EAI Endorsed Transactions

on Pervasive Health and Technology Research Article

Skin Disease Classification Using CNN Algorithms

Raghav Agarwal1 and Deepthi Godavarthi2, *

1
School of Computer Science & Engineering (SCOPE), VIT-AP University, Amravati, Andhra Pradesh, India
2
School of Computer Science & Engineering (SCOPE), VIT-AP University, Amravati, Andhra Pradesh, India

Abstract

INTRODUCTION: Dermatological disorders, particularly human skin diseases, have become more common in recent
decades. Environmental factors, socioeconomic problems, a lack of a balanced diet, and other variables have all
contributed to an increase in skin diseases in recent years. Skin diseases can cause psychological suffering in addition to
physical injury, especially in people with scarred or disfigured faces.
OBJECTIVES: The use of artificial intelligence or computer-based technologies in the detection of face skin disorders has
advanced dramatically over time. Even for highly experienced doctors and dermatologists, identifying skin disorders can
be tricky since many skin diseases have a visual affinity with the surrounding skin and lesions.
METHODS: Today, the majority of skincare specialists rely on time-consuming, traditional methods to identify disorders.
Even though several research have demonstrated promising results on the picture classification job, few studies compare
well-known deep learning models with various metrics for categorizing human skin disorders.
RESULTS: This study examines and contrasts various skin illnesses in terms of cosmetics and common skin concerns.
Our dataset includes over 25000 of the eight most common skin disorders. Convolutional neural networks have shown
imaging performance that is comparable to or greater than that of humans. We used 11 different network algorithms to
identify the illnesses in the sample and compared the results.
CONCLUSION: To adjust the format of incoming photographs, we do certain image pre-processing and image scaling for
each model. ResNet152 beat other deep learning methods in terms of recall, accuracy, and precision on a test dataset of
1930 images.

Keywords: Skin disease classification, Transfer Learning, Deep Learning, Medical Images, Clinical Decision Support System

Received on 19 June 2023, accepted on 07 September 2023, published on 02 October 2023

Copyright © 2023 R. Agarwal et al., licensed to EAI. This is an open access article distributed under the terms of the CC BY-NC-SA
4.0, which permits copying, redistributing, remixing, transformation, and building upon the material in any medium so long as the
original work is properly cited.

doi: 10.4108/eetpht.9.4039

Corresponding author. Email: deepthi.g@vitap.ac.in

*

doctor unless they have a serious illness. As a result,

addressing skin diseases too late may worsen the situation
1. Introduction and exacerbate the symptoms.
According to the 2018 English Skin Establishment
Human skin disease is a chronic ailment that affects people Report, 5.4 million new cases of skin sickness are recorded
of all ages all over the world. Skin disorders that are severe in the United States each year, and one in every five persons
enough to necessitate medical treatment or that receive is believed to have a lifelong cutaneous risk. An estimated
insufficient attention might result in bodily damage or even 60% of English people suffer from a skin ailment. Skin
death. Clinicians must still detect most skin problems by problems can be fatal and severely impair a person's daily
hand and examine the patient's symptoms. However, the activities, interpersonal interactions, and internal organs.
following factors make it difficult to diagnose skin This illness may also be characterized by dysfunctional
conditions. Skin disease classification necessitates a behaviors that may lead to suicide, social isolation, or even
considerable amount of information; thus, a diagnosis given despair.
by a non-expert doctor may result in inadequate therapy or Skin disorders affect 10% to 12% of Indians. The skin
other complications. Second, reaching hospitals or basic functions as a sensor of its environment in addition to
treatment in rural places might be difficult, especially during providing physical protection. It is the biggest organ in the
the COVID-19 epidemic. Furthermore, few people go to the

