Autor

Noél Pérez, Miguel A. Guevara, Mario Vaz, Raúl Ramos, Manuel Rubio, Francisco Castrillo 

Medio de Publicación

Congreso: Instituto Politécnico de Setúbal
Año: 2009
Tipo de publicación: Oral

Absctract

Mammography is the best method used for screening; it is a test producing no inconvenience and with small diagnostic doubts of breast cancer since the preclinical phase. For this reason, reliable Computer-Aided-Diagnosis (CAD) systems for automated detection/classification of Phatological Lesions (PL) will be very useful and helpful, providing a valuable "second opinion" to medical personnel. Several detection and diagnosis methods have been developed and represent important approaches to improve (in major o minor degree) the mammography image analysis process. However, these techniques only classify (by separation) some specific class of masses or calcifications without tackling the whole classification problem.

Distributed computing systems (GRID) represent a valuable model to be used in this context. In particular, the GRID framework offers huge storage capacity (high-resolution images and its replicas) and high performance computing power (to execute methods and algorithms) to manage shared resources deployed across collaborating institutions. This paper describes a novel CAD tool that combines digital image processing and artificial neural networks among others techniques to diagnose mammography PL (as bening or malignant tissues) on GRID environments. This CAD tool was tested statistical and experimentally in a representative repository (dataset) from the Mammography Image Analysis Society database.