International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 09 Issue: 07 | July 2022 www.irjet.net p-ISSN: 2395-0072

Skin Disease Detection using Convolutional Neural Network

Srujan S A1, Chirag M Shetty2 , Mohammed Adil3 ,Sarang P K4 , Roopitha C H5

1,2,3,4Students, Department of CS&E, MITE, Karnataka, India

5Assist. Professor, Department of CS&E, MITE, Karnataka, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Dermatology remains one of the foremost 2. LITERATURE REVIEW
branches of science that is uncertain and complicated
because of the sheer number of diseases that affect the skin [1] Classification of skin cancer using CNN analysis of
and the uncertainty surrounding their diagnosis. The Raman Spectra.
variation in these diseases can be seen because of many
environmental, geographical, and gene factors and also the The performance of convolutional neural networks is
human skin is considered one of the most uncertain and compared and also the projection on latent structures with
troublesome terrains particularly due to the presence of discriminant analysis for discriminating carcinoma using
hair, its deviations in tone and other similar mitigating the analysis of Raman spectra with a high
factors. Skin disease diagnosis at present includes a series of autofluorescence background stimulated by a 785 nm
pathological laboratory tests for the identification of the laser. They’ve registered the spectra of 617 cases of skin
correct disease and among them, cancers of the skin are neoplasms (615 patients, 70 melanomas, 122 basal cell
some of the worst. Skin cancers can prove to be fatal, carcinomas, 12 epithelial cell carcinomas and 413 benign
particularly if not treated at the initial stage. The tumors) in vivo with a conveyable Raman setup and
Convolutional Neural Network system proposed in this created classification models both for convolutional neural
paper aims at identifying seven skin cancers: Melanocytic networks and projection on latent structures approaches.
Nevi, Melanoma, Benign keratosis-like lesions, Basal cell To test the classification model’s stability, a 10-fold cross-
carcinoma, Actinic keratoses, Vascular lesions, and validation was performed for all created models. To avoid
Dermatofibroma. The dataset used is "Skin Cancer MNSIT: model overfitting, the info was divided into a training set
HAM10000" and was obtained from Kaggle. It has a (80% of spectral dataset) and a test set (20% of spectral
disproportionate number of images for each disease class, dataset). The results for various classification tasks
some have well over a thousand while others have a few demonstrate that the convolutional neural networks
hundreds. significantly (p<0.01) outperforms the projection on latent
structures.
Key Words: Dermatology, Skin Disease, Cancer,
Convolutional Neural Network, MNSIT: HAM10000 [2] Channel Attention based Convolutional Network
for skin disease classification.
1. INTRODUCTION
This research aims to develop a system for detecting skin
Skin Cancers have wreaked havoc since the early ages and diseases employing a Convolution Neural Network (CNN).
it is particularly because of the sheer number of cancers The proposed model named Eff2Net is constructed on
that are present that they pose such a high risk - It is EfficientNetV2 in conjunction with the Efficient Channel
difficult to diagnose them without a laboratory test. In our Attention (ECA) block. This research attempts to switch
attempt to bring about a change in this ecosystem, we the quality Squeeze and Excitation (SE) block within the
have proposed an automatic skin cancer classification EfficientNetV2 architecture with the ECA block. By doing
system that can help people in identifying the particular so, it had been observed that there was a big call for the
type of pigmented lesion that has taken over their skin. full number of trainable parameters. The proposed CNN
The idea behind this project is to make it possible for a learnt around 16 M parameters to classify the disease,
common man to get a sense of the disease affecting which is relatively less than the present deep learning
his/her skin so they can get a head start in preparing for approaches reported within the literature. This disease
its betterment and also the doctor in charge can get an classification was performed on four classes: acne,
idea about the type of cancer, which ultimately helps in keratosis (AK), melanoma, and psoriasis.
faster and efficient diagnosis. We make use of a
[3] The Automated skin lesion segmentation using
Convolutional Neural Network that uses Batch
attention based deep Convolutional Neural Network
Normalization to normalize the layer’s inputs and also
makes use of an Adam optimizer. The dataset used is open To advance the digital process of segmentation, a deep
source obtained from Kaggle and of all the lesions in the learning-based end-to-end framework is proposed for
dataset, more than 50% has been confirmed through automatic dermoscopic image segmentation. The
histopathology.
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 2162
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 07 | July 2022 www.irjet.net p-ISSN: 2395-0072