EAI Endorsed Transactions on

Pervasive Health and Technology
1 2023 | Volume 9
R. Agarwal and D. Godavarthi

human body, with seven layers of ectodermic tissue automated procedures usually make use of machine learning
protecting bones, muscles, and internal organs. Poor technology based on artificial intelligence. This paper
personal hygiene, rising pollution, climate change, and presents an automated deep learning-based technique for
hazardous UV radiation can all aggravate skin disorders. diagnosing skin diseases. Diseases are predicted using a
Cancer incidence may increase by two to three percent for deep convolutional neural network (CNN).
every percentage point decrease in ozone. In India, The following is the paper's structure: In this section,
photosensitive and infectious skin disorders are extremely beginning with section 2, we have provided a quick synopsis
widespread. It is critical to address skin disorders as soon as of the literature review. Section 3 then outlines the article's
they emerge to prevent complications that impact more than strategy. Section 4 presents the results and opinions. Section
just the skin. People's quest for efficient solutions is 5 closes the papers with a brief comment on future research.
motivated by the following reasons: Given India's explosive
population growth, it is critical to begin providing high-
quality care to each individual as soon as possible. Skin
problems mirror diseases such as AIDS, tuberculosis, and
others that can be fatal if not treated promptly. Even mild
skin issues are expensive, limiting treatment possibilities. As
a result, it is critical to create economical and effective
approaches for identifying skin disorders. Manual traditional
equipment, including that employed in the medical sector
and other disciplines, has been largely supplanted by
automated technology in the modern world.
No age group is more vulnerable to skin disorders. Skin
disorders affect people of all ages, including babies and the
elderly. Skin disorders can be difficult to identify, especially
when many conditions display the same or similar
symptoms. Because of the intricacy of the skin's structure
and the illnesses' seeming proximity, determining the actual
kind of skin problem can be difficult. Diagnosing skin
disorders can be difficult due to a variety of variables,
including the difficulty in segmenting and examining the
skin due to the presence of hair, sweat, and other
unattractive aspects. Because many skin conditions resemble
one another, different dermatological conditions can be
difficult to distinguish from colored photographs. This is a
significant problem for computer vision. One of the biggest
disadvantages of large-scale biomedical image processing
features is the difficulty in structuring and extracting critical
information from data. Digital photographs taken with a
camera, which may have noisy images, poor quality, or
inconsistent lighting, aggravate skin problems. Because of
lesion irregularities such as skin and hair coloring, it may be
difficult to detect skin disorders. [1–5] Skin issues [6] are Figure 1. Sample representation of the dataset.
usually unexpected and difficult to diagnose. Some skin
disorders may be difficult to identify and categorize. Many
people in other nations cannot afford to see a dermatologist. 2. Related Work
[7]. As a result of mobile phone use, skin disease detection
has become more economical in underdeveloped nations. A lot of previous articles have discussed dermatological
Image processing in devices such as cameras and cell categorization. [9] carried conducted the study by improving
phones is used to evaluate skin disorders. We can address an a deep convolutional neural network (CNN) that has
issue in two steps by developing a purpose. In the first stage, previously been taught to spot skin cancer. [10] used the
image processing is used to identify skin problems, and then DermNet dataset to test the efficiency of traditional
a deep learning system is used. Skin illnesses are difficult to machine-learning approaches.
diagnose in both the early and late stages due to changes in [11] used transfer learning with AlexNet to identify a
the distinguishing properties of the skin, such as color and melanoma, typical nevi, and atypical nevi. [12] used transfer
texture [8]. Deep learning technologies developed for learning with VGGNet to classify skin lesions as benign or
analyzing skin disease sample data might be applied to solve malignant. [13] used transfer learning on Google Net to
this problem. classify the eight basic kinds of skin illnesses (basal cell
The use of computer-assisted technology in this industry carcinoma, melanoma, melanocytic nevus, and so on).
facilitates diagnostics and helps to reduce mistakes. These

EAI Endorsed Transactions on

Pervasive Health and Technology
2 2023 | Volume 9
 Skin Disease Classification Using CNN Algorithms

