Mathematical Modelling of Engineering Problems

Vol. 10, No. 4, August, 2023, pp. 1097-1104

Journal homepage: http://iieta.org/journals/mmep

Lung-Related Diseases Classification Using Deep Convolutional Neural Network

Joy Oluwabukola Olayiwola1,2 , Joke A. Badejo1,3 , Kennedy Okokpujie1,3* , Morayo E. Awomoyi4
1
Department of Electrical and Information Engineering, Covenant University, Ota 112101, Nigeria
2
Department of Computer Engineering, Federal Polytechnic, Ilaro 111101, Nigeria
3
Covenant Applied Informatics and Communication African Center of Excellence, Covenant University, Ota 112101, Nigeria
4
US School of International Service, American University, Washington, Tenleytown 20016, USA

Corresponding Author Email: [email protected]

https://doi.org/10.18280/mmep.100401 ABSTRACT

Received: 8 February 2023 Accurate diagnosis is a crucial first step in the management and treatment of lung
Revised: 28 March 2023 diseases, which include infectious diseases such as COVID-19, viral pneumonia, lung
Accepted: 6 April 2023 opacity, tuberculosis, and bacterial pneumonia. Despite these conditions sharing similar
Available online: 30 August 2023 manifestations in chest X-ray images, it is imperative to correctly identify the disease
present. This study, therefore, sought to develop a convolutional neural network (CNN)-
based model for the classification of lung diseases. Four distinct CNN models, namely
Keywords: MobileNetV2, ResNet-50, ResNet-101, and AlexNet, were rigorously evaluated for
deep learning, diagnosis, lung disease, their ability to classify lung diseases from chest X-ray images. These models were tested
MobileNetV2, ResNet-50, transfer learning against three classification schemes to examine the impact of high interclass similarity:
a 4-subclass classification (COVID-19, viral pneumonia, lung opacity, and normal), a
5-subclass classification (COVID-19, viral pneumonia, lung opacity, tuberculosis, and
normal), and a 6-subclass classification (COVID-19, lung opacity, viral pneumonia,
tuberculosis, bacterial pneumonia, and normal). The retrained ResNet-50 architecture
yielded the best results, achieving a classification accuracy of 97.22%, 92.14%, and
96.08% for the 6-subclass, 5-subclass, and 4-subclass classifications respectively.
Conversely, ResNet-101 demonstrated the lowest classification accuracy for the 6-
subclass and 5-subclass classifications, with 78.12% and 79.49% respectively, while
MobileNetV2 had the lowest accuracy for the 4-subclass classification, with 88.89%.
These results suggest that, despite high interclass similarity, the ResNet-50 model can
effectively classify lung-related diseases from chest X-ray images. This finding
supports the use of computer-aided detection (CAD) systems as decision-support tools
in the early classification of lung-related diseases.

1. INTRODUCTION severe pulmonary implications, including difficulty in

breathing among those infected. Moreover, the causative agent
Lung or pulmonary diseases constitute a spectrum of of COVID-19, along with other viruses or bacteria, can
pathological disorders that adversely affect the organs and precipitate pneumonia, a form of lung disease.
tissues instrumental in respiration, thereby impeding efficient The integration of radiological imaging procedures and
gas exchange. These afflictions compromise various computer-aided diagnosis (CAD) systems holds great promise
components of the respiratory system, including the pleurae, for improving the detection and classification of lung diseases.
the pleural cavity, respiratory nerves, breathing muscles, Given that these diseases primarily affect the lung passages,
bronchi, bronchioles, alveoli, and trachea [1]. The spectrum of the application of deep learning modalities for early detection
lung diseases spans minor, self-limiting conditions such as the and classification is particularly apt. The use of machine vision,
common cold, influenza, and pharyngitis, to life-threatening image-processing techniques, and deep learning algorithms as
diseases like bacterial pneumonia, lung opacity, tuberculosis, diagnostic tools can significantly reduce the risk of
acute asthma, lung cancer, and severe acute respiratory misdiagnosis [5]. These tools offer superior accuracy,
syndromes such as COVID-19 [2]. portability, and affordability, thereby enhancing their
The burden of lung diseases is particularly significant in applicability in clinical settings. This can substantially
developing and low- to middle-income countries, where mitigate lung-related health risks, especially in regions where
millions grapple with extreme poverty and poor air quality. As access to quality healthcare is limited. The deployment of
estimated by the World Health Organization, over four million CAD systems can potentially improve the quality of healthcare
individuals succumb prematurely each year to illnesses services, decreasing mortality rates amidst increasing disease
associated with lung diseases such as asthma and pneumonia. prevalence.
This underscores the urgency for effective preventive The concept of transfer learning in image classification
measures and diagnostic systems for early detection and posits that a model trained on a sufficiently large and diverse
management of lung diseases [3, 4]. dataset can serve as a generalized model of the visual world.
Since the outbreak in late 2019, COVID-19 has exhibited Once the feature maps are learned, they can be used to

