Extraction of Important Features for Mandibular Reconstruction by Enumeration of Lasso Solutions

SCI '19 第 63 回 システム制御情報学会研究発表講演会 (2019年5月22日～24日, 大阪)As implicit medical knowledge and experience are used to determine surgical plans, clarification of decision making is important for systematization of surgical procedures. We propose extraction methods of low dimensional features which are important to determine the number of fibular segments in mandibular reconstruction by enumeration of Lasso solutions. Experiments showed that the extracted four-dimensional features can correctly estimate 91.25% surgeons’plans