[14] used Google Net as a transfer learning strategy to 3. Methodology

classify different skin diseases. In [15], many deep learning
models were utilized, including VGG16, MobileNet,
VGG19, and InceptionV3. 3.1. Data Collection
[16] proposed developing an Android app that could
diagnose seven ailments using transfer learning and the The dataset serves as the foundation for all algorithms,
MobileNet methodology. They employed an uneven dataset models, and systems. Approximately 25000 color images of
using data augmentation strategies such as under-sampling, skin disorders were imported for this project using several
oversampling, and default preprocessing. They use a simple Kaggle [24] datasets, each of which focuses on a different
technique and have encouraging results; however, they don't type of skin disease. Fig.1 shows the sample representation
provide the datasets they used or the specific categories of of the dataset.
skin illnesses that were useful. The approach has a 94.4% Eight common skin conditions are covered by around
accuracy rate. 25000 colored photographs in the collection:
[17] categorized seven disorders using two alternative dermatofibroma, melanocytic nevus, melanoma, squamous
methods. The first way is the Alone CNN model, while the cell carcinoma, actinic keratosis, basal cell carcinoma,
second is the One-versus-All and CNN strategy. The benign keratosis, and vascular lesion. The photographs in
preprocessing step is skipped in this manner. Combining the the collection depict various body parts. Following the
one-versus-all-and-alone CNN themes is smart and import of the dataset, 80% of the images are used for
wonderful. It was feasible to achieve accuracy rates of up to training, 10% for validation, and 10% for testing.
92.9%.
[18] described a strategy for categorizing the seven skin
disorders. The steps of this CNN-based approach are pre-
3.2. Pre-processing
processing, a deep learning algorithm, model training,
Image pre-processing is the process of improving image
model validation, and classification. They didn't say
quality and producing the desired outcomes. Images may
anything about the conditions or the dataset they utilized.
contain disturbing components such as air bubbles, hair, and
Neither CNN nor data pre-processing was mentioned.
other noises. Image Pre-processing includes:
Approximately 93% of the categories were correct.
[19] used the fine-tuning direct technique and pre-trained
Image resizing
Google's EfficientNet-b4 with seven auxiliary classifiers
Because the gathered photos come in a range of sizes, the
applied to each intermediate layer group to identify fourteen
input image into the recommended CNN is shrunk to 224 by
diseases with 94.8% accuracy. To expose the hidden image
224 dimensions. The image's data was still intact at this size.
properties that were explored, the proposed approach
employs t-distributed Stochastic Neighbour Embedding. The
Image Standardization
dataset and sickness kind are not supplied.
As part of data normalization, the input image is turned into
ANN, CNN, KNN, and GAN are included in [20] for
a set of pixels in the [0,1] range. The pixel values of each
classifying graphic images. Each calculation has its
picture, which vary from 0 to 255, are normalized to a range
advantages and merits. Deep CNN performs comparative
of pixels from 0 to 1 by dividing each pixel by 255.
analysis with the highest accuracy of 98.79%. It gives the
best results compared to other algorithms.
For the two-class classification of [21], the accuracy 3.3. Methods
achieved by REsNet50 is 50.50%, VGG is 82.49%, SVM is
83.48%, Vit 84.31%, and CNN 97.61%. Transfer learning is a strategy for enhancing model
Regarding three-class classification, [22] studied many performance on smaller datasets by reusing models trained
machine learning and deep learning algorithms. The support on larger datasets. In this part, we'll look at ways to improve
vector machine achieved the highest accuracy among all classification accuracy by combining transfer learning with
machine learning approaches (92.54%), and the pre-trained previously trained CNN models and feature extraction from
Xception model achieved the highest accuracy (99.33%) MRI images.
among all deep learning models.
In [23], the CNN model Inception-ResNetV2 is mainly VGG16
used to solve the 23-class skin classification problem. The well-known CNN model architecture VGG16 was
Additionally, using a data augmentation approach mitigates introduced at ILSVRC-2014. Figure 2 depicts the model's
the overfitting problem. According to the data, Inception- construction. It outperforms AlexNet by sequentially
ResNetV2 achieves 54.68% accuracy, outperforming replacing many 3x3 kernel-sized filters with large kernel-
VGG16 and Inception-V3 models. sized filters. To build the VGG16 model, we first set the
input image size to 224x224. The top of the model is
removed, the global average layer is added, and a prediction
layer with softmax activation is added before importing the
pre-trained VGG16 model from Keras using the Imagenet

