Improving cancer subtype diagnosis and grading using clinical decision support system based on computer-aided tissue image analysis
Chaudry, Qaiser Mahmood
MetadataShow full item record
This research focuses towards the development of a clinical decision support system (CDSS) based on cellular and tissue image analysis and classification system that improves consistency and facilitates the clinical decision making process. In a typical cancer examination, pathologists make diagnosis by manually reading morphological features in patient biopsy images, in which cancer biomarkers are highlighted by using different staining techniques. This process is subjected to pathologist's training and experience, especially when the same cancer has several subtypes (i.e. benign tumor subtype vs. malignant subtype) and the same cancer tissue biopsy contains heterogeneous morphologies in different locations. The variability in pathologist's manual reading may result in varying cancer diagnosis and treatment. This Ph.D. research aims to reduce the subjectivity and variation existing in traditional histo-pathological reading of patient tissue biopsy slides through Computer-Aided Diagnosis (CAD). Using the CAD, quantitative molecular profiling of cancer biomarkers of stained biopsy images are obtained by extracting and analyzing texture and cellular structure features. In addition, cancer sub-type classification and a semi-automatic grade scoring (i.e. clinical decision making) for improved consistency over a large number of cancer subtype images can be performed. The CAD tools do have their own limitations and in certain cases the clinicians, however, prefer systems which are flexible and take into account their individuality when necessary by providing some control rather than fully automated system. Therefore, to be able to introduce CDSS in health care, we need to understand users' perspectives and preferences on the new information technology. This forms as the basis for this research where we target to present the quantitative information acquired through the image analysis, annotate the images and provide suitable visualization which can facilitate the process of decision making in a clinical setting.