计算机科学
人工智能
学习迁移
分类器(UML)
深度学习
模式识别(心理学)
脑瘤
机器学习
胶质瘤
召回
计算机辅助设计
病理
医学
哲学
工程类
工程制图
癌症研究
语言学
标识
DOI:10.1016/j.compbiomed.2019.103345
摘要
Brain tumor classification is an important problem in computer-aided diagnosis (CAD) for medical applications. This paper focuses on a 3-class classification problem to differentiate among glioma, meningioma and pituitary tumors, which form three prominent types of brain tumor. The proposed classification system adopts the concept of deep transfer learning and uses a pre-trained GoogLeNet to extract features from brain MRI images. Proven classifier models are integrated to classify the extracted features. The experiment follows a patient-level five-fold cross-validation process, on MRI dataset from figshare. The proposed system records a mean classification accuracy of 98%, outperforming all state-of-the-art methods. Other performance measures used in the study are the area under the curve (AUC), precision, recall, F-score and specificity. In addition, the paper addresses a practical aspect by evaluating the system with fewer training samples. The observations of the study imply that transfer learning is a useful technique when the availability of medical images is limited. The paper provides an analytical discussion on misclassifications also.
科研通智能强力驱动
Strongly Powered by AbleSci AI