EAI Endorsed Transactions on

Pervasive Health and Technology
3 2023 | Volume 9
R. Agarwal and D. Godavarthi

dataset weights. We use the Adam optimizer, cross-entropy applications. The MobileNet model employs depthwise
loss, and 50 iterations to train the model. separable convolutions to minimize processing costs and
parameter counts while retaining excellent accuracy.
VGG19 When compared to other deep learning models, the
To adapt the VGG19 model to our objective, the input layer MobileNet model is relatively simple, with only 28 layers.
will be modified to take images as input. The model's output The ImageNet dataset was used to pre-train the MobileNet
layer will be replaced with a new softmax layer with eight model, which provides an appropriate initialization for
nodes to represent the various forms of skin disease. To transfer learning to our aim of recognizing skin conditions.
focus on our aim, we will alter the weights of the remaining
layers while freezing the weights of the model's initial few Xception
layers, which are in charge of feature extraction. Because it The Xception model has excelled in a variety of image
has more layers than the VGG16 model, the VGG19 model classification applications, including medical image
can capture images with more complex attributes. analysis. By utilizing depth-wise separable convolutions, the
model's capacity to generalize to new data is improved while
ResNet the number of parameters is reduced.
In this study, three alternative ResNet models—ResNet50, We classified MRI scans for skin cancer into groups with
ResNet101, and ResNet152—were employed. Deep distinct subgroups in this investigation. The model's
convolutional neural network models known as ResNet hyperparameters will be changed, and its effectiveness will
models have succeeded in image classification applications. be measured using accuracy, recall, and precision. We will
The skip connections in the ResNet models allow the train, validate, and test the model using the Kaggle dataset,
models to create and learn residual functions. which includes images of skin cancers. Table 1 lists the
ResNet50 has 50 layers and is somewhat shallower than hyperparameters reported in each model.
the other ResNet models. ResNet101 is a deeper model than
ResNet50 since it includes 101 layers. ResNet152 is the
most complex ResNet model, with 152 layers. The ResNet Table 1. Hyperparameter for models
models were pre-trained using the ImageNet dataset, which
offers a solid basis for employing transfer learning to fulfill
our aim of recognizing skin illness. Hyper-parameter Value

InceptionV3 Image size 224 x 224

The InceptionV3 model's convolutional neural network
design was initially revealed at the ImageNet Large-Scale Weight ImageNet
Visual Recognition Challenge. To extract complex
characteristics from input data, the InceptionV3 model Epochs 50
includes convolutional layers with variable filter sizes and
average pooling layers. Batch size 32
The ImageNet dataset's more than one million annotated
Optimizer Adam
pictures from 1,000 categories were used as pre-training
data for the InceptionV3 model's 48 convolutional layers. Learning rate 0.001
DenseNet Loss Categorical cross-entropy
In our analysis, we employed three distinct DenseNet
models: DenseNet121, DenseNet169, and DenseNet201.
DenseNet models, a type of convolutional neural network,
primarily addressed the vanishing gradient issue in deep 3.4. Fine-tuning
networks. DenseNet models incorporate direct skip
connections between each layer, which improves gradient Transfer learning models' pre-trained weights were
flow and network information flow during backpropagation. downloaded. The models are made up of an input layer, a
DenseNet121 has 121 layers, DenseNet169 has 169 layers, convolutional layer, a batch normalization layer, and an
and DenseNet201 has 201 layers. As the number of layers activation layer. There were no obvious higher strata. The
grows, models get more complicated and contain more Transfer learning model's layers were completely frozen, but
layers. Because all DenseNet models were previously the weights survived. The Transfer learning model has been
trained on the ImageNet dataset, transfer learning provides a pre-trained except for the final layer. The output layer is
solid foundation for classifying skin disorders. made up of the global_average_pooling layer and a dense
layer with a softmax activation function. The Adam
MobileNet optimizer and the categorical_crossentropy loss function
In our research, we employed the MobileNet model. were used to generate the model [25]. The model was
MobileNet, a lightweight convolutional neural network trained using 50 epochs and 32 batches. The model's
model, was created for embedded and mobile vision

EAI Endorsed Transactions on

Pervasive Health and Technology
4 2023 | Volume 9
 Skin Disease Classification Using CNN Algorithms

efficacy was assessed using the test set. Figure 2 depicts the proposed fine-tuned design.

