Multi-task semantic segmentation of CT images for COVID-19 infections using DeepLabV3+ based on dilated residual network

Abstract

COVID-19 is a deadly outbreak that has been declared a public health emergency of international concern. The massive damage of the disease to public health, social life, and the global economy increases the importance of alternative rapid diagnosis and follow-up methods. RT-PCR assay, which is considered the gold standard in diagnosing the disease, is complicated, expensive, time-consuming, prone to contamination, and may give false-negative results. These drawbacks reinforce the trend toward medical imaging techniques such as computed tomography (CT). Typical visual signs such as ground-glass opacity (GGO) and consolidation of CT images allow for quantitative assessment of the disease. In this context, it is aimed at the segmentation of the infected lung CT images with the residual network-based DeepLabV3+, which is a redesigned convolutional neural network (CNN) model. In order to evaluate the robustness of the proposed model, three different segmentation tasks as Task-1, Task-2, and Task-3 were applied. Task-1 represents binary segmentation as lung (infected and non-infected tissues) and background. Task-2 represents multi-class segmentation as lung (non-infected tissue), COVID (GGO, consolidation, and pleural effusion irregularities are gathered under a single roof), and background. Finally, the segmentation in which each lesion type is considered as a separate class is defined as Task-3. COVID-19 imaging data for each segmentation task consists of 100 CT single-slice scans from over 40 diagnosed patients. The performance of the model was evaluated using Dice similarity coefficient (DSC), intersection over union (IoU), sensitivity, specificity, and accuracy by performing five-fold cross-validation. The average DSC performance for three different segmentation tasks was obtained as 0.98, 0.858, and 0.616, respectively. The experimental results demonstrate that the proposed method has robust performance and great potential in evaluating COVID-19 infection.

MADR-Net: multi-level attention dilated residual neural network for segmentation of medical images

“…The disjoint sets are defined as follows: True positive (TP) set as , True negative (TN) set as , False positive (FP) set as and False negative (FN) set as . In medical segmentation, the region of interest will be too small compared to the entire image, so TP will be low, and the background or non-infected region will be represented as TN 64 . This may lead to misleading performance, and so to overcome the class imbalance, it is necessary to focus on DSC and IoU metrics that robustly reflect the performance of the model.…”

Brain functional connectivity based on phase lag index of electroencephalography for automated diagnosis of schizophrenia using residual neural networks

“…However, the deeper neural networks are more difficult to train due to the problem of vanishing and exploding gradients. 27 , 37 , 38 , 39 In this study, ResNet architecture, which overcomes these problems, is used for feature extraction and classification processes.…”

CADNet: an advanced architecture for automatic detection of coronary artery calcification and shadow border in intravascular ultrasound (IVUS) images

“…Every pixel in the image is classified using semantic segmentation into one of the specified classes [ 26 ]. In the medical field, deep learning-based methods are frequently used in the diagnosis of breast tumors [ 27 ], Covid-19 lung infection [ 28 ], coronary segmentation [ 19 – 21 , 29 – 35 ], Alzheimer disease prediction [ 36 ], skin lesion segmentation [ 37 ], dermatological diseases [ 38 ] and polyp segmentation [ 39 ] to mention a few. Table 1 depicts various areas of medical image segmentation where deep learning is widely used.…”