International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 07 | July 2024
p-ISSN: 2395-0072
www.irjet.net
Skin disease detection using deep learning and machine learning algorithms Vanshi Nrupesh Patel1, Stephy Patel2 1 Student, Dept. of Computer Engineering (Software Engineering), Lok Jagruti Kendra University, Ahmedabad,
Gujarat, India
2 Professor, Dept. of Computer Engineering, Lok Jagruti Kendra University, Ahmedabad, Gujarat, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Human skin is an unpredictable and almost
more suffering to the patient body due to not only long term consumption of medicines but also application of ointments for quite a long time. Therefore, it is necessary for the patient to at least diagnose the skin disease that patient is suffering from. So once if patient is aware about the skin disease that he/she is aching through then he/she can prioritise the duration in which he/she want to visit the dermatologists clinic to cure the skin disease. [3] Early detection is skin disease can help not only to reduce the cost of dermatologists clinic but also reduce the suffering duration of patient. As a result, if there is one system in which a user can upload his/her skin lesion image and can receive the name and certain information regarding the skin disease that patient might be suffering through which is detected through the skin lesion image uploaded by the user then this can accelerate the process for patient to have a blur knowledge regarding his/her affected skin condition. Moreover, it is well known fact that it takes ample amount of time investment in waiting area in dermatologists clinics. So, if such skin disease detection system is available in reception areas of dermatologists clinics then the receptionist could take the pictures of skin lesions of patients and form a report for the same and deliver it to the dermatologist to acknowledge him regarding number of patients with common skin disease issues and remaining with intense skin regarding issues. So that doctor could do time management in such manner that he/she could give appropriate time to each patient depending upon patient’s skin disease intensity. To prepare such system the classifier used in the proposed work is convolutional neural network as they work efficiently in the case of images. [5] Also to train the model on large data sets of dermatological images, deep learning algorithms such as convolutional neural network is most efficient.
intricate landscape due to its irregular lesion patterns, moles, varying tone, presence of thick hair, and other confusing features. Identifying the infected skin area and identifying the type of skin disease helps in early awareness. This study deals with the detection system that allows users to detect and identify skin disease by providing an image of the affected area as input. In this study, there is one website on which there is a section called upload image inside which user can upload the image of his/her skin lesion, that input image can be taken from the database or by a user. Now, that image goes in backend where image undergoes preprocessing such as resizing and finally the array format of that specific image enter into convolutional neural network, which provides an output between zero to eleven. Each digit is affiliated to specific skin disease for instance Acne_Level_0 is for output digit 0, Acne_Level_1 is for output digit 1. Acne_Level_2 is for output digit 2, Blister is for output digit 3 and so no. Key Words: Convolutional Neural Networks, Image preprocessing, max-pooling, fully-connected layer, data augmentation, deep learning.
1.INTRODUCTION Computer intervention has become inevitable in all fields in recent years. One sector that heavily relies on computers for diagnostic purposes is the medical field. Skin diseases account for the majority of all illnesses worldwide. Despite being widespread, diagnosing it is highly challenging and requires extensive domain knowledge. [3] Also dermatologists clinic fees are expensive therefore major patients decide to delay the treatment of any skin disease encountered as patient thinks that skin disease is avoidable. But it is necessary to grasp knowledge that if certain skin diseases treatment are not started on time then delay in diagnosing and treating of that specific skin disease might result in increase in its intensity and as a result it might turn chronic and later when patient decides to treat that specific skin disease then it takes years of ointment applications and years of medicinal consumptions to treat that skin disease as it might have turned chronic due to delay in visiting dermatologists. At last it gets more expensive to treat such skin disease which has turned chronic and requires higher intervention of dermatologist and long term of dermatologists bills hit and at the end it turns more and
© 2024, IRJET
|
Impact Factor value: 8.226
2. METHODOLOGY 2.1 Dataset In this research, we used a custom dataset for skin disease classification. The dataset is divided into training and validation sets. Training dataset is within train folder and the model performance is assessed using validation dataset which is inside val directory.
|
ISO 9001:2008 Certified Journal
|
Page 371