scholarly journals COVID-CT-Mask-Net: Prediction of COVID-19 from CT Scans Using Regional Features

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
High Capacity ◽  
Ground Glass Opacity ◽  
Ct Scans ◽  
Regional Features ◽  
Bounding Boxes ◽  
Normal Sensitivity ◽  
Two Stages ◽  
And Control ◽  
And Training ◽  
Very High

AbstractWe present COVID-CT-Mask-Net model that predicts COVID-19 from CT scans. The model works in two stages: first, it detects the instances of ground glass opacity and consolidation in CT scans, then predicts the condition from the ranked bounding box detections. To develop the solution for the three-class problem (COVID, common pneumonia and control), we used the COVIDx-CT dataset derived from the dataset of CT scans collected by China National Center for Bioinformation. We use about 5% of the training split of COVIDx-CT to train the model, and without any complicated data normalization, balancing and regularization, and training only a small fraction of the model’s parameters, we achieve a 90.80% COVID sensitivity, 91.62% common pneumonia sensitivity and 92.10% normal sensitivity, and an overall accuracy of 91.66% on the test data (21182 images), bringing the ratio of test/train data to 7.06, which implies a very high capacity of the model to generalize to new data. We also establish an important result, that ranked regional predictions (bounding boxes with scores) in Mask R-CNN can be used to make accurate predictions of the image class. The full source code, models and pretrained weights are available on https://github.com/AlexTS1980/COVID-CT-Mask-Net.

scholarly journals COVID-CT-Mask-Net: Prediction of COVID-19 from CT Scans Using Regional Features

2020 ◽  
Author(s):  
Aram Ter-Sarkisov
Keyword(s):  
High Capacity ◽  
Ground Glass Opacity ◽  
Ct Scans ◽  
Regional Features ◽  
Bounding Boxes ◽  
Normal Sensitivity ◽  
Two Stages ◽  
And Control ◽  
And Training ◽  
Very High

Abstract We present COVID-CT-Mask-Net model that predicts COVID-19 from CT scans. The model works in two stages: first, it detects the instances of ground glass opacity and consolidation in CT scans, then predicts the condition from the ranked bounding box detections. To develop the solution for the three-class problem (COVID, common pneumonia and control), we used the COVIDx-CT dataset derived from the dataset of CT scans collected by China National Center for Bioinformation. We use about 5% of the training split of COVIDx-CT to train the model, and without any complicated data normalization, balancing and regularization, and training only a small fraction of the model’s parameters, we achieve a 90.80% COVID sensitivity, 91.62% common pneumonia sensitivity and 92.10% normal sensitivity, and an overall accuracy of 91.66% on the test data (21182 images), bringing the ratio of test/train data to 7.06, which implies a very high capacity of the model to generalize to new data. We also establish an important result, that ranked regional predictions (bounding boxes with scores) in Mask R-CNN can be used to make accurate predictions of the image class. 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 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.

Interference mitigation in point to point wireless sensor networks using LSP protocol and time division multiplexing approach

2020 ◽  
Vol 39 (4) ◽  
pp. 5449-5458
Author(s):  
A. Arokiaraj Jovith ◽  
S.V. Kasmir Raja ◽  
A. Razia Sulthana
Keyword(s):  
Energy Consumption ◽  
Large Scale ◽  
Control Parameter ◽  
Interference Mitigation ◽  
Wireless Sensor ◽  
Point To Point ◽  
A Minor ◽  
Two Stages ◽  
Time Division ◽  
And Control

