A deep convolutional structure-based approach for accurate recognition of skin lesions in dermoscopy images

International Journal of Electrical and Computer Engineering (IJECE)
Vol. 13, No. 5, October 2023, pp. 5792~5803
ISSN: 2088-8708, DOI: 10.11591/ijece.v13i5.pp5792-5803  5792
Journal homepage: http://ijece.iaescore.com
A deep convolutional structure-based approach for accurate
recognition of skin lesions in dermoscopy images
Shimaa Fawzy1
, Hossam El-Din Moustafa1
, Ehab H. AbdelHay1
, Mohamed Maher Ata2
1
Department of Communications and Electronics Engineering, Faculty of Engineering, Mansoura University, Mansoura, Egypt
2
Department of Communications and Electronics Engineering, MISR Higher Institute for Engineering and Technology,
Mansoura, Egypt
Article Info ABSTRACT
Article history:
Received Feb 26, 2023
Revised Apr 11, 2023
Accepted Apr 14, 2023
One-third of all cancer diagnoses worldwide are skin malignancies. One of
the most common tumors, skin cancer can develop from a variety of
dermatological conditions and is subdivided into different categories based
on its textile, color, body, and other morphological characteristics. The most
effective strategy to lower the mortality rate of melanoma is early
identification because skin cancer incidence has been on the rise recently. In
order to categorize dermoscopy images into the four diagnosis classifications
of melanoma, benign, malignant, and human against machine (HAM) not
melanoma, this research suggests a computer-aided diagnosis (CAD)
system. Experimental results show that the suggested approach enabled
97.25% classification accuracy. In order to automate the identification of
skin cancer and expedite the diagnosis process in order to save a life, the
proposed technique offers a less complex and cutting-edge framework.
Keywords:
Classification
Convolutional neural network
Deep learning
Feature extraction
Preprocessing
Skin cancer This is an open access article under the CC BY-SA license.
Corresponding Author:
Mohamed Maher Ata
Department of Communications and Electronics Engineering, MISR Higher Institute for Engineering and
Technology
Mansoura-35516, Egypt
Email: mmaher844@yahoo.com
1. INTRODUCTION
One of the major cancers, skin cancer, has had a rising prevalence over the past skin cancer is one of
the worst cancers and is the most common variety in the world. Over the past few decades, its prevalence has
increased. The aberrant expansion of cells is linked to the development of skin cancer. Melanoma, malignant,
human against machine (HAM), and the International Skin Imaging Collaboration (ISIC) are a few examples
of the various kinds of skin cancer. The most aggressive form of cancer among these several types is
melanoma, which spreads swiftly throughout the body, has a tendency to spread early, and often takes many
lives if it is discovered in the later stages. The presence of moles is a risk factor for melanoma. Most people
have benign moles or nevi, but some can increase the risk of melanoma. An expert dermatologist must
compare different skin lesions in order to make the diagnosis of skin cancer. Effective illness management
and therapy are made easier by prompt diagnosis [1].
Although cancer can exist anywhere on the body, skin cancer is a frequent kind that often manifests
in the skin that has been exposed to sunlight on a regular basis. Skin cancer is quite obvious since it starts in
the epidermis, the top layer of skin [2]. This shows that computer-aided diagnosis (CAD) systems may use
photos of skin lesions to make a preliminary diagnosis without considering any other pertinent data. The
performance of the dermoscopy imaging approach improved by 50%, aiding the specialist in the early
diagnosis of some kinds of skin cancer.