1097
construct a different model using a significant dataset, and
eliminating the need to start from scratch. Transfer learning (2) The development of a CNN-based multi-classification
involves leveraging knowledge gained from one model to model for lung-related diseases using chest X-ray images from
improve the efficacy of another model [6, 7]. multiple sources to enhance the analysis of chest radiography
Accordingly, in this study, deep learning techniques images.
employing transfer learning were utilized to classify lung
diseases. Chest X-ray images were fed into deep learning
architectures (MobileNetV2, ResNet-50, ResNet-101, and 2. METHODS
AlexNet) to identify and extract the pertinent features of these
diseases. The objectives of this study were to curate hybrid The dataset that was used for this study were gotten from
chest X-ray image datasets of lung diseases from benchmark different publicly available online repository. The coronavirus
datasets, preprocess these images to address inconsistencies in Radiography Database by Chowdhury et al. [12] from Kaggle
data formats, develop a CNN-based model for classifying lung [13] were accessed and collated in order to fulfill the
diseases from these chest X-ray images, and evaluate the requirements of this investigation. 3,616 X-rays of chest
performance of this classification model. images belong to the COVID-19 class, 1,345 images of chest
In their seminal work, Zak and Krzyżak [8] presented a X-rays that show viral pneumonia, 6,012 X-rays images of
bifurcated deep learning approach for the classification of lung lung opacity, and 10,192 chest X-rays images that are normal
disorders (namely, pneumonia and tuberculosis) based on in this dataset. From this repository, 1,000 images were put to
chest X-ray images extracted from the Shenzhen dataset. The use from each class of images.
initial phase involved the utilization of U-Net to segment the Chest X-Ray for Tuberculosis is the second dataset. CXR
target lung region, followed by the employment of three images for Tuberculosis (TB) positive cases as well as normal
transfer learning models (InceptionV3, VGG16, and ResNet- images are included in the dataset from the Kaggle repository
50) with previously trained ImageNet weights. Notably, the that belongs to investigators from Qatar University in Doha,
InceptionV3 model exhibited superior performance with an Qatar, and the University of Dhaka, Bangladesh, as well as
accuracy and specificity of 82.00% and a sensitivity of 82.33%. their collaborators from Malaysia and in collaboration with
Similarly, Apostolopoulos and Mpesiana [9] employed medical professionals from Hamad Medical Corporation and
transfer learning for classifying lung disorders (normal, Bangladesh. This chest X-ray data is available in the study [14].
COVID-19, and pneumonia) using chest X-ray images from a The dataset includes 700 X-rays of chest images that were
publicly available dataset. They used five models (Inception, gotten for the tuberculosis class. Another tuberculosis data was
MobileNetV2, Xception, Inception ResNetV2, and VGG19), obtained from the TBX11K dataset [15] which consists of five
and VGG19 emerged as the most effective with a specificity categories of tuberculosis diseases in this dataset, i.e., Healthy,
of 98.75%, accuracy of 93.48%, and sensitivity of 92.85%. Sick but Non-TB, Latent TB, Active TB, and Uncertain TB,
In an innovative approach, Pham [10] combined three active TB dataset were extracted for this research. The last
databases, namely the COVID-19 Chest X-Ray Dataset dataset that was used for the research which is bacterial
Initiative, the COVID-19 Radiography Database, and the pneumonia was also obtained from the Kaggle repository [16].
IEEE 8023/COVID Chest X-Ray dataset, for conducting 2- The total dataset contains 6,000 datasets which were
class (COVID-19 and normal cases) and 3-class (COVID-19, divided into 6 classes. Each class in the dataset contains 1,000
viral pneumonia, and normal) classifications. Three pre- data samples. The data samples in each class were split in the
trained CNNs, AlexNet, SqueezeNet, and GoogleNet, were proportion of 80:20 to train and test, while 10 percent of the
incorporated into the network. The model achieved an training samples were used for validation to prevent
accuracy of 95%, specificity of 97%, and sensitivity of 90% in overfitting. Table 1 illustrates the distribution of the dataset.
detecting COVID-19. Notably, among the applied pre-trained Figure 1 depicts the images of chest X-rays of viral pneumonia,
CNNs, SqueezeNet and AlexNet demanded the least training COVID-19, normal, tuberculosis, lung opacity, and bacterial
and prediction time. pneumonia.
Further, an end-to-end learning method was proposed by
Kim et al. [11], wherein raw chest X-ray images were directly
input into a deep learning model (EfficientNet v2-M) using
transfer learning to extract significant features for lung disease
classification. The method was validated using three classes of
data from the National Institutes of Health (NIH) dataset:
normal, pneumonia, and pneumothorax. The validation results
demonstrated a loss of 0.6933, accuracy of 82.15%, sensitivity
of 81.40%, and specificity of 91.65%. The method was further
tested on a dataset from the Cheonan Soonchunhyang
University Hospital (SCH), which comprised four classes:
normal, pneumothorax, tuberculosis, and pneumonia. The
testing accuracy for the normal, pneumonia, pneumothorax,
and tuberculosis classes was reported as 63.60%, 82.30%,
82.80%, and 89.90%, respectively.
In light of the aforementioned studies, the present study
seeks to make the following contributions: Figure 1. Included in the sample images of chest X-ray used
(1) The curation of a hybrid chest X-ray image dataset in this study (a) COVID 19; (b) Pneumonia; (c) Normal; (d)
encompassing COVID-19, bacterial pneumonia, viral Tuberculosis positive; (e) lung-opacity; (f) bacterial
pneumonia, lung opacity, tuberculosis, and healthy instances; pneumonia