framework has the modified kind of U-Net, which higher accuracy even with the smaller training data size,
effectively uses Group Normalization (GN) within the they’ve used image data augmentation. The binary
encoder and also the decoder layers. Attention Gates (AG) classifier achieved an accuracy of 73% with 220 epochs of
that specializes in minute details within the skip coaching, and therefore the multiclass classifier's accuracy
connection, later incorporates Tversky Loss (TL) because is 72% with 200 epochs. Finally, the results are shown to
the output loss function is added. Rather than Batch the user, including the kind of disease, spread, and
Normalization (BN), GN is employed to extract the feature severity.
maps generated by the encoding path efficiently. To tell
apart high dimensional information from low-level 3. METHODOLOGY
irrelevant background regions within the input image, AGs
are used. Tversky Index (TI)-based TL is applied to We begin by resizing the images to 28,28 for better
accomplish better alliance between recall and precision. learning and then proceed to add the names and labels
To further strengthen feature propagation and encourage after which the plot parameters are set. The image pixels
feature reuse, atrous convolutions are applied within the are stored as a dependent variable while the target label is
connecting bridge between the encoder path and therefore stored as an independent feature. The data is divided into
the decoder path of the network. The proposed model is train and test split and it is reshaped to handle the
evaluated on the ISIC 2018 image dataset. imbalance issues (it can only be handled if the data is 2
Dimensional), after which Random Oversampler does the
[4] Deep learning approach to skin layer segmentation job of handling the imbalance. The data is fit on the train
in inflammatory dermatoses set and the new shape is checked, following which it is
reshaped again to 3 Dimension in order to train the
In practice, the analysis, including segmentation, is Convolutional Neural Network.
sometimes performed manually by the physician with all
drawbacks of such an approach, e.g., extensive time After checking if the shape is acceptable, the Convolutional
consumption and lack of repeatability. Recently, HFUS has Neural Network model is defined and subsequently the
become common in dermatological practice, yet it's barely first layer of the CNN is plotted. Max pooling is used to
supported by the use of automated analysis tools. To select the maximum features as identified by the
satisfy the necessity for skin layer segmentation and convolutional filter. After this, Batch Normalization does
measurement, they have developed an automatic the job of making the Artificial Neural Network faster and
segmentation method of both epidermis and SLEB layers. more stable by normalization of the layer’s inputs through
It consists of a fuzzy c-means clustering-based re-scaling and re-centering. The Convolutional Neural
preprocessing step followed by a U-shaped convolutional Network is Flattened to feed the fully connected Artificial
neural network. The network employs batch Neural Network and Dropout is used to avoid overfitting.
normalization layers adjusting and scaling the activation After all this, the first Artificial Neural layer is defined.
to make the segmentation more robust. The obtained Softmax is used as the activation function to the output
segmentation results are verified and compared to state- layer which has 7 neurons. For optimization, the learning
of-the-art methods addressing the skin layer rate is set to 0.001 and Adam optimizer is used.
segmentation. The obtained Dice coefficient adequate 0.87
and 0.83 for the epidermis and SLEB, respectively, proves The model is compiled with accuracy as metric and loss as
the developed framework’s efficiency, outperforming the Sparse categorical since we have multiple outputs and
opposite approaches. then after that the data is trained using sample validation
split as 0.2. The model is predicted on the test set and the
[5] Skin cancer detection using Convolutional and predicted probability is converted to classes. The model is
Artificial Neural Networks then finally evaluated. The parameters amount to about
half a million of which about a thousand are non-trainable
This paper focuses on the event of classifiers capable of params.
detecting skin cancer(s) given dermoscopic images. The
dataset used for the training is a part of the 2019 ISIC
Challenge, and consists of over 25,000 labeled
dermoscopic images. Specifically, classifying dermoscopic
images accounts for nine different diagnostic categories:
melanoma, melanocytic nevus, basal cell carcinoma,
keratosis, benign keratosis, dermatofibroma, vascular
lesion, and epithelial cell carcinoma, a number of which
are benign. They've developed classifiers -a binary
classifier and a multiclass classifier -on the Google Cloud
Platform using Convolutional Neural Networks (CNNs). To
forestall the classifiers from overfitting and to attain
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 2163
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 07 | July 2022 www.irjet.net p-ISSN: 2395-0072

Chart -2: Model Accuracy Chart

The 2 graphs depict the decreasing model loss and the

increasing accuracy of the system respectively.

