011 and May of 2015. Patients were classified according to their immune status. Two thoracic radiologists reviewed the scans in order to determine the number of lesions, as well as their distribution, size, and contour, together with halo thickness and any other associated findings. Results: Of the 85 patients evaluated, 53 were immunocompetent and 32 were immunosuppressed. Of the 53 immunocompetent patients, 34 (64%) were diagnosed with primary neoplasm. Of the 32 immunosuppressed patients, 25 (78%) were diagnosed with aspergillosis. Multiple and randomly distributed lesions were more common in the immunosuppressed patients than in the immunocompetent patients (p < 0.001 for both). Halo thickness was found to be greater in the immunosuppressed patients (p < 0.05). Conclusions: Etiologies of the halo sign differ markedly between immunocompetent and immunosuppressed patients. Although thicker halos are more likely to occur in patients with infectious diseases, the number and distribution of lesions should also be taken into account when evaluating patients presenting with the halo sign.

scholarly journals Diagnostic value of lung ultrasound versus chest CT in COVID-19

2020 ◽  
Vol 11 (2) ◽  
pp. 5-18
Author(s):  
S. S. Petrikov ◽  
I. E. Popova ◽  
V. M. Abuchina ◽  
R. Sh. Muslimov ◽  
L. T. Khamidova ◽  
...  
Keyword(s):  
Diagnostic Accuracy ◽  
Lung Diseases ◽  
Structural Changes ◽  
Lung Ultrasound ◽  
Ground Glass Opacity ◽  
Diagnostic Value ◽  
Chest Ct ◽  
Ct Scans ◽  
Chest Computed Tomography ◽  
Ct Data

Lung ultrasound demonstrates a high diagnostic value in the assessment of lung diseases.Aim. To determine the diagnostic accuracy of lung ultrasound compared to chest computed tomography (CT) in the diagnosis of lung changes in COVID-19. Materials and methods. The retrospective study included 45 patients (28 men) aged 37 to 90 years who underwent polypositional lung ultrasound with an assessment of 14 zones. The study compared lung echograms with chest CT data in assessing the prevalence of the process and the nature of structural changes. The diagnostic accuracy, sensitivity, and specificity of lung ultrasound in comparison with CT scans were determined, 95% confidence intervals (CI) were calculated.Results. In 44 patients (98%), CT revealed pathological changes with subpleural localization in both lungs. Of these, in 30 cases, the inflammation was limited only to the subpleural parts, and in 14 cases, the changes spread to the basal parts of the lungs, while ultrasound revealed changes at the depth of the lesion no more than 4 cm. The lesion of 10–11 zones according to lung ultrasound corresponds to CT 1–2 degrees, the lesion of 13–14 zones — CT 3–4 degrees. The sensitivity of ultrasound to detect lung changes of various types was ≥ 92%. The highest sensitivity of 97.9% (95% CI: 92.8–99.8%) was determined for small consolidations on the background of interstitial changes (degree 1A+, 1B+), which corresponded to “crazy-paving” pattern on CT. The specificity depended on the nature of the changes and varied from 46.7 to 70.0%. Diagnostic accuracy was ≥ 81%, the maximum values of 90.6% (95% CI: 85.6–94.2%) were obtained for moderate interstitial changes (grade 1A) corresponding to ground-glass opacity (type one) according to CT data.Conclusion. The sensitivity of ultrasound to detect lung changes in COVID-19 is more than 90%. Lung ultrasound has some limitations: inability to determine the prevalence of the process clearly and identify centrally located areas of changes in the lung tissue.

scholarly journals Agreement between low-dose and ultra-low-dose chest CT for the diagnosis of viral pneumonia imaging patterns during the COVID-19 pandemic

2022 ◽  
Vol 53 (1) ◽  
Author(s):  
Hooman Bahrami-Motlagh ◽  
Yashar Moharamzad ◽  
Golnaz Izadi Amoli ◽  
Sahar Abbasi ◽  
Alireza Abrishami ◽  
...  
Keyword(s):  
Radiation Dose ◽  
Ct Scan ◽  
Predictive Value ◽  
Low Dose ◽  
Ground Glass Opacity ◽  
Chest Ct ◽  
Viral Pneumonia ◽  
Ct Scans ◽  
Low Dose Ct ◽  
Chest Ct Scan

Abstract Background Chest CT scan has an important role in the diagnosis and management of COVID-19 infection. A major concern in radiologic assessment of the patients is the radiation dose. Research has been done to evaluate low-dose chest CT in the diagnosis of pulmonary lesions with promising findings. We decided to determine diagnostic performance of ultra-low-dose chest CT in comparison to low-dose CT for viral pneumonia during the COVID-19 pandemic. Results 167 patients underwent both low-dose and ultra-low-dose chest CT scans. Two radiologists blinded to the diagnosis independently examined ultra-low-dose chest CT scans for findings consistent with COVID-19 pneumonia. In case of any disagreement, a third senior radiologist made the final diagnosis. Agreement between two CT protocols regarding ground-glass opacity, consolidation, reticulation, and nodular infiltration were recorded. On low-dose chest CT, 44 patients had findings consistent with COVID-19 infection. Ultra-low-dose chest CT had sensitivity and specificity values of 100% and 98.4%, respectively for diagnosis of viral pneumonia. Two patients were falsely categorized to have pneumonia on ultra-low-dose CT scan. Positive predictive value and negative predictive value of ultra-low-dose CT scan were respectively 95.7% and 100%. There was good agreement between low-dose and ultra-low-dose methods (kappa = 0.97; P < 0.001). Perfect agreement between low-dose and ultra-low-dose scans was found regarding diagnosis of ground-glass opacity (kappa = 0.83, P < 0.001), consolidation (kappa = 0.88, P < 0.001), reticulation (kappa = 0.82, P < 0.001), and nodular infiltration (kappa = 0.87, P < 0.001). Conclusion Ultra-low-dose chest CT scan is comparable to low-dose chest CT for detection of lung infiltration during the COVID-19 outbreak while maintaining less radiation dose. It can also be used instead of low-dose chest CT scan for patient triage in circumstances where rapid-abundant PCR tests are not available.

Sign in / Sign up

Export Citation Format

Share Document