1098
 Table 1. Distribution of the dataset TensorFlow Keras library while ResNet101 and AlexNet were
implemented on PyTorch library on Intel(R) Core (TM) i5-
Class
Total Number of Training Validation
Testing
2,520M central processing unit operating at 2.50 GHz, and
CXIs/Class Set Set RAM comprised of 8GB 2,400 MHz DDR4 modules.
COVID-19 1000 720 80 200 The model was trained for 6-subclass classification
Viral (Coronavirus, tuberculosis, viral pneumonia, bacterial lung
1000 720 80 200
Pneumonia
opacity, and pneumonia) and 5-subclass classification
Tuberculosis 1000 720 80 200
Lung Opacity 1000 720 80 200
(COVID-19, viral pneumonia, bacterial pneumonia, lung
Bacterial opacity, and tuberculosis) for all the models. The model was
1000 720 80 200 trained for 4-subclass classification (COVID-19, Viral
Pneumonia
Normal 1000 720 80 200 Pneumonia, lung opacity, and bacterial pneumonia) for
MobileNetV2 and ResNet-50. Each model has a batch size of
The normalize function from the Python preprocessing 32, drop out of 0.6, a learning rate of 0.001, global average
library was used to normalize the data as part of the pooling, and Adam optimiser was employed for the precise
preprocessing of the dataset. It accepts an array as input and classification.
normalizes its values to range from 0 to 1. After that, it gives The models will be evaluated using five criteria which are
back an output array with the same size as the input array. accuracy, recall, precision, specificity, and F1-score. They are
Normalization is the process of rescaling attributes of real- as described in Eqs. (1)-(4):
valued images into a 0 to 1 range. To lessen the sensitivity of
model training to feature scale, deep learning uses data TP + TN
Accuracy = (1)
preprocessing. This makes it possible for the model in TP + TN + FP + FN
converging to improved weights, which leads to a more
accurate model. Also, the dataset was in different sizes, TP
therefore, they were pre-processed to the same size of Precision = (2)
224×224. TP + FP
The study was implemented with a transfer learning
technique by manipulating MobileNetv2, ResNet50, TP
Recall = (3)
ResNet101, and AlexNet as a base model and defined the new TP + FP
top layers as a fine-tuning model. The conventional pipeline
of transfer learning is initially to excerpt features from the 2* precision *Re call
source dataset and then fine-tune them on the targeted dataset. F1 Score = (4)
Pr ecision + Re call
Figure 2 shows the transfer learning pipeline.
where, TP (True Positive), TN (True Negative), FN (False
Negative) and FP (False Positive).
Also, confusion matrices analysis was used in validating the
model [17]. The confusion matrix provides extra details on the
illustration of the classification model. The confusion matrix
has two types of elements: diagonal and off-diagonal. The
diagonal elements depict how many points have predicted
labels that match actual labels, indicating that the point is
correctly classified. Off-diagonal elements represent how
many of the points for which the classifier mislabelled or
misclassified the data. The more accurate the predictions that
the model was able to make, indicated by the diagonal
numbers of the confusion matrix indicates the prediction that
the model was able to make accurately.
The model was iterated using the forward propagation as
well as the backward propagation utilising Adam optimiser.
The optimised repetitions of forward and backward
propagation making the model optimised for classification of
Figure 2. The transfer learning pipeline the lung diseases.
Adam Optimizer is defined as shown in Eq. (5):
Four cutting-edge pretrained networks—MobileNetv2,
ResNet50, ResNet101, and AlexNet—were used in this study Vt
in extracting the deep features from the previously trained wt =  *g (5)
networks. These networks were refined using the chest x-ray
St + 
dataset after being pre-trained on the ImageNet dataset,
containing one million images divided into 1,000 classes. In η: Initial learning rate;
this study, the chest X-ray images will be fed into each pre- Gt: Gradient at time t along wj;
trained network individually in extracting the highlighted Vt: Exponential average of squares of gradients along wj;
vectors at the fully connected layer. All training was done St: Exponential average of squares of gradients along wj; and
using the Python programming language in Jupyter Notebook. ∈: Smoothing term.
MobileNetv2 and ResNet50 was implemented with