Int J Elec & Comp Eng ISSN: 2088-8708 
A deep convolutional structure-based approach for accurate recognition of skin lesions … (Shimaa Fawzy)
5793
Deep convolutional neural network (DCNN) algorithms have been extensively utilized in the
proposed study to analyze and correctly identify pigmented skin lesions in dermoscopy images, diagnose skin
lesions as early as possible, and demonstrate robust results. A big dataset labeled by a dermatologist and an
ensemble of many CNN models, including ResNet, DenseNet, MobileNet, VGG 19, Xception, EfficientNet,
and Inception-V3, demonstrate CNN’s supremacy. The suggested CNN-based deep neural network model
performed better than alternative methods in the classification of dermoscopy images.
In summary, the following are the paper’s major contributions:
− Using dermoscopy images, a CNN-established model is created that can accurately categorize the
patient’s type of skin cancer.
− In order to build a deep neural network (DNN), the validation set is subjected to a large number of
experimental trials in order to maximize the network’s depth. Sub-blocks are repeated in a specific ratio
to achieve this.
− The stride, number of kernels, and size of the filter are some of the parameters that each network block
uses to produce low and high-level quality information from lesions.
− Combining information and image features was also suggested as a way to increase classification
accuracy. Additionally, Adam optimizer was used to increase the proposed method’s effectiveness while
lowering the issue of hyper-tuning.
− The classification of skin lesions is investigated using a variety of pre-trained CNNs, including ResNet,
DenseNet, MobileNet, VGG 19, Xception, EfficientNet, and Inception-V3. The impacts of adding data
augmentation to all pre-trained CNN models under consideration are evaluated using a number of
evaluation metrics, such as the area under the receiver operator characteristic (ROC) curve, accuracy,
sensitivity, and precision, as well as the F1-score and computing time.
− The proposed model outperforms other cutting-edge techniques on the datasets while using fewer filters
and learnable parameters. As a result, it is a straightforward network for categorizing a huge dataset of
skin cancer cases.
A full article usually follows a standard structure: section 2 is the suggested system framework and
methods. Section 3 classification using CNN model architectures and performance method. Section 4 the
experimental results and examined. Section 5 brings the work to a close, discusses its limits, and offers
suggestions for further research on this topic.
In order to learn increasingly complex and fine-grained patterns from lesion photos,
Jaisakthi et al. [3] have presented the transport learning-based EfficientNet architecture. it automatically
increases the depth, size, and resolution of the network. According to the area under the curve, the suggested
system had a score of 0.9681. Ranger optimizer was used to improve EfficientNet-performance B6’s and
lessen the need to change hyperparameters.
Hameed et al. [4] have suggested a classification technique to categorize skin lesions into seven
classes using data augmentation and image preparation approaches. Various approaches were put forth in the
Dermatology pigmented lesion classification for the separation of melanocytic lesions from normal ones. The
proposed model had an accuracy rate of 92.5%. A comparison of the findings with previously published
methods on the same dataset.
Saifan and Jubair [5] have a method for categorizing color images of skin lesions using
convolutional neural networks. To distinguish between six skin conditions, it uses a DCNN that has already
been trained. Additionally, the holdout approach was utilized to calculate this accuracy, with 90% of the
images being used for training and 10% being used for out-of-sample accuracy testing. As an additional
interface to their proposed system, we created and implemented an Android application. Up to 81.75%
accuracy was attained, which is encouraging.
For the purpose of training images, Bhimavarapu and Battineni [6] suggested the vague-based
GrabCut-stacked convolutional neural networks (GC-SCNN) model. Lesion categorization and image feature
extraction were carried out on various publicly accessible datasets. The fuzzy GCSCNN combined with the
support vector machines (SVM) provided 100% sensitivity and specificity as well as 99.75% classification
accuracy. Results further show that compared to existing methods, the proposed model could more accurately
and quickly identify and classify the lesion parts.
Kaur et al. [7] suggest an automated melanoma classifier that can distinguish between malignant
and benign melanoma. The proposed DCNN classifier performed well, achieving accuracy rates on the ISIC
2016-2020 datasets of 81.41%, 88.23%, and 90.42%, respectively. In order to automate the detection of
melanoma and speed up the diagnosis process in order to save a life, this proposed approach may offer a less
complicated and sophisticated framework.
Salma and Eltrass [8] suggest a unique automated CAD system with excellent classification
execution employing accuracy low computing complication and using image processing approaches and data
augmentation is getting higher performance than collecting new images. The experimental results show that

 ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 5, October 2023: 5792-5803
5794
the suggested framework performs better than other modern methodologies in terms of the F1-score (97.3%),
the area under the ROC curve (99.52%), accuracy (99.87%), sensitivity (98.87%), and precision (98.77%). It
also takes less time to run (3.2 s), compared to other methodologies. This demonstrates how the suggested
structure might be put to use to aid medical professionals in categorizing various skin lesions.
Alkarakatly et al. [9] have suggested a 5-layer convolutional neural network (CNN). it aims to the
classification of skin lesions into three groups, including melanoma belonging to deadly skin cancer. On the
dataset that was created, the CNN-based classifier was trained and tested. The outcomes demonstrated high
accuracy. Rates were 95%, 94%, 97%, and 100% for accuracy, sensitivity, specificity, and area under the
curve (AUC).
Nawaz et al. [10] ground-breaking method incorporates a modern deep learning-based methodology,
and two examples are quicker region-based convolutional neural networks (RCNN) and fuzzy k-means
clustering (FKM). The method presented here first preprocesses the dataset photographs to reduce noise and
illumination concerns and enhance the visual information before learning using the quicker RCNN to create the
advantage vector with a constant length. The melanoma-affected skin region was then divided into parts of
varied sizes and shapes using FKM.
A fresh deep-learning method for the identification of melanoma is proposed by
Khouloud et al. [11] pre-processing, segmentation, and classification are the three phases that make up the
system. The invention of two new deep learning network architectures, W-net and Inception-Resnet, to tackle
the segmentation and classification problems, respectively. The recommended approach is more precise.
The skin lesion photos were classified using machine learning and CNN approaches in
Shetty et al. [12] proposed’s work. According to the findings, the customized CNN performed better at
classifying the given data set and had an accuracy of 95.18%. Seven groups of skin illnesses are made easier
to recognize early, which may be verified and properly treated by medical professionals over time.
2. METHOD
Medical diagnostics frequently make use of convolutional neural networks. It was trained on small
sample sizes of highly changeable, distinctive picture datasets, such as dermoscopic image datasets. The
neural network was used to create an automated system for categorizing various types of skin lesions. The
three main stages of the suggested framework for identifying skin lesions are pre-processing of dermoscopy
images, feature extraction, and classification. The block diagram of the proposed system framework is shown
in Figure 1.
Figure 1. Skin cancer classification based on the suggested system framework
2.1. Data preprocessing
The data pre-processing methods used to prepare the dataset for deep learning tasks are disputed in
this section, and the following image pre-processing steps were used in the framework [13].
− Step 1 Order the dataset: The dataset which comprises 24014 skin lesion images split into four types. The
Benign (ISIC) skin cancer dataset and the melanoma, malignant, not melanoma (HAM) dataset was used
in the proposed work.
− Step 2 Image resizing: There are various sizes with a resolution of (Benign: 224×224 pixels, Melanoma:
224×224 pixels, malignant: 224×224 pixels and Not Melanoma: 600×450 pixels) in the original skin
lesion images from the skin cancer dataset. Therefore, all images are scaled to the same size, which is
224×224, prior to training. After that, edge detection filters are applied to the images.

