Wal, B.W. van de (2007) Automatic Classification of Hand-Dermatitis. Master's Thesis / Essay, Computing Science.
|
Text
Scriptie_Barend_van_de_Wal.pdf - Published Version Download (5MB) | Preview |
Abstract
An important problem in the evaluation of new treatments for hand-eczema or hand-dermatitis lies in finding an objective measure that tells us how badly a hand is affected. When evaluating the degree of dermatitis, dermatologists are influenced by a number of subjective factors; e.g. patient populations differ, knowledge of treatments varies, the dermatologist might be having a particularly good or bad day. In an attempt to overcome these problems during clinical trials, usually several dermatologists grade the same sets of hands or images thereof and their diagnoses are averaged out to get more reliable results. This thesis is a pilot study for building an automated system for the gradation of hand-dermatitis using photo images, with a high degree of reproducibility of results. The automated system that was implemented had an accuracy of 66 %, and of the wrongly identified images 66.6 % was off by only one class. This is a substantial improvement on the expected accuracy for guessing (16.67% in a 6 class problem) or always picking the largest class (47.4 % of the examples belonged to the same class). Though the automated system developed in this pilot study cannot compete with dermatologists it is a step into the right direction.
| Item Type: | Thesis (Master's Thesis / Essay) |
|---|---|
| Degree programme: | Computing Science |
| Thesis type: | Master's Thesis / Essay |
| Language: | English |
| Date Deposited: | 15 Feb 2018 07:44 |
| Last Modified: | 15 Feb 2018 07:44 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/9394 |
Actions (login required)
 |
View Item |