1099
3. RESULT For each of the 6-subclass classifications, the
implementation for lung-related disease cases was trained and
In this section, the distinct metrics for the classification of validated. For all of them, the model was trained for 10 epochs
lung-related diseases were observed, followed by the accuracy with a batch size of 32. The results of the analysis of the 6-
observation. The model was trained in detecting and subclass lung-related diseases with a balanced dataset with the
classifying lung disease-confirmed cases by means of chest X- deep feature extraction pipeline are presented.
rays images. The dataset was built with a randomly chosen Figure 3 shows the trained and validated losses of each of
evenly distributed amount of chest x-rays from the chosen the models, whereas Figure 4 showed the trained and validated
repository to eliminate bias effects. Different dataset was used accuracy for each of the models. In each of the epochs,
for each of the training, validation, and testing. In essence, the ResNet50 had the lowest training loss and also had the highest
test dataset has not been known aforehand by the model, this training accuracy. This implies that the lung diseases were able
will make the model perform excellently on the new dataset. to train well on the ResNet50 model.

Figure 3. Trained and validated accuracy model for six-subclass classification

Figure 4. Trained and validated loss model for six-subclass classification

The confusion matrix for the 6-subclass classification of

lung-related disease dataset with MobileNetv2 is presented in
Figure 5. COVID-19, viral pneumonia, and bacteria
pneumonia achieved good classification performance in
comparison to the other classes of diseases. The model is able
to classify the occurrence of one lung disease. This will enable
the patient to quickly take adequate precautions. Furthermore,
classification measures for each class in terms of precision,
recall, specificity, F1-score were also determined (Table 2).
The highest precision was for COVID-19, viral pneumonia,
and bacterial pneumonia, which demonstrates the method's
potential for use in assisting with the quick classification of
bacterial, viral, and COVID-19 pneumonia. Table 3 displays
the ResNet50 results for macro average metrics for 6-subclass
classification. Nevertheless, Table 4 shows the MobileNetv2
results for macro average metrics for 5-subclass classification.
In addition, Figures 6-8 show the confusion matrix of
ResNet50, ResNet101 and AlexNet respectively. The models
have a good classification for each of the lung-related diseases.
Figure 5. Confusion matrix for MobileNetv2 for 6-subclass
classification for classification of lung-related diseases

1100
Table 2. MobileNetv2 results for macro average metrics for
6-subclass classification

Dataset Precision Recall F1-Score Support

Bacteria Pneumonia 0.93 1.00 0.96 186
Lung_Opacity 0.70 0.64 0.67 222
Viral Pneumonia 0.94 0.56 0.70 336
COVID-19 0.99 0.78 0.87 255
Normal 0.10 0.79 0.17 24
Tuberculosis 0.82 0.93 0.87 177

Figure 7. Confusion matrix for ResNet101 for 6-subclass

classification for classification of lung-related diseases

Figure 6. Confusion matrix for ResNet50 for 6-subclass

classification for the classification of lung-related diseases

Table 3. ResNet50 results for macro average metrics for 6-

subclass classification

Dataset Precision Recall F1-Score Support

Bacteria Pneumonia 0.94 0.99 0.97 191
Lung_Opacity 0.89 0.73 0.80 243
Viral Pneumonia 0.94 0.65 0.77 286
COVID-19 0.91 0.89 0.90 205
Normal 0.20 0.95 0.33 42
Tuberculosis 0.94 0.81 0.87 233

Table 4. MobileNetv2 results for macro average metrics for

5-subclass classification

Dataset Precision Recall F1-Score Support

Lung-Opacity 0.70 0.86 0.77 163
Viral Pneumonia 0.94 0.75 0.83 251
COVID 0.99 0.84 0.91 235 Figure 8. Confusion matrix for AlexNet for 6-subclass
Normal 0.55 0.91 0.69 121 classification for detection of lung-related diseases
Tuberculosis 0.76 0.62 0.68 170

Figure 9. Trained and validated loss model for five-subclass classification

1101
 Figure 10. Trained and validated accuracy model for five-subclass classification

Figure 11. Confusion matrix for MobileNetv2 for 5-subclass

classification for the classification of lung-related diseases
Figure 13. Confusion matrix for ResNet50 for 5-subclass
Moreover, MobileNetV2, ResNet50, ResNet101, and classification for detection of lung-related diseases
AlexNet architectures were trained on a total of 5,000 datasets
for five-subclass classification. Chest x-ray consists of 1,000
images each for the lung-related disease (Lung-Opacity,
COVID-19, Viral pneumonia, normal, and tuberculosis) with
the dataset divided in the proportion of 80:20 for training and
testing, the training dataset was additionally divided into
training and validation.
Figure 9 illustrates the trained and validated loss of each of
the model, whereas Figure 10 shows the trained and validated
accuracy for each of the model. In each of the epochs,
ResNet50 had lowest training loss and had the highest training
accuracy. This implies that the lung diseases were able to train
well on the ResNet50 model.