5795
− Step 3 Data augmentation: Small datasets result in models that overfit the training dataset, making it
impossible to generalize the findings. We used a data-augmentation technique to increase the dataset and
produce additional “data” in order to prevent this issue. to generalize more effectively in order to build deep
learning models and boost accuracy rates. The image generator has the ability to enhance data based on a
variety of criteria, including a rotation range of 40, image flipping (horizontally or vertically) of True, zoom
range of 0.2, and brightness range of (0.5, 1.5). As a result, models with data augmentation have a higher
likelihood of picking up more significant distinguishing qualities than models without data augmentation.
− Step 4 Data split: The dataset comprises 24,014 skin lesion images split into four types The Benign
contained 6,024 samples, the melanoma contained 7,056 samples, the malignant contained 6,479 samples,
and not melanoma (HAM) contained 4,455 samples. All of the datasets were split into a training set with
a ratio of 70%, a validation set with a ratio of 5%, and a test set with a ratio of 15%.
2.2. Feature extraction
The dimensionality reduction approach of feature extraction divides a starting set of raw data into
smaller groups that may be processed more easily. Feature extraction is a useful strategy when less
processing power is required without losing important or relevant data. Using feature extraction, it is possible
to reduce the amount of duplicate data for a given inquiry. Additionally, the speed of the learning and
generalization processes in the deep learning process, as well as the data reduction. Feature representation
vectors were created after CNN models were trained using pre-learned weights, which used the layers of max
pooling, flatten, and dense layers with a sigmoidal activation function.
2.3. Classification
Numerous automatic classification methods have tried to determine the kind of skin lesion based on
image analysis. Skin cancer detection is made easier for dermatologists and doctors by automatic
classification. In addition to training and testing the image dataset with a CNN model, a number of other
criteria, such as accuracy, precision, recall, and F1-score, were used to evaluate the performance [14].
3. PROPOSED CNN ARCHITECTURE
The specifics of the suggested CNN design are covered in this section. The primary objective was to
create the optimal CNN architecture for the test set that can predict the four classifications of skin lesions.
CNN is made up of many levels. The main types of layers used to create the suggested CNN architectures
included multi-convolutional, dropout, dense layers, pooling layers, and fully-connected layers in order to fit
an efficient model with greater performance than earlier architectures. The pre-processed image itself served
as the input, and the network automatically extracted the essential visual attributes from it.
The CNN architecture employed in this study is highlighted in Figure 2, which also shows the whole
structure of the convolutional model we propose. It features five convolutional layers with filters of sizes and
(153, 153, 512, 768, and 1,024) as well as input shapes of (124, 124, and 1) with kernels of size 5×5 for the
first four convolutional layers and 1×1 for the final convolutional layers. After each convolutional layer,
batch normalization is useful. After each convolution layer, we added a maximum pooling layer with a size
(2×2). In this model, a batch size of 32 was employed, the number of training epochs has been 50, the
learning rate of (0.0000001), and the network contains a total of 64,296,852 trainable parameters.
Figure 2. Proposed CNN layering system
The network is then made up of two dense layers that each include 1,024 and 512 units. The
convolutional layers maintain each neuron with a 0.3 probability of dropout regularization. The entire
network uses the rectified linear unit (ReLU) function as an activation function, while Adam, the study’s
optimizer, measures loss with the best precision possible using a cross-entropy function. Include L2

