Application of artificial neural network models in segmentation and classification of nodules in digital images of breast ultrasound

Author(s)

Karem Daiane Marcomini

• Document type: Master's Dissertation
• Press: São Carlos.
• Institution: Universidade de São Paulo (USP). Escola de Engenharia de São Carlos (EESC/SBD)
• Defense date: 2013-03-26

Examining board members

Homero Schiabel; Antonio Adilton Oliveira Carneiro; Maria Cristina Ferreira de Oliveira

• Advisor: Homero Schiabel

Abstract

Many procedures have been developed to assist in the early diagnosis of breast cancer. In this context, ultrasound has become an indispensable tool to distinguish benign and malignant lesions. Due to the subjectivity on interpreting images, CAD schemes have provided to the specialist a second opinion more accurate and reliable. Thus, this research presents a methodology for the automatic detection and characterization of breast sonographic findings. The tests were based the use of images obtained by simulators and, as considerable results, were applied to clinical examinations. The process was started employing of a preprocessing (wiener filter, equalization and median filter) to minimize noise. Then, five segmentation techniques were investigated to determine the most concise representation of the lesion contour, enabling to consider the neural network SOM the most relevant. After the delimitation of the object, the most expressive features were defined to the morphological description of the finding, generating the input data to the neural classifier MLP. The accuracy achieved during training with simulated images was 94.2%, producing an Az of 0.92. To evaluating the data generalization, the classification was performed with a group of unknown images to the system, both to simulators as to clinical trials, resulting in an accuracy of 90% and 81%, respectively. The proposed classifier proved to be an important tool for the diagnosis in ultrasonography breast. (AU)

• FAPESP's process: 10/14194-3 - Application of artificial neural networks models on the classification of nodules in breast ultrasound images
• Grantee: Karem Daiane Marcomini
• Support Opportunities: Scholarships in Brazil - Master