Figure 14. Confusion matrix for AlexNet for 5-subclass

classification for the classification of lung-related diseases

Figure 11 depicts Confusion matrix for MobileNetv2 for 5-

subclass classification for the classification of lung-related
diseases while Figure 12 shows the confusion matrix for
ResNet101 for 5-subclass classification for detection of lung-
related diseases while Figures 12 and 13 show the confusion
matrix for ResNet50 and ResNet101 for 5-subclass
classification for detection of lung-related diseases
respectively. Finally, Figure 14 shows the confusion matrix for
Figure 12. Confusion matrix for ResNet101 for 5-subclass AlexNet for 5-subclass classification for the classification of
classification for detection of lung-related diseases lung-related diseases.

1102
4. DISCUSSION accurately.
For six classifications, ResNet-50 outperformed
In this study, MobileNetV2 and ResNet50 were built as MobileNetV2 in terms of accuracy, as evidenced by the
lightweight deep-learning models for lung-related disease confusion matrices and average accuracies. In ten (10) epochs,
identification and classification in chest X-rays. Also, the accuracy of the ResNet-50 is 97%, whereas that of the
ResNet101 and AlexNet models were also trained using the MobileNetV2 is 86.35%. However, MobileNet's training has
same dataset with PyTorch libraries. The architecture of the the lowest time. In the process of increasing the number of
model is used in the classification of the most prevalent and epochs, accuracy will undoubtedly continue to rise. In
pandemic lung-related diseases: bacterial pneumonia, lung comparison to MobileNet, ResNet-50, ResNet101, and
opacity, viral pneumonia, COVID, normal, and tuberculosis. AlexNet, ResNet50 has the highest training and testing
The model compared the performance of the accuracy, recall, accuracy for the detection of lung-related diseases. The
precision, F1 score. The highest training accuracy was accuracy of each of the model is shown in Table 5.
achieved 97% and the testing accuracy is 94% indicating that Furthermore, Table 6 shown the comparison of the training
model is able to classify each of the lung-related diseases time for each of the model and classification.

Table 5. Comparison with earlier literature on classification of lung-related diseases with X-ray images of the chest

Author(s) Dataset Techniques Used Accuracy

5-subclasses:
435 COVID-19/ 439 normal/ 439
Al-Timemy et al. pneumonia bacterial/ 434 Tuberculosis/ 439 pneumonia viral The subspace discriminant 91.6
[18] 3-subclasses: classifier ensemble of Resnet50 98.6
435 COVID-19/ 434
Tuberculosis / 439 normal
5-subclasses
1770 COVID-19, 1436 Tuberculosis, 1345 viral pneumonia, 1700
bacterial pneumonia and 1341 normal
99.6
4-subclasses
Jia et al. [19] Modified MobileNet 99.9
1700 COVID-19, 1341, viral pneumonia, 1436 tuberculosis, 1341
99.7
healthy
3-subclasses
1770 COVID-19, 1341 viral pneumonia, and 1341 healthy
3-subclass:
Ozturk et al. [20] 125 COVID-19/ 500 DarkCOVIDNet CNN 87.2
Pneumonia/ 500 Normal
4-subclass:
Khan et al. [21] 284 COVID-19/ 310 normal/ 330 CoroNet CNN 89.6
pneumonia bacterial/ 327 pneumonia viral
MobileNetV2 92
6 subclasses
ResNet50 97
This study 1000 Bacteria pneumonia, 1000 Lung-Opacity, 1000 Viral pneumonia,
ResNet101 82
1000 COVID, 1000 normal, 1000 tuberculosis
AlexNet 85
5 subclasses MobileNetV2 84
1000 Lung-Opacity, 1000 Viral pneumonia, 1000 COVID, 1000 ResNet50 91
normal, 1000 tuberculosis ResNet101 79
4 subclasses AlexNet 84
1000 Lung-Opacity, 1000 Viral pneumonia, 1000 COVID, 1000 MobileNetV2 88
normal ResNet50 96

