医用画像情報学会雑誌
Online ISSN : 1880-4977
Print ISSN : 0910-1543
ISSN-L : 0910-1543
最新号
選択された号の論文の2件中1~2を表示しています
依頼総説(特別講演 II)
原著論文
  • -パターン認識アプローチの適用限界-
    吉岡 拓弥, 内山 良一
    2020 年 37 巻 1 号 p. 5-10
    発行日: 2020/03/23
    公開日: 2020/03/30
    ジャーナル 認証あり

    A cancer treatment plan has been determined based on TNM classification by considering patient's age and medical history. However, the ability to predict recurrence risk would be contributed for the selection of an appropriate treatment and follow-up plan. The purpose of this study is to develop a method for the prediction of recurrence risk of patients with lung cancer by using pattern recognition. The public database NSCLC-Radiogenomics was used in this study. Sixty one patients (24 recurrences and 37 no recurrences) selected from the public database, and their pretreatment CT images were obtained. First, we selected one slice from the largest tumor area and segmented the tumor region manually. We subsequently determined 367 radiomic features. Seven radiomic features were selected by using least absolute shrinkage and selection operator (Lasso). Linear discriminant analysis (LDA) and support vector machine (SVM) with 7 radiomic features were employed for the estimation of recurrence risk. The experimental result showed that the area under the curve (AUC) values were 0.79 with LDA and 0.91 with SVM, respectively. Our scheme can predict the recurrence risk of lung cancers by using non-invasive image examination. However, we found that pattern recognition was not practical for the prediction problems containing censored time.

feedback
Top