Fig -1: Data Flow diagram 5. CONCLUSION

The above figure represents how the data transitions into We observe that the CNN model is accurate to an extent
various stages as required for the prediction. and going forward, with careful scrutiny and a more
reliable dataset, the model can be tweaked to attain a
4. RESULTS greater degree of accuracy. By storing the results in an H5
file, one can build an application around it that could give
The system was seen to predict the said diseases with an quick predictions on the go to users who upload an image
accuracy of around 74% - 75%. The model loss can be of the diseased part of their skin.
seen decreasing abruptly at first and then gradually
towards the end. Accordingly, the accuracy rises abruptly REFERENCES
and consolidates towards the end. The epoch was set to 50
after carefully testing for different epoch values. [1] Ivan Bratchenko, Lyudmela Bratchenko, Yulia
Khristoforova. Classification of skin cancer using CNN
analysis of Raman Spectra. ScienceDirect, November
2021.

[2] Karthik R, Tejas Vaichole and Sanika Kulkarni. Channel

Attention based Convolutional Network for skin disease
classification. ScienceDirect, August 2021.

[3] Ridhi Arora , Balasubramanian Raman and Ruchi

Awasthi. The Automated skin lesion segmentation using
attention based deep Convolutional Neural Network. May
2020.

[4] Pawel Budura, Anna Platkowska and Joanna

Czajowska. Deep learning approach to skin layer
segmentation in inflammatory dermatoses. IEEE. July
2020
Chart -1: Model Loss Chart
[5] Joshua John, Mallia Galatti and Gillian Lee. Skin cancer
detection using Convolutional and Artificial Neural
Network. Journal of Computing Sciences. January 2020

[6] Mohammed Al-Masni, Don-Hyung Kim and Tae Seong

Kim. Multiple skin lesion diagnosis via integrated deep

© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 2164
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 07 | July 2022 www.irjet.net p-ISSN: 2395-0072

CNN for segmentation and classification. ResearchGate,

March, 2020.

[7] Ling Fang Lee, Xu Wang, Neal N. Xiaong and others.

Deep Learning in Skin Disease Image Recognition. IEEE,
November 2020.

[8] Vipul Dhabi, Vipul Goswami, Harshad Kumar. Skin

Disease Classification from Image. IEEE, March 2020.

[9] Md Al Mamun, Mohammed Sharif. A Comparative

Study Among Segmentation Techniques for Skin Disease
Detection Systems. ResearchGate, January 2021.

[10] Akhtar Jamil, Merve Gun, Alaa Ali Hamid. Skin Lesions
Segmentation and Classification for Medical Diagnosis.
Researchgate April 2021.

[11] Yunendah Nur Fu'adah1, NK Caecar Pratiwi1,

Muhammad Adnan Pramudito1 and Nur Ibrahim.
Automatic Skin Cancer Classification System. IOP Science.

[12] Kamil Dililler, Boran Sekeroglu. Skin Lesion

Classification Using CNN-based Transfer Learning Model.
Journal of Science, January 2022.

[13] Akash kumar, Jalluri Rama, James Jing Khan.

Classification of Skin Disease Using Deep Learning Neural
Networks with MobileNet V2 and LSTM. mdpi, September
2021.

[14] Jessica Valesco, Cherry Pascheon, Jonathan Apuang. A

Smartphone-Based Skin Disease Classification Using
MobileNet CNN. ResearchGate, October 2019.

[15] C Pabitha, B Vinitha. Deep learning based severity

grading for skin related issues, AIP Conference
Proceedings, May 2022.

© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 2165