Table 6. Comparison of the training time diseases based on chest X-ray images. The overarching aim
was to augment the diagnostic capabilities of Computer-Aided
Model Classification Time (Sec) Diagnostics (CADs), particularly in terms of accuracy and
ResNet50 6- Subclass 190.18 efficiency.
MobileNetV2 6 - Subclass 71 Lung disorders exhibit chest X-ray characteristics that bear
ResNet50 5- Subclass 150.56 a striking similarity to each other, rendering early
MobileNetV2 5 - Subclass 40.86 classification a matter of paramount importance. In this
ResNet50 4- Subclass 189.53
MobileNetV2 4 - Subclass 34.54
context, the present study's focus on the utilization of pre-
trained networks, devoid of GPU support, emerges as a key
feature. A CPU-enabled computer was harnessed for the
efficient extraction of deep features from chest X-ray images.
5. CONCLUSION Moreover, the classification accuracy was evaluated across
six, five, and four subclass classifications. This study
In the present study, a multi-pronged approach was adopted, demonstrated the potential of pipelines requiring minimal
leveraging the capabilities of MobileNetv2, ResNet50, computational power, thereby contributing to the
ResNet101, and AlexNet models. These models, endowed classification of lung-related diseases using chest X-ray
with the previously trained weights of ImageNet and fine- images, particularly in scenarios where more sophisticated or
tuned, were deployed for the categorization of lung-related computationally intensive techniques might be unavailable.