Interference in Wireless Sensor Network (WSN) predominantly affects the performance of the WSN. Energy consumption in WSN is one of the greatest concerns in the current generation. This work presents an approach for interference measurement and interference mitigation in point to point network. The nodes are distributed in the network and interference is measured by grouping the nodes in the region of a specific diameter. Hence this approach is scalable and isextended to large scale WSN. Interference is measured in two stages. In the first stage, interference is overcome by allocating time slots to the node stations in Time Division Multiple Access (TDMA) fashion. The node area is split into larger regions and smaller regions. The time slots are allocated to smaller regions in TDMA fashion. A TDMA based time slot allocation algorithm is proposed in this paper to enable reuse of timeslots with minimal interference between smaller regions. In the second stage, the network density and control parameter is introduced to reduce interference in a minor level within smaller node regions. The algorithm issimulated and the system is tested with varying control parameter. The node-level interference and the energy dissipation at nodes are captured by varying the node density of the network. The results indicate that the proposed approach measures the interference and mitigates with minimal energy consumption at nodes and with less overhead transmission.

scholarly journals Portable DC Supply Based on SiC Power Devices for High-Voltage Marx Generator

Electronics ◽  
2021 ◽  
Vol 10 (3) ◽  
pp. 313
Author(s):  
Jacek Rąbkowski ◽  
Andrzej Łasica ◽  
Mariusz Zdanowski ◽  
Grzegorz Wrona ◽  
Jacek Starzyński
Keyword(s):  
High Voltage ◽  
Specific Area ◽  
Input Voltage ◽  
Marx Generator ◽  
Power Devices ◽  
Laboratory Setup ◽  
High Voltage Gain ◽  
Main Components ◽  
Two Stages ◽  
Very High

The paper describes major issues related to the design of a portable SiC-based DC supply developed for evaluation of a high-voltage Marx generator. This generator is developed to be a part of an electromagnetic cannon providing very high voltage and current pulses aiming at the destruction of electronics equipment in a specific area. The portable DC supply offers a very high voltage gain: input voltage is 24 V, while the generator requires supply voltages up to 50 kV. Thus, the system contains two stages designed on the basis of SiC power devices operating with frequencies up to 100 kHz. At first, the input voltage is boosted up to 400 V by a non-isolated double-boost converter, and then a resonant DC-DC converter with a special transformer elevates the voltage to the required level. In the paper, the main components of the laboratory setup are presented, and experimental results of the DC supply and whole system are also shown.

The effect of practicing yoga during pregnancy on labor stages length, anxiety and pain: a randomized controlled trial

2020 ◽  
Vol 0 (0) ◽  
Author(s):  
Ellahe Mohyadin ◽  
Zohreh Ghorashi ◽  
Zahra Molamomanaei
Keyword(s):  
Controlled Trial ◽  
Intervention Group ◽  
Total Duration ◽  
Measurement Data ◽  
Labor Pain ◽  
Control Group ◽  
Chi Square ◽  
Nulliparous Women ◽  
Two Stages ◽  
And Control

AbstractBackgroundAnxiety and fear of labor pain has led to elevated cesarean section rate in some countries. This study was conducted to investigate the effect of yoga in pregnancy on anxiety, labor pain and length of labor stages.MethodsThis clinical trial study was performed on 84 nulliparous women who were at least 18 years old and were randomly divided into two groups of yoga and control groups. Pregnancy Yoga Program consisting of 6 60-min training sessions was started every 2 weeks from week 26 of pregnancy and continued until 37 weeks of gestation. Anxiety severity at maternal admission to labor was measured by the Spielbergers State-Trait Anxiety Inventory, and labor pain was measured by Visual Analogue Scale (VAS) at dilatation (4–5 cm) and 2 h after the first measurement. Data were analyzed using Chi-Square and t-test.ResultsIntervention group reported less pain at dilatation (4–5 cm) (p=0.001) and 2 h after the first measurement (p=0.001) than the control group. Stat anxiety was also lower in intervention group than the control group (p=0.003) at the entrance to labor room. Subjects in the control group required more induction compared to intervention group (p=0.003). Women in intervention group experienced shorter duration of the first phase of the labor than the control group (p=0.002). Also, the total duration of two stages of labor was shorter in intervention group than the control group (p=0.003).ConclusionsPracticing yoga during pregnancy may reduce women’s anxiety during labor; shorten labor stages, and lower labor pain.

Sign in / Sign up

Export Citation Format

Share Document