Rancang Bangun Sistem Identifikasi Psoriasis Vulgaris Berbasis Fitur Warna dan Tekstur Menggunakan Metode GLCM dann Deep Learning

Muslimah, Rizka Umi (2021) Rancang Bangun Sistem Identifikasi Psoriasis Vulgaris Berbasis Fitur Warna dan Tekstur Menggunakan Metode GLCM dann Deep Learning. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

	Text 02311740000041-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (2MB) | Request a copy
	Text 02311740000041-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (2MB) | Request a copy

Abstract

Psoriasis adalah peradangan kronis pada kulit yang disebabkan oleh masalah sistem kekebalan tubuh yang semakin parah dapat mengakibatkan komplikasi kardiovaskular, hati, dan ginjal. Psoriasis vulgaris adalah psoriasis yang paling umum dialami hampir 90% pasien psoriasis. Penyakit psoriasis vulgaris dapat didiagnosis dengan gambaran klinis dengan melihat permukaan kulit luar. Tetapi ada penyakit kulit lain seperti seborrhea, lichen planus, tinea corporis, pityriasis dan eksema yang memiliki gambaran klinis serupa. Oleh karena itu, penulis merancang sistem algoritma untuk mengidentifikasi psoriasis yang lebih mudah, akurat, menggunakan ekstraksi fitur momen warna dan tekstur GLCM dengan metode deep neural network. Pada penelitian ini digunakan data gambar dari International Skin Imaging Collaboration (ISIC) dan Kaggle sebagai data pelatihan dan data pengujian. Digunakan 115 gambar data pengujian dan 200 gambar data pengujian. Model deep neural network dilatih menggunakan model loss binary-cross entropy dengan metode optimasi Adam. Pada penelitian ini didapatkan nilai akurasi diagnosis sebesar 81,75% dengan sensitivitas 75% dan spesifitas 87,08%. Sedangkan pada data pengujian diperoleh akurasi diagnosis sebesar 83% dengan sensitivitas 79,60% dan spesifitas 86%.
================================================================================================
Psoriasis is an immune chronic inflammatory disease that worsens and can lead to cardiovascular, hepatic, and renal complications. Psoriasis vulgaris is the most common psoriasis in nearly 90% of psoriasis patients. Psoriasis vulgaris can be diagnosed clinically by looking at the skin surface. But there are other skin diseases such as seborrheic dermatitis (seborrhea), Lichen planus, Tinea corporis, Pityriasis and Eczema which have a similar clinical picture. Therefore, the authors designed a yahoo system to identify psoriasis more easily, accurately, using color moment feature extraction and GLCM with deep neural network methods. In this study, image data from the International Skin Imaging Collabortion (ISIC) and Kaggle were used as training data and test data. Used 115 data testing images and 200 data testing images. The deep neural network model uses a binary-cross entropy loss model with Adam's optimization method. In this study, the diagnostic accuracy value was 81.75% with a sensitivity of 75% and a specificity of 87.08%. While the test data obtained a diagnosis accuracy of 83% with a sensitivity of 79.60% and a specificity of 86%.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Psoriasis vulgaris, momen warna, tekstur, GLCM, deep neural network
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. T Technology > T Technology (General)
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Rizka Umi Muslimah
Date Deposited:	02 Sep 2021 05:42
Last Modified:	02 Sep 2021 05:42
URI:	http://repository.its.ac.id/id/eprint/90202

Actions (login required)

View Item