1103
ACKNOWLEDGEMENT https://doi.org/10.1007/s13755-020-00135-3
[11] Kim, S., Rim, B., Choi, S., Lee, A., Min, S., Hong, M.
The authors acknowledge the part sponsorship of the (2022). Deep learning in multi-class lung diseases’
Covenant University Centre for Research, Innovation, and classification on chest X-ray images. Diagnostics, 12(4):
Discovery (CUCRID), Ota, Ogun State, Nigeria. 915. https://doi.org/10.3390/DIAGNOSTICS12040915.
[12] Chowdhury, M.E., Rahman, T., Khandakar, A., Mazhar,
R., Kadir, M.A., Mahbub, Z.B., Islam, M.T. (2020). Can
REFERENCES AI help in screening viral and COVID-19 pneumonia?
IEEE Access, 8: 132665-132676.
[1] Veale, D. (2006). Chronic respiratory care and https://doi.org/10.1109/ACCESS.2020.3010287
rehabilitation in France. Chronic Respiratory Disease, [13] COVID-19 radiography database.
3(4): 215-216. https://www.kaggle.com/tawsifurrahman/COVID19-
https://doi.org/10.1177/1479972306070070 radiography-database.
[2] Yimer, F., Tessema, A.W., Simegn, G.L. (2021). [14] Rahman, T., Khandakar, A., Kadir, M.A., Islam, K.R.,
Multiple lung diseases classification from chest X-ray Islam, K.F., Mazhar, R., Chowdhury, M.E. (2020).
images using deep learning approach. International Reliable tuberculosis detection using chest X-ray with
Journal of Advanced Trends in Computer Science and deep learning, segmentation and visualization. IEEE
Engineering, 10(5): 2936-2946. Access, 8: 191586-191601.
https://doi.org/10.30534/ijatcse/2021/021052021 https://doi.org/10.1109/ACCESS.2020.3031384
[3] Mondal, M.R.H., Bharati, S., Podder, P., Podder, P. [15] Liu, Y., Wu, Y.H., Ban, Y., Wang, H., Cheng, M.M.
(2020). Data analytics for novel coronavirus disease. (2020). Rethinking computer-aided tuberculosis
Informatics in Medicine Unlocked, 20: 100374. diagnosis. In Proceedings of the IEEE/CVF Conference
https://doi.org/10.1016/j.imu.2020.100374 on Computer Vision and Pattern Recognition, Seattle,
[4] Adetiba, E., Abolarinwa, J.A., Adegoke, A.A., Taiwo, WA, USA, pp. 2646-2655.
T.B., Ajayi, O.T., Abayomi, A., Badejo, J.A. (2022). https://doi.org/10.1109/CVPR42600.2020.00272
DeepCOVID-19: A model for identification of COVID- [16] Kermany, D.S., Goldbaum, M., Cai, W., Valentim, C.C.,
19 virus sequences with genomic signal processing and Liang, H., Baxter, S.L., Zhang, K. (2018). Identifying
deep learning. Cogent Engineering, 9(1): 2017580. medical diagnoses and treatable diseases by image-based
https://doi.org/10.1080/23311916.2021.2017580 deep learning. Cell, 172(5): 1122-1131.
[5] Xie, Y., Wu, Z., Han, X., Wang, H., Wu, Y., Cui, L., https://doi.org/10.1016/J.CELL.2018.02.010
Chen, Z. (2020). Computer-aided system for the [17] Ruuska, S., Hämäläinen, W., Kajava, S., Mughal, M.,