 ISSN: 2088-8708
5796
regularization (weight decay). Because it reduced training loss and eliminated over-fitting from the model,
the setting of (0.0001) produced the best results for us. The final layer of this model is a dense layer with a
“softmax” activation function. This activation function is utilized in the final dense layer to deliver the
multiclass classification commission’s most likely class for the input windows.
Algorithm 1 introduces the suggested system of the CNN model. the schematic for producing
discriminative and pertinent attribute interpretations for the cancer detection method is presented. The dataset
that was used is first given a brief explanation. Also included are preprocessing methods and the fundamental
architecture, along with the specifics of how the suggested model would be implemented.
Algorithm 1
Input: Reading in skin lesion image from the dataset.
Output: Skin cancer classification results, Confusion matrix, Accuracy, Precision, Recall,
F1-score.
1. Define hyper-parameter:
I=skin image, Aug=Augmentation, Pre=preprocessing, Rt=rotation, Sc=scaling, Zr=zoom range,
Sr=shear range, Hf=horizontal flip, ×train=training dataset, ytrain=label training dataset,
×test=testing dataset, ytest=label testing dataset, ypred=prediction data, ytrue =the ground
truth image.
Start Procedure
2. Browse (𝐼)
3. Apply (𝑃𝑟𝑒):
3.1: Resize (𝐼).
3.2: Aug (𝐼).
3.3: Normalize (𝐼).
4. Apply (Aug): Sc, Zr, Sr, Hf, Rt.
4.1: Perform Sc.
4.2: Perform Zr.
4.3: Perform Sr.
4.4: Perform Hf.
4.5: Perform Rt.
5. Split (dataset): Prepare training, testing, and validating.
6. Make a validation dataset from the training dataset.
7. Feature extraction (max pooling, flatten, dense layer, and sigmoidal function)
8. Adjust model parameters by adding
Model. add (Conv2D ())
Model. add (MaxPooling2D ())
Model. add (Dense ())
9. Set hyper-parameter
9.1: Batch size: 32
9.2: Epochs: 50
9.3: Optimizer: Adam
9.4: Learning rate: 0.0000001
10. Training the CNN model.
For 𝑘=1: numepochs
mm=randper(𝑖);
For 𝑙=1: numbatches
batch − ×= ×train (mm((l − 1) ∗ size + 1: l ∗ size), : );
batch − y = ytrain(mm((l − 1) ∗ size + 1: l ∗ size), : );
𝑍 = 𝑛𝑓(mm, batch − ×, batch − y)
End
End
Train the model
𝑚𝑜𝑑𝑒𝑙. 𝑓𝑖𝑡 ( ×train, ytrain)
11. Load the proposed model.
𝐹𝑜𝑟 𝐼 = 1: 𝑛𝑢𝑚 𝑡𝑒𝑠𝑡 𝑑𝑎𝑡𝑎𝑠𝑒𝑡𝑠
𝑚𝑜𝑑𝑒𝑙 𝑒𝑣𝑎𝑙𝑢𝑎𝑡𝑒 ( ×𝑡𝑒𝑠𝑡, 𝑦𝑡𝑒𝑠𝑡)
ypred = 𝑚𝑜𝑑𝑒𝑙 𝑝𝑟𝑒𝑑𝑖𝑐𝑡 ( ×test)
𝐴𝑐𝑐 = 𝑎𝑐𝑐𝑢𝑟𝑎𝑐𝑦_𝑠𝑐𝑜𝑟𝑒 ( 𝑦𝑡𝑒𝑠𝑡, 𝑦𝑝𝑟𝑒𝑑)
𝐿𝑜𝑠𝑠 = ( 𝑦𝑡𝑟𝑢𝑒 𝑙𝑜𝑔 ( 𝑦𝑝𝑟𝑒𝑑) + (1 − 𝑦𝑡𝑟𝑢𝑒)𝑙𝑜𝑔 (1 − 𝑦𝑝𝑟𝑒𝑑))
𝐶𝑜𝑚𝑝𝑢𝑡𝑒 (𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛, 𝑅𝑒𝑐𝑎𝑙𝑙, 𝐹1 − 𝑠𝑐𝑜𝑟𝑒)
𝐶𝑜𝑚𝑝𝑢𝑡𝑒 (𝐶𝑜𝑛𝑓𝑢𝑠𝑖𝑜𝑛 𝑚𝑎𝑡𝑟𝑖𝑥)
End
12. Classification of skin cancer images.
13. Prediction=classification (Train CNN, Test dataset)
14. Return prediction.
15. Train (ResNet, DenseNet, MobileNet, VGG 19, Xception, EfficientNet, and Inception-V3).
16. Compare the models.
17. Evaluation for all models: Compute (Confusion matrix, Accuracy, Precision, Recall, and
F1-score).
End Procedure

5797
4. COMPARISON WITH STATE-OF-THE ART CNN’s USED FOR SKIN LESION IMAGES
CNN has significantly advanced only image processing techniques. The classification of CNN
advancements includes regularization, design innovations, learning methods, and optimization [15]. The most
prevalent CNN architectures are viewed in this section as they progress.
− ResNet (residual network block), which has 152 layers, employs residual learning. It creates a quick
connecting procedure and an efficient method for deep network training [16].
𝑇𝑚+1
𝑘
= 𝑔𝑐(𝑇1→𝑚
𝑘
, 𝑘1→𝑚) + 𝑇𝑖
𝑘
𝑚 ≥ 𝐼 (1)
𝑇𝑚+1
𝑘
= 𝑔𝑎(𝑇𝑚+1
𝑘 ) (2)
𝑔𝑐(𝑇1→𝑚
𝑘
, 𝑘1→𝑚) = 𝑇𝑚+1
𝑘
− 𝑇𝑖
𝑘
(3)
where (𝑇𝑖
𝑘
) is an input of i the layer 𝑔𝑐(𝑇1→𝑚
𝑘
, 𝑘1→𝑚) , 𝑔𝑐(𝑇1→𝑚
𝑘
, 𝑘1→𝑚) is a transformed signal, the output
results (𝑇𝑚+1
𝑘
), and the next layer after adding the activation function 𝑔𝑎.
− DenseNet: the vanishing gradient issue is lessened by the DenseNet model, enhances feature propagation,
encourages feature reuse, and minimizes the number of parameters, which are all reasons why the
DenseNet design is well-liked [17]. All features in this architecture are concatenated in a sequential layer.
following is a definition of the concatenation procedure in mathematics:
𝑥1 = ∅1([𝑥0, 𝑥1, … , 𝑥𝑙−1]) (4)
where (∅1) is a nonlinear transform by a ReLU activation function. the convolution process of 3×3 is
([𝑥0, 𝑥1, … , 𝑥𝑙−1]), which refers to layer l-1.
− MobileNet: the inverted bottleneck MBConv is the fundamental component of the MobileNet family.
Since the MBconv block is an inverted residual block that contains layers that first extend and then spend
the channels, direct connections are employed between bottlenecks that connect fewer channels than
extension layers [18]. ReLU activation function was replaced with a new activation function called Swish
activation to increase performance.
− VGG was composed of 19 layers deep, in order to recreate the relationship between depth and the network’s
potential for imitation, the VGG was composed of 19 layers deep. The benefit of representation depth for
classification accuracy has been proven [19]. The use of 138 million parameters, which makes it extremely
expensive and challenging to deploy on low-resource technology, was the fundamental issue with VGG.
− Xception is a theory that produces cross-channel correlations and spatial linkages within CNN feature
maps that are completely decoupled. Swish, a new activation function, has been utilized to develop the
conventional activation function and to classify the initial diagnosis of skin cancer [20]. The following is
a mathematical formulation of the Swish activation function:
𝑆 = 𝑖 × 𝑠𝑖𝑔𝑚𝑜𝑖𝑑(𝜇 × 𝑖) (5)
where μ denotes a configurable per-channel value, i input dataset, and (μ×i) evaluation of the sigmoid
function.
− EfficientNet: They are known as EfficientNets because they outperform CNN in terms of accuracy and
efficiency, and Considering the depth, width, and resolution dimensions, a suitable scaling factor is
determined [21]. Depth: d=ε∂, width: w=α∂, resolution: r=μ∂. (ε≥1, α≥1, μ≥1) where ε, α, μ are constant
using a grid search, ∂ used as controllers availability of resources for model scaling.
− Inception-V3 is called GoogLeNet, a 22 layers-deep network, that is used to evaluate the performance of
classification and detection systems [22]. The goal was to lower the computational cost of deep networks
while maintaining generality.
5. PERFORMANCE EVALUATION METHODS
The usefulness of skin lesion cancer diagnosis is evaluated by calculating the appropriate accuracy,
arithmetic time, and complexity level. In this study, numerous evaluation criteria have been employed to
gauge how well the suggested system has performed at various phases [23]. We can determine how changing
a parameter will impact the model’s performance during the training process by looking into deep learning
techniques. The most prominent performance measurements are precision, F1-score, sensitivity (recall), and
accuracy. True positives (TP), false positives (FP), true negatives (TN), and false negatives are the four
variables needed by the evaluation methods (FN).

