Advanced search
Start date

Test oracle automation for medical images of three-dimensional models

Full text
Misael Costa Junior
Total Authors: 1
Document type: Master's Dissertation
Press: São Carlos.
Institution: Universidade de São Paulo (USP). Instituto de Ciências Matemáticas e de Computação (ICMC/SB)
Defense date:
Examining board members:
Márcio Eduardo Delamaro; Helton Hideraldo Biscaro; Simone do Rocio Senger de Souza; Auri Marcelo Rizzo Vincenzi
Advisor: Márcio Eduardo Delamaro

Automated testing activities significantly contribute to cost reduction and project productivity during the software development process. Automated testing reduces human effort and increases the quality of the final product. However, the high complexity of contemporary SUT outputs limits, in some cases, the application of automated test strategies. Systems with graphical/audio outputs, three-dimensional imaging, graphical user interfaces and some web applications are examples of systems with complex outputs. The lack of automated test oracles leads to the application of manual tests performed by the tester (human oracle) in an informal, ad-hoc and unproductive manner. One possible contribution to alleviate the testers efforts is to implement test oracles based on the extraction of characteristics from the SUT outputs. Results from previous work show that such an approach contributes to increase test productivity, mitigating and complementing manual efforts. This dissertation proposes and evaluates automated test oracles for systems whose outputs consist of three-dimensional medical images. To do so, the framework O-FIm/CO (Oracle for Images and Complex Outputs) that uses CBIR (Content-Based Image Retrieval) concepts is explored as a strategy to automate test oracles. In addition to adaptations and extensions of the framework, plug-ins were developed which represent extractors of three-dimensional medical image characteristics of blood vessels. To demonstrate the efficacy of the approach, two experimental studies were conducted to evaluate the efficacy and accuracy of feature-based test oracles in the evaluation of such images. Moreover, an experimental study was conducted to compare feature-based test oracles with human oracles. The results show the efficacy of the approach as a promising strategy to automate testing activities, contributing to the reduction of time and effort generated by manual approaches during the quality assessment of three-dimensional medical imaging systems. (AU)

FAPESP's process: 15/11844-0 - Test oracle automation for medical image processing of three-dimensional models
Grantee:Misael Costa Junior
Support Opportunities: Scholarships in Brazil - Master