detection of multicategory pulmonary tuberculosis in Matilainen, P., Mononen, J. (2018). Evaluation of the
radiographs. Journal of Healthcare Engineering. confusion matrix method in the validation of an
https://doi.org/10.1155/2020/9205082 automated system for measuring feeding behaviour of
[6] Okokpujie, K., John, S., Ndujiuba, C., Badejo, J.A., cattle. Behavioural Processes, 148: 56-62.
Noma-Osaghae, E. (2021). An improved age invariant https://doi.org/10.1016/j.beproc.2018.01.004
face recognition using data augmentation. Bulletin of [18] Al-Timemy, A.H., Khushaba, R.N., Mosa, Z.M.,
Electrical Engineering and Informatics, 10(1): 179-191. Escudero, J. (2021). An efficient mixture of deep and
https://doi.org/10.11591/eei.v10i1.2356 machine learning models for COVID-19 and tuberculosis
[7] Okokpujie, K., John, S., Ndujiuba, C., Noma-Osaghae, E. detection using x-ray images in resource limited settings.
(2020). Development of an adaptive trait-aging invariant Artificial Intelligence for COVID-19, 77-100.
face recognition system using convolutional neural https://doi.org/10.1007/978-3-030-69744-0_6
networks. In Information Science and Applications: [19] Jia, G., Lam, H.K., Xu, Y. (2021). Classification of
ICISA 2019, Singapore, pp. 411-420. COVID-19 chest X-Ray and CT images using a type of
https://doi.org/10.1007/978-981-15-1465-4_41 dynamic CNN modification method. Computers in
[8] Zak, M., Krzyżak, A. (2020). Classification of lung Biology and Medicine, 134: 104425.
diseases using deep learning models. In International https://doi.org/10.1016/J.COMPBIOMED.2021.104425
Conference on Computational Science, Amsterdam, the [20] Ozturk, T., Talo, M., Yildirim, E.A., Baloglu, U.B.,
Netherlands, pp. 621-634. https://doi.org/10.1007/978-3- Yildirim, O., Acharya, U.R. (2020). Automated detection
030-50420-5_47 of COVID-19 cases using deep neural networks with X-
[9] Apostolopoulos, I.D., Mpesiana, T.A. (2020). COVID- ray images. Computers in Biology and Medicine, 121:
19: Automatic detection from x-ray images utilizing 103792.
transfer learning with convolutional neural networks. https://doi.org/10.1016/j.compbiomed.2020.103792
Physical and Engineering Sciences in Medicine, 43: 635- [21] Khan, A.I., Shah, J.L., Bhat, M.M. (2020). CoroNet: A
640. https://doi.org/10.1007/S13246-020-00865-4 deep neural network for detection and diagnosis of
[10] Pham, T.D. (2021). Classification of COVID-19 chest X- COVID-19 from chest x-ray images. Computer Methods
rays with deep learning: New models or fine tuning? and Programs in Biomedicine, 196: 105581.
Health Information Science and Systems, 9: 1-11. https://doi.org/10.1016/j.cmpb.2020.105581

1104