 ISSN: 2088-8708
5798
− Accuracy: this is the percentage of cases that were correctly identified out of all the cases.
𝐴𝑐𝑐𝑢𝑟𝑎𝑐𝑦 =
𝑇𝑃+𝑇𝑁
𝑇𝑃+𝑇𝑁+𝐹𝑃+𝐹𝑁
(6)
− Precision: it measures the proportion of accurately predicted positive outcomes to all its.
Precision =
𝑇𝑃
𝑇𝑃+𝐹𝑃
(7)
− Recall: it is the proportion of accurately predicted events among the foreseen data.
Recall =
𝑇𝑃
𝑇𝑃+𝐹𝑁
(8)
− F1-score: it is the average of recall and precision weighted together.
F1-score = 2 ×
𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 ×𝑅𝑒𝑐𝑎𝑙𝑙
𝑝𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛+𝑅𝑒𝑐𝑎𝑙𝑙
(9)
6. RESULTS AND DISCUSSION
Eight thorough tests based on various classical CNN deep learning models, including ResNet,
DenseNet, MobileNet, VGG 19, Xception, EfficientNet, and Inception-V3, as well as the suggested CNN
model, have been carried out in this study. The suggested CNN has been tested using the following
performance metrics: recall, F1-score, and precision. The PC used to analyze all trials had the following
specifications: Microsoft Windows 10 operating system, AMD Fx-8370, 8-core processor @ 4.0 GHz, 32 GB
of RAM, NVidia GeForce GTX1050 6GB GPU. The proposed system has been established in sate of art of
many types of skin lesions from Kaggle [24], [25].
6.1. Experiment 1: the traditional CNN models architectures
We implemented Eight distinct architectures to show the ability of CNN: ResNet, DenseNet,
MobileNet, VGG 19, Xception, EfficientNet, Inception-V3, and the suggested CNN model. In Table 1 (see in
Appendix), the results of CNN performance experiments employing model accuracy and weighted averages
of precision, recall, and F1-score are displayed. According to the results, EfficientNet had the lowest
accuracy (24%), followed by Xception (43%), DenseNet (48%), InceptionV3 (54%), ResNet50 (55%),
Mobile Net (57%), and VGG19 (57%), before proposed model (97.25%), which had the highest accuracy.
6.2. Experiment 2: the confusion matrix for the traditional CNN architectures
By training the skin lesion datasets, the suggested CNN model is tested to see if it can anticipate the
most effective optimizer to attain exceptional performance. With the aid of the Adam optimizer and sparse
categorical cross-entropy, we assembled and fitted the suggested model. Figure 3 shows the outcomes of the
accuracy and loss curves of the eight CNN architectures with the loss of the ResNet50 model in Figure 3(a)
and after the accuracy of the ResNet50 model in Figure 3(b), the loss of the DenseNet model in Figure 3(c)
and after the accuracy of the DenseNet model in Figure 3(d), the loss of the MobileNet model in Figure 3(e)
and after the accuracy of the MobileNet model in Figure 3(f), the loss of the VGG19 model in Figure 3(g)
and after the accuracy of the VGG19 model in Figure 3(h), the loss of the Xception model in Figure 3(i) and
after the accuracy of the Xception model in Figure 3(j), the loss of the EfficientNet model in Figure 3(k) and
after the accuracy of the EfficientNet model in Figure 3(l), the loss of the InceptionV3 model in Figure 3(m)
and after the accuracy of the InceptionV3 model in Figure 3(n), the loss of the Proposed model in Figure 3(o)
and after the accuracy of the proposed model in Figure 3(p).
Figure 4 shows the outcomes of the confusion matrix by comparing the benefits and cons of the
eight CNN architectures. The ResNet50 model is in Figure 4(a) and the DenseNet model is in Figure 4(b).
The MobileNet model is in Figure 4(c) and the VGG19 model is in Figure 4(d). The Xception model is in
Figure 4(e) and the EfficientNet model is in Figure 4(f). Finally, The InceptionV3 model is in Figure 4(g) and
the proposed model is in Figure 4(h).
The outcomes demonstrate that the suggested model architecture produces the greatest results.
A thorough comparison of all of these CNN architectures, including VGG-16, ResNet50, ResNetX,
InceptionV3, and MobileNet, shows that the suggested model architecture performs better and requires less
computing power. We have already looked at the majority of the pre-trained CNN structures, which are
widely known to exist.