Figure 2. Proposed Fine-tuned Architecture

4. Results & Discussion

4.1. Experimental Setup

The training plan used for this continuing work meets the
following criteria: AMD Ryzen 7 5800H with a 3.20 GHz
Radeon graphics processor. Our computer was built with a
512 GB SSD, a 64-bit operating system, and 32 GB of
RAM. The experiment was carried out using an NVIDIA
RTX 3050 GPU.

4.2. Performance Evaluation

Accuracy
This formula determines the proportion of occurrences
properly categorized in the test dataset. When there is a class
imbalance or when misclassifying one class would cost
more than misclassifying the other, accuracy might be
deceptive.
(1)

Precision and Recall

Precision is the ratio of observed positives to expected
positives, as opposed to recall, which is the percentage of
projected positives to actual positives. When there is a class
imbalance, these metrics are useful because they give a more
thorough knowledge of how each class is doing using the
model.

(2)
(3)

True positives (tP), true negatives (tN), false positives (fP),

and false negatives (fN) are the units of measurement for the
aforementioned metrics [26]. Fig.3 provides a proper Fig. 3. Precision, Recall, Accuracy, and Loss Plot: (a)
comparison of all deep learning models. VGG16 (b) VGG19 (c) ResNet50 (d) ResNet101 (e)
ResNet152 (f) Xception (g) DenseNet121 (h)

EAI Endorsed Transactions on

Pervasive Health and Technology
5 2023 | Volume 9
1st author’s surname • 2nd author’s surname • 3rd author’s surname

DenseNet169 (i) DenseNet201 (j) MobileNet (k) InceptionV3.

Table 2. Metric measurement of several transfer learning models

medical pictures [27]. Component extraction substantially

4.3. Results aids in the classification of skin disorders. We do
extensive testing using seven deep-learning approaches.
Deep learning techniques that were recently released were
used to analyze and automatically learn the distinguishing ResNet152 outperformed all others with a great accuracy
qualities required to find hidden patterns in raw data. We of 74.24% in validation and 73.01% in the test dataset,
used a large, eight-class dataset of dermatological precision of 75.45% in validation and 75.30% in the test
disorders that was freely available to the public for our dataset, recall of 73.04% in validation and 71.71% in the
study. We provide standard measures for the model test dataset.
prediction module, such as accuracy, precision, and recall, The bulk of writers concentrate on identifying and
to evaluate the competency of each model. According to categorizing a specific set of skin problems. The proposed
the outcomes of the investigations, all deep learning- strategy efficiently overcomes several difficulties,
based models outperform traditional machine learning including unequal distribution of photos for each kind of
models. Furthermore, we can classify skin diseases using skin disease in the dataset under consideration,
the Restnet152 model at a rate of nearly 75% across all differences in image lighting, similarities between some
metrics. of the skin conditions in the dataset, and others.
RestNet152 looks to beat all other deep learning Furthermore, because the method is used for multiple
models because the more layers it has, the better it is at body parts, training the network is more difficult.
detecting hidden properties in difficult input data. Table 2 Because there hasn't been much research in this field,
examines the precision, recall, and accuracy of each skin there's a lot of research fields for new categorization or
disease categorization model. automatic machine analysis methods to be developed.
Future performance enhancements will include increasing
the size of our database and introducing new deep neural
5. Conclusion and Future Remarks network models, both of which will increase the model's
accuracy.
Skin problems are a common and unpleasant ailment. As
a consequence of recent advances in deep learning in
medical imaging, we are encouraged to employ digital
images to diagnose skin issues. Transfer learning is one of
the most popular and advanced approaches for assessing

EAI Endorsed Transactions on

Pervasive Health and Technology
6 2023 | Volume 9
 Skin Disease Classification Using CNN Algorithms

