DATE: Wed, Nov 13, 2013
TIME: 11:45 am
PLACE: Council Room (SITE 5-084)
TITLE: The Use of Artificial Intelligence tools to guide the surgical treatment of Adolescent Idiopathic Scoliosis
PRESENTER: Philippe Phan, M.D. FRCS(C)
The Ottawa Hospital, Civic Campus, University of Ottawa
ABSTRACT:

AIS (Adolescent Idiopathic Scoliosis) is a three dimensional deformity of the spine. The complexity of this pathology is related to its variable geometry, its unknown etiology and progression. Variability in surgical treatment of AIS has been repeatedly documented despite accepted clinical classifications but clear guidelines are lacking. Several applications based on computer algorithms have been developed in order to assist assessment and treatment of AIS. While their clinical usefulness is sound, their clinical applicability has been challenged. A review of those applications and the lessons learned will be presented.
Simple check-lists have proven to improve of surgical treatment and patients outcome. The development of a simple decision tree has permitted the improvement of clinical classification of AIS by increasing accuracy and classification speed and will be presented. In order to guide treatment, a surgical strategy decision tree (SSDT) based on rules extracted from the literature was developed, and its ability to cover surgical treatment alternatives in an area of great variability will be presented.
A Kohonen-Self-organizing map was developed based on common AIS measurements. The classification extracted was compared to the clinical classification in its ability to predict surgical treatment and assess surgical strategy patterns. This classification and the SSDT have been integrated in a Matlab platform to guide surgeons in their surgical treatment. In orthopaedics like many fields in medicine, multiple centers are collaborating in order to build large database and do prospective studies. Yet limited effort is done to gather data efficiently, generate classifications from those data, and integrate advance algorithms to improve management and prognosis prediction. Future research efforts in this direction will be presented.

Bio: Philippe Phan is a PhD candidate at the University of Montreal under the co-supervision of doctor Hubert Labelle and Jacques de Guise. His research interests include the study of spine pathologies, particularly spinal deformity and the application of computer science to the orthopedics field. He received his HB.Sc. in software engineering from the University of Toronto and pursued his medical studies at the University of Ottawa, residency in orthopedics at the University of Montreal and fellowship in spine surgery at Harvard University.