5799
(a) (b) (c)
(d) (e) (f)
(g) (h) (i)
(j) (k) (l)
(m) (n) (o)
(p)
Figure 3. Training and validation versus the number of epochs for the traditional CNN architectures (a) loss
of ResNet50 model, (b) accuracy of ResNet50 model, (c) loss of DenseNet model (d) accuracy of DenseNet
model, (e) loss of MobileNet model, (f) accuracy of MobileNet model, (g) loss of VGG19 model,
(h) accuracy of VGG19model, (i) loss of Xception model, (j) accuracy of Xception model, (k) loss of
EfficientNet model, (l) accuracy of EfficientNet model, (m) loss of InceptionV3 model, (n) accuracy of
InceptionV3 model, (o) loss of proposed model, and (p) accuracy of proposed model

 ISSN: 2088-8708
5800
(a) (b) (c) (d)
(e) (f) (g) (h)
Figure 4. The confusion matrix for the traditional CNN architectures (a) ResNet50, (b) DenseNet,
(c) MobileNet, (d) VGG19, (e) Xception, (f) EfficientNet, (g) InceptionV3, and (h) proposed model
More computational training is needed for CNN models with increased depth. Using deeper layers also
introduces more free parameters, which could lead to over-fitting issues and performance decrease. The CNN
models chosen for this investigation reflect an appropriate trade-off between speed, accuracy, and diagnosis of
skin cancer. The data variability in the current study is lower than in other image classification implementations.
To better show the recommended method’s practicality, its effectiveness was compared to that of
other approaches already in use. Table 2 demonstrates that, in terms of performance, the proposed technique
outperformed other networks. Aiming at about 97.25%, the suggested strategy.
Table 2. Comparison with other approaches overall performance
Reference Year Accuracy
Saifan and Jubair [5] 2022 81.75%
Nawaz et al. [10] 2021 93.10%
Gouda et al. [13] 2022 83.2.%
Ameri [26] 2020 84.00%
Kim et al. [27] 2021 80.00%
Gouabou et al. [28] 2021 76.60%
Polat and Koc [29] 2020 92.90%
Chaturvedi et al. [30] 2020 91.11%
Proposed model 2023 97.25%
7. CONCLUSION AND FUTURE WORK
The classification issue gets increasingly difficult as the number of people with skin diseases rises
daily. particularly after gaining success in it. We suggest a system to help dermatologists and people diagnose
skin conditions. used this model to determine the kind of skin illness present in a particular image. Images of
skin lesions were classified using CNN techniques in the proposed work The Benign (ISIC) skin cancer
dataset and the melanoma, malignant, not melanoma (HAM) dataset were used in the tests. The images were
pre-processed, before the training and testing phase, after which they were split into feature and target values,
creating data augmentation. According to the results, the customized CNN had an accuracy rate of 97.25%.

