Development of Computer-Aided Diagnosis System for Periapical Lesions


Rodrigo Esteban Salas

Oral Defence Date: 



HH 301


Professors Okada & Petkovic


This project designs, implements, and evaluates a software system which solves the problem of non-invasive diagnosis of large periapical lesions using cone-beam computed tomography (CBCT) scans. Current solutions for computer-aided diagnosis tools in medical fields requires expensive proprietary software and large computers to perform diagnosis on medical conditions using CBCT data, due to the large amount of data and complexity of the image segmentation and classification algorithms. The Periapical Lesions Computer-Aided Diagnosis (PL-CAD) system, proposes an alternative to the design of medical diagnosis applications which is cost effective, distributed and available to a large number of users with minimum software/hardware requirements. The architecture of the PL-CAD system effectively separates the graphical user interface from the diagnosis process to identify large periapical lesions. The result is a system that enables users to set-up the diagnosis parameters, and freeing them from the time consuming task of classifying periapical lesions as cysts or granulomas, while avoiding invasive procedures to achieve the same goal.


Cone beam computed tomography, periapical lesions, cyst, granuloma, computer aided diagnosis, random walks algorithm, adaptive boosting algorithm, linear discriminant analysis algorithm.


Rodrigo Esteban Salas