Expert System with an Embedded Imaging Module for Diagnosing Lung Diseases

Lung diseases are one of the major causes of suffering and death in the world. Improved
survival rate could be obtained if the diseases can be detected at its early stage. Specialist
doctors with the expertise and experience to interpret medical images and diagnose
complex lung diseases are scarce. In this work, a rule-based expert system with an
embedded imaging module is developed to assist the general physicians in hospitals and
clinics to diagnose lung diseases whenever the services of specialist doctors are not
available. The rule-based expert system contains a large knowledge base of data from
various categories such as patient's personal and medical history, clinical symptoms,
clinical test results and radiological information. An imaging module is integrated into
the expert system for the enhancement of chest X-Ray images. The goal of this module is
to enhance the chest X-Ray images so that it can provide details similar to more
expensive methods such as MRl and CT scan. A new algorithm which is a modified
morphological grayscale top hat transform is introduced to increase the visibility of lung
nodules in chest X-Rays. Fuzzy inference technique is used to predict the probability of
malignancy of the nodules. The output generated by the expert system was compared
with the diagnosis made by the specialist doctors. The system is able to produce results
which are similar to the diagnosis made by the doctors and is acceptable by clinical
standards.