5801
Using accuracy, precision, recall, and F1-Score, the customized CNN approaches were assessed after the
tests. This shows that the suggested CNN performs more effectively at classifying the data set than the
current CNN. The recommended approach has less loss and error and is more accurate than the one that has
been shown to be most useful in the literature. In comparison to other cutting-edge systems’ performance, it
is a competitive framework. Researchers can further develop CNN design and implementation by adjusting
hyperparameters like the number of layers, the kind of layers, and the hyperparameter values for the layers,
as well as by investigating other pre-trained CNN models. Additional activities might be added, other
aggregations of the activities could be encountered, and future studies will concentrate on merging more
sophisticated deep structures for precise cancer classification and speed.
APPENDIX
Table 1. The classification report for traditional CNN architectures
ResNet50 DenseNet
Dataset Precision Recall F1-score Precision Recall F1-score
Benign 0.68 0.79 0.73 0.89 0.49 0.63
Melanoma 0.44 0.59 0.51 0.40 0.95 0.56
malignant 0.48 0.26 0.34 0.36 0.09 0.14
Not Melanoma 0.60 0.52 0.56 0.51 0.29 0.37
Over all accuracy 0.55 Over all accuracy 0.48
Mobile Net VGG19
Benign 0.80 0.75 0.78 0.70 0.71 0.71
Melanoma 0.42 0.46 0.44 0.54 0.56 0.55
malignant 0.51 0.63 0.56 0.59 0.43 0.50
Not Melanoma 0.62 0.34 0.44 0.43 0.58 0.49
Xception EfficientNet
Benign 0.53 0.68 0.59 0.67 0.24 0.35
Melanoma 0.38 0.82 0.52 0.28 0.08 0.12
malignant 0.22 0.01 0.02 0.26 0.09 0.14
Not Melanoma 0.33 0.01 0.02 0.18 0.75 0.28
InceptionV3 Proposed Model
Benign 0.71 0.74 0.72 1.00 0.96 0.98
Melanoma 0.45 0.72 0.55 0.96 0.98 0.97
malignant 0.57 0.36 0.44 0.98 0.98 0.98
Not Melanoma 0.48 0.22 0.30 0.95 0.97 0.96
REFERENCES
[1] G. Wang, P. Yan, Q. Tang, L. Yang, and J. Chen, “Multiscale feature fusion for skin lesion classification,” BioMed Research
International, 2023, doi: 10.1155/2023/5146543.
[2] W. Abbes and D. Sellami, “Deep neural networks for melanoma detection from optical standard images using transfer learning,”
in Procedia Computer Science, 2021, vol. 192, pp. 1304–1312, doi: 10.1016/j.procs.2021.08.134.
[3] S. M. Jaisakthi, P. Mirunalini, C. Aravindan, and R. Appavu, “Classification of skin cancer from dermoscopic images using deep
neural network architectures,” Multimedia Tools and Applications, vol. 82, no. 10, pp. 15763–15778, 2023, doi: 10.1007/s11042-
022-13847-3.
[4] A. Hameed et al., “Skin lesion classification in dermoscopic images using stacked convolutional neural network,” Journal of
Ambient Intelligence and Humanized Computing, vol. 14, no. 4, pp. 3551–3565, Apr. 2023, doi: 10.1007/s12652-021-03485-2.
[5] R. Saifan and F. Jubair, “Six skin diseases classification using deep convolutional neural network,” International Journal of
Electrical and Computer Engineering, vol. 12, no. 3, pp. 3072–3082, 2022, doi: 10.11591/ijece.v12i3.pp3072-3082.
[6] U. Bhimavarapu and G. Battineni, “Skin lesion analysis for melanoma detection using the novel deep learning model fuzzy
GC-SCNN,” Healthcare (Switzerland), vol. 10, no. 5, 2022, doi: 10.3390/healthcare10050962.
[7] R. Kaur, H. Gholamhosseini, R. Sinha, and M. Lindén, “Melanoma classification using a novel deep convolutional neural network
with dermoscopic images,” Sensors, vol. 22, no. 3, 2022, doi: 10.3390/s22031134.
[8] W. Salma and A. S. Eltrass, “Automated deep learning approach for classification of malignant melanoma and benign skin
lesions,” Multimedia Tools and Applications, vol. 81, no. 22, pp. 32643–32660, 2022, doi: 10.1007/s11042-022-13081-x.
[9] T. Alkarakatly, S. Eidhah, M. Al-Sarawani, A. Al-Sobhi, and M. Bilal, “Skin lesions identification using deep convolutional
neural network,” in 2019 International Conference on Advances in the Emerging Computing Technologies (AECT), Feb. 2020,
pp. 1–5, doi: 10.1109/AECT47998.2020.9194205.
[10] M. Nawaz et al., “Skin cancer detection from dermoscopic images using deep learning and fuzzy k‐means clustering,”
Microscopy Research and Technique, vol. 85, no. 1, pp. 339–351, Jan. 2022, doi: 10.1002/jemt.23908.
[11] S. Khouloud, M. Ahlem, T. Fadel, and S. Amel, “W-net and inception residual network for skin lesion segmentation and
classification,” Applied Intelligence, vol. 52, no. 4, pp. 3976–3994, Mar. 2022, doi: 10.1007/s10489-021-02652-4.

