19
Implementing DEWA Framework for Early Diagnosis of Melanoma Setiawan Hadi, Bernard Y Tumbelaka, Budi Irawan, Rudi Rosadi Universitas Padjadjaran

Implementing DEWA framework for early diagnosis of melanoma

Embed Size (px)

Citation preview

Page 1: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early

Diagnosis of MelanomaSetiawan Hadi, Bernard Y Tumbelaka, Budi Irawan, Rudi Rosadi

Universitas Padjadjaran

Page 2: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

AGENDA• Overview• DEWA Framework• Experiment• Result• Conclusion and future works

Page 3: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Overview• As one type of skin cancer, Melanoma can be considered as a deadliest

type of skin cancer. The number of people infected are increasing year by year and are geographically distributed world wide. • Computer vision researchers are interested in the automatic, vision-

based, early diagnostic system of skin cancer. By developing a such system which is easily used by non-paramedic educated person, the people can visually check their skin as initial diagnosis without supporting of formal trained or skillful people.• In this paper, a lightweight, lowcost application has been designed, based

on the previous developed, integrated computer vision and image processing algorithms, called DEWA.

Page 4: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Page 5: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Overview• As one type of skin cancer, Melanoma can be considered as a deadliest

type of skin cancer. The number of people infected are increasing year by year and are geographically distributed world wide. • Computer vision researchers are interested in the automatic, vision-

based, early diagnostic system of skin cancer. By developing a such system which is easily used by non-paramedic educated person, the people can visually check their skin as initial diagnosis without supporting of formal trained or skillful people.• In this paper, a lightweight, lowcost application has been designed, based

on the previous developed, integrated computer vision and image processing algorithms, called DEWA.

Page 6: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Page 7: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Overview• As one type of skin cancer, Melanoma can be considered as a deadliest

type of skin cancer. The number of people infected are increasing year by year and are geographically distributed world wide. • Computer vision researchers are interested in the automatic, vision-

based, early diagnostic system of skin cancer. By developing a such system which is easily used by non-paramedic educated person, the people can visually check their skin as initial diagnosis without supporting of formal trained or skillful people.• In this paper, a lightweight, lowcost application has been designed, based

on the previous developed, integrated computer vision and image processing algorithms, called DEWA.

Page 8: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Page 9: Implementing DEWA framework for early diagnosis of melanoma

METHODS

Page 10: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

DEWA• Image Segmentation• Color Transformation• Morphological Filter

• Image Filtering• K-Means/Improved K-Means• Connected Component Labeling

• Object Classification • Ellips detection

Page 11: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

ABCD Rule• Assymetric • Border• Color• Diameter

• TDS Index

Page 12: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Illustration

Page 13: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Simple AlgorithmLoad skin cancer imageApply DEWA algorithm to obtain noise-free imageImage analysis based on ABCD Rule• Determine the asymmetric value (A)• Calculate the border irregularity (B)• Get the color distribution (C) of skin image• Compute the diameter (D)Computer TDS Index

Page 14: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Data for Experiment

Page 15: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Page 16: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Result

Page 17: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

SISIK

Page 18: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Conclusion• Early diagnosis of melanoma is important step that can be useful to

decrease the danger of skin cancer. By using public device such as smart phone, the diagnosis mechanism can be done by common people. • DEWA as framework, utilized three multi aspects and can be considered

as light framework because it has been deployed on small portable device successfully. • The integrated methods that have been applied, even though it has been

tested, and give promising result (83%). Some problems still found in the methods, such as inaccuracy in detection especially in color aspect need to be corrected.

Page 19: Implementing DEWA framework for early diagnosis of melanoma

Implementing DEWA Framework for Early Diagnosis of Melanoma, presented at ICCSCI 2015

Thank you for your attention.