References Learning," in IEEE Access, vol. 8, pp. 114822- 114832,

2020, doi: 10.1109/ACCESS.2020.3003890.
[1] S. Akyeramfo-Sam, A. addo Philip, Derrick Yeboah, [14] Bhadoria R.K., Biswas S. (2020) A Model for
Nancy Nartey, et. al., "A Web-Based Skin Disease Classification of Skin Disease Using Pretrained
Diagnosis Using Convolutional Neural Networks," I. J. Convolutional Neural Network. In: Mandal J.,
Information Technology and Computer Science, vol. 11, Mukhopadhyay S. (eds) Proceedings of the Global AI
pp. 54-60, 2019. Congress 2019. Advances in Intelligent Systems and
[2] D. A. Shoieb, S. M. Youssef, and W, M. Aly, Computing, vol 1112. Springer, Singapore.
"ComputerAided Model for Skin Diagnosis Using Deep https://doi.org/10.1007/978-981-15-2188-1_14.
Learning," Journal of Image and Graphics, vol. 4, no. 2, [15] Md. Aminur Rab Ratul, Mohammad Hamed Mozaffari,
2016. Enea Parimbelli, WonSook Lee, "Atrous Convolution with
[3] S. Kumar Patnaik, M, Singh Sidhu, Yaagyanika Gehlot, Transfer Learning for Skin Lesions Classification”.
Bhairvi Sharma, and P Muthu*, "Automated Skin Disease [16] J. Velasco, C. Pascion, J, Wilmar Alberio, Jonathan
Identification using Deep Learning Algorithm," Apuang, John Stephen Cruz, Mark Angelo Gomez,
Biomedical & Pharmacology Journal, vol. 11(3), pp. 1429- Benjamin Jr. Molina, Lyndon Tuala, August Thio-ac, and
1436, 2018. Romeo Jr. Jorda, "A Smartphone-Based Skin Disease
[4] R. Yasir, Md. Ashiqur Rahman, and Nova Ahmed, Classification Using MobileNet CNN," International
"Dermatological Disease Detection Using Image Journal of Advanced Trends in Computer Science and
Processing and Artificial Neural Network," in 8th Engineering, vol. 8, 2019.
International Conference on Electrical and Computer [17] K.l Polat, and K. Onur Koc, "Detection of the skin
Engineering 20-22 December 2014, IEEE 2014, Dhaka, Diseases from Dermoscopy Image Using the combination
Bangladesh, 2014. of Convolutional Neural Network and One-versus-All,"
[5] P. Bose, S.K. Bandyopadhyay, A. Bhaumik, and Dr. Journal of Artificial Intelligence and Systems, vol. 2, pp.
Sandeep Poddar, "Skin Disease Detection: Machine 80-97, 10 February 2020.
Learning vs Deep Learning," Preprints, 13 September [18] K. Sujay Rao, P.Suresh Yelkar, O. Narayan Pise, and Dr.
2021. Swapna Borde, "Skin Disease Detection using Machine
[6] Jyothilakshmi K. K and Jeeva J. B, "Detection of Learning," [1] Kritika Sujay Rao, Pooja Suresh Yelkar,
malignant skin diseases based on the lesion segmentation," Omkar Narayan Pise, and Dr. Swapna Borde, "Skin
2014 International Conference on Communication and Disease Detection using Machine Learning", International
Signal Processing, 2014, pp. 382-386, doi: Journal of Engineering Research & Technology (IJERT)
10.1109/ICCSP.2014.6949867. ISSN: 2278- 0181 Published by, www.ijert.org, vol. 9, no.
[7] P. R. Hegde, M. M. Shenoy, and B. H. Shekar, 3, 2021.
"Comparison of Machine Learning Algorithms for Skin [19] C.Yu Zhu, Yu-Kun Wang, Hai-Peng Chen, Kun-Lun Gao,
Disease Classification Using Color and Texture Features," Chang Shu, Jun-Cheng Wang, Li-Feng Yan, Yi-Guang
2018 International Conference on Advances in Computing, Yang, Feng-Ying Xie, and Jie Liu, "A Deep Learning
Communications and Informatics (ICACCI), 2018, pp. Based Framework for Diagnosing Multiple Skin Diseases
1825-1828, doi: 10.1109/ICACCI.2018.8554512. in a Clinical Environment," Frontiers in Medicine, vol. 8,
[8] S. Kolkur and D. R. Kalbande, "Survey of texture based 16 April 2021.