 ISSN: 2088-8708
5802
[12] B. Shetty, R. Fernandes, A. P. Rodrigues, R. Chengoden, S. Bhattacharya, and K. Lakshmanna, “Skin lesion classification of
dermoscopic images using machine learning and convolutional neural network,” Scientific Reports, vol. 12, no. 1. Nature
Publishing Group UK London, 2022, doi: 10.1038/s41598-022-26516-0.
[13] W. Gouda, N. U. Sama, G. Al-Waakid, M. Humayun, and N. Z. Jhanjhi, “Detection of skin cancer based on skin lesion images
using deep learning,” Healthcare (Switzerland), vol. 10, no. 7, 2022, doi: 10.3390/healthcare10071183.
[14] A. K. Waweru, K. Ahmed, Y. Miao, and P. Kawan, “Deep learning in skin lesion analysis towards cancer detection,” in Proceedings
of the International Conference on Information Visualisation, 2020, pp. 740–745, doi: 10.1109/IV51561.2020.00130.
[15] S. Singha and P. Roy, “Skin cancer classification and comparison of pre-trained models performance using transfer learning,” Journal
of Information Systems Engineering and Business Intelligence, vol. 8, no. 2, pp. 218–225, 2022, doi: 10.20473/jisebi.8.2.218-225.
[16] M. A. Khan, M. Sharif, T. Akram, R. Damaševičius, and R. Maskeliūnas, “Skin lesion segmentation and multiclass classification
using deep learning features and improved moth flame optimization,” Diagnostics, vol. 11, no. 5, 2021, doi:
10.3390/diagnostics11050811.
[17] X. Lu and Y. A. F. A. Zadeh, “Deep learning-based classification for melanoma detection using XceptionNet,” Journal of
Healthcare Engineering, 2022, doi: 10.1155/2022/2196096.
[18] F. S. Hanoon and A. H. Hassin Alasadi, “A modified residual network for detection and classification of Alzheimer’s disease,”
International Journal of Electrical and Computer Engineering (IJECE), vol. 12, no. 4, pp. 4400–4407, 2022, doi:
10.11591/ijece.v12i4.pp4400-4407.
[19] M. K. Islam et al., “Melanoma skin lesions classification using deep convolutional neural network with transfer learning,” in 2021
1st International Conference on Artificial Intelligence and Data Analytics, CAIDA 2021, 2021, pp. 48–53,
doi: 10.1109/CAIDA51941.2021.9425117.
[20] A. C. Salian, S. Vaze, P. Singh, G. N. Shaikh, S. Chapaneri, and D. Jayaswal, “Skin lesion classification using deep learning
architectures,” in 2020 3rd International Conference on Communication System, Computing and IT Applications (CSCITA), Apr.
2020, pp. 168–173, doi: 10.1109/CSCITA47329.2020.9137810.
[21] F. Afza, M. Sharif, M. A. Khan, U. Tariq, H. S. Yong, and J. Cha, “Multiclass skin lesion classification using hybrid deep features
selection and extreme learning machine,” Sensors, vol. 22, no. 3, 2022, doi: 10.3390/s22030799.
[22] P. B. S. Varma, S. Paturu, S. Mishra, B. S. Rao, P. M. Kumar, and N. V. Krishna, “SLDCNet: Skin lesion detection and
classification using full resolution convolutional network‐based deep learning CNN with transfer learning,” Expert Systems, vol.
39, no. 9, Nov. 2022, doi: 10.1111/exsy.12944.
[23] P. Ghosh et al., “SkinNet-16: A deep learning approach to identify benign and malignant skin lesions,” Frontiers in Oncology,
vol. 12, 2022, doi: 10.3389/fonc.2022.931141.
[24] M. A. Scarlat, “Melanoma,” Kaggle.com, 2018. https://www.kaggle.com/datasets/drscarlat/melanoma (accessed Feb. 01, 2023).
[25] F. Ullah, “Skin lesion dermis dataset,” Kaggle.com, 2021. https://www.kaggle.com/datasets/farhatullah8398/skin-lesion-dermis-
dataset (accessed Feb. 01, 2023).
[26] A. Ameri, “A deep learning approach to skin cancer detection in dermoscopy images,” Journal of Biomedical Physics and
Engineering, vol. 10, no. 6, pp. 801–806, 2020, doi: 10.31661/jbpe.v0i0.2004-1107.
[27] C. Il Kim, S. M. Hwang, E. Bin Park, C. H. Won, and J. H. Lee, “Computer-aided diagnosis algorithm for classification of
malignant melanoma using deep neural networks,” Sensors, vol. 21, no. 16, 2021, doi: 10.3390/s21165551.
[28] A. C. Foahom Gouabou, J.-L. Damoiseaux, J. Monnier, R. Iguernaissi, A. Moudafi, and D. Merad, “Ensemble method of
convolutional neural networks with directed acyclic graph using dermoscopic images: melanoma detection application,” Sensors,
vol. 21, no. 12, Jun. 2021, doi: 10.3390/s21123999.
[29] K. Polat and K. Onur Koc, “Detection of skin diseases from dermoscopy image using the combination of convolutional neural
network and one-versus-all,” Journal of Artificial Intelligence and Systems, vol. 2, no. 1, pp. 80–97, 2020, doi:
10.33969/ais.2020.21006.
[30] S. S. Chaturvedi, J. V Tembhurne, and T. Diwan, “A multi-class skin cancer classification using deep convolutional neural
networks,” Multimedia Tools and Applications, vol. 79, no. 39–40, pp. 28477–28498, 2020, doi: 10.1007/s11042-020-09388-2.
BIOGRAPHIES OF AUTHORS
Shimaa fawzy is an Assistant lecturer at MISR Higher institute for Engineering
and technology, Mansoura, Egypt. She received the M. Sc. degree in communications
engineering from Mansoura University 2018. She is currently a Ph.D. student in Faculty of
Engineering, Mansoura University. Her research interests are in the area of image processing,
communication systems, and both machine and deep learning methodologies. She can be
contacted at email: shimaafawzy89@gmail.com.
Hossam El-Din Moustafa is a Professor at the Department of Electronics and
Communications Engineering, the founder and former executive manager of Biomedical
Engineering Program (BME) at the Faculty of Engineering, Mansoura University. He is an
IEEE senior member. Research interests include biomedical imaging, image processing
applications, and bioinformatics. He can be contacted at hossam_moustafa@mans.edu.eg.

5803
Ehab H. AbdelHay is an Associate professor at Faculty of Engineering,
Mansoura University, Egypt. He is a Programs Director of Faculty of Engineering, Mansoura
National University, Egypt. He received the B.Sc. degree in Comm. Engineering from
Mansoura University, Egypt in 2005. He received M. Sc. degree from the same university in
2010. He received Ph.D. degree from the same university in 2015. He worked as a
Demonstrator at Department of comm. and electronics-Faculty of Engineering, Mansoura
University, from 2006, Lecture assistant from 2011, Assistant Professor from 2015 to
May2022, and Associate Professor from May 2022 till now. His research interest the area of
5G and Beyond, WSNs, IOT, Cloud Computing, AI, and Cyber Security He can be contacted
at email: ehababdelhay@mans.edu.eg.
Mohamed Maher Ata is an Assistant professor at MISR Higher institute for
Engineering and technology, Mansoura, Egypt. He has received his Ph.D. from the faculty of
Engineering, Electrical communication and electronics department, Tanta University, Egypt
with the cooperation of Regina University in Canada. His research area of interest was
utilized in the field of signal processing, image processing, Multimedia, machine and deep
learning, video processing, and computer vision. He has published many indexed research
articles (SJR indexed-ISI indexed) in the state-of-the art of computer vision, biomedical
engineering, astrophysics, electrical communication, bioinformatics, encryption, cyphering and
intelligent transportation systems (ITS). He can be contacted at mmaher844@yahoo.com.

A deep convolutional structure-based approach for accurate recognition of skin lesions in dermoscopy images

More Related Content

Similar to A deep convolutional structure-based approach for accurate recognition of skin lesions in dermoscopy images (20)

More from IJECEIAES (20)

Recently uploaded (20)

A deep convolutional structure-based approach for accurate recognition of skin lesions in dermoscopy images