Unsupervised Lesion Detection with Locally Gaussian Approximation

• Published in: Machine Learning in Medical Imaging pp. 355 - 363
• Main Authors: Chen, Xiaoran, Pawlowski, Nick, Glocker, Ben, Konukoglu, Ender
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing 10.10.2019
• Series: Lecture Notes in Computer Science
• ISBN: 9783030326913; 3030326918
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-32692-0_41

Generative models have recently been applied to unsupervised lesion detection, where a distribution of normal data, i.e. the normative distribution, is learned and lesions are detected as out-of-distribu-tion regions. However, directly calculating the probability for the lesion region using the normative distribution is intractable. In this work, we address this issue by approximating the normative distribution with local Gaussian approximation and evaluating the probability of test samples in an iterative manner. We show that the local Gaussian approximator can be applied to several auto-encoding models to perform image restoration and unsupervised lesion detection. The proposed method is evaluated on the BraTS Challenge dataset, where the proposed method shows improved detection and achieves state-of-the-art results.