feature extraction for skin disease detection," 2016 [20] V. Rawat, D. P. Singh, N. Singh, P. Kumar and T. Goyal,
International Conference on ICT in Business Industry & "A Comparative Study of various Skin Cancer using Deep
Government (ICTBIG), 2016, pp. 1-6, doi: Learning Techniques," 2022 International Conference on
10.1109/ICTBIG.2016.7892649. Computational Intelligence and Sustainable Engineering
[9] Aditya, W., Luthfil Hakim, N., Shih, T. K., Enkhbat, A., & Solutions (CISES), Greater Noida, India, 2022, pp. 505-
Thaipisutikul, T.. IC4Windows–hand gesture for 511, doi: 10.1109/CISES54857.2022.9844409.
controlling MS windows. 2020 - 5th International [21] I. K. Pious and R. Srinivasan, "A Review on Early
Conference on Information Technology (InCIT), 2020, 2. Diagnosis of Skin Cancer Detection Using Deep Learning
[10] Esteva, A., Kuprel, B., Novoa, R. A., Ko, J., Swetter, S. Techniques," 2022 International Conference on Computer,
M., Blau, H. M., & Thrun, S.. Dermatologist-level Power and Communications (ICCPC), Chennai, India,
classification of skin cancer with deep neural networks. 2022, pp. 247-253, doi:
Nature, 2017, 2–5. 10.1109/ICCPC55978.2022.10072274.
[11] K. M., M. A. Kassem, and M. M. Foaud, "Skin Cancer [22] T. R. Anik, P. Talukder, I. Faruki, I. S. Ibn Rahman and E.
Classification using Deep Learning and Transfer Learning Hossain, "Analysis of Automated Skin Disease
Hosny," 2018 9th Cairo International Biomedical Classification Exploiting Different Machine Learning
Engineering Conference (CIBEC), Cairo, Egypt, 2018, pp. Techniques," 2023 IEEE 13th Annual Computing and
90-93, doi: 10.1109/CIBEC.2018.8641762. Communication Workshop and Conference (CCWC), Las
[12] A. Romero Lopez, X. Giro-i-Nieto, J. Burdick and O. Vegas, NV, USA, 2023, pp. 1145-1151, doi:
Marques, "Skin lesion classification from dermoscopic 10.1109/CCWC57344.2023.10099354.
images using deep learning techniques," 2017 13th [23] X. Xiang and T. Chen, "Skin Disease Classification Using
IASTED International Conference on Biomedical Inception-ResNetV2 and Data Augmentation," 2022 14th
Engineering (BioMed), Innsbruck, Austria, 2017, pp. 49- International Conference on Computer Research and
54, doi: 10.2316/P.2017.852-053. Development (ICCRD), Shenzhen, China, 2022, pp. 42-47,
[13] M. A. Kassem, K. M. Hosny, and M. M. Fouad, "Skin doi: 10.1109/ICCRD54409.2022.9730242.
Lesions Classification Into Eight Classes for ISIC 2019 [24] Dataset link:
Using Deep Convolutional Neural Network and Transfer https://www.kaggle.com/datasets/bhanuprasanna/isic-2019
[25] Lakshmanaprabu, S. K., Mohanty, S. N., Shankar, K.,
Arunkumar, N., & Ramirez, G. (2019). Optimal deep

EAI Endorsed Transactions on

Pervasive Health and Technology
7 2023 | Volume 9
1st author’s surname • 2nd author’s surname • 3rd author’s surname

learning model for classification of lung cancer on CT

images. Future Generation Computer Systems, 92, 374-
382.
[26] Agarwal, R., Suthar, J., Panda, S. K., & Mohanty, S. N.
(2023). Fuzzy and Machine Learning based Multi-Criteria
Decision Making for Selecting Electronics Product. EAI
Endorsed Transactions on Scalable Information
Systems, 10(5). https://doi.org/10.4108/eetsis.3353
[27] Lakshmanaprabu, S. K., Mohanty, S. N., Krishnamoorthy,
S., Uthayakumar, J., & Shankar, K. (2019). Online clinical
decision support system using optimal deep neural
networks. Applied Soft Computing, 81, 105487.

EAI Endorsed Transactions on

Pervasive Health and Technology
8 2023 | Volume 9