Margin-Aware Intra-Class Novelty Identification for Medical Images

TL;DR

This paper introduces TEND, a hybrid model for intra-class novelty detection in medical images that leverages autoencoder features and a margin-based learning approach, outperforming existing methods without requiring out-of-distribution training data.

Contribution

The paper presents a novel hybrid model, TEND, combining autoencoder and classifier techniques with margin-based learning for intra-class novelty detection in medical images.

Findings

TEND outperforms state-of-the-art methods on natural and medical image datasets.

The margin-based objective improves separation between in-distribution and out-of-distribution features.

The approach does not require out-of-distribution training data.

Abstract

Traditional anomaly detection methods focus on detecting inter-class variations while medical image novelty identification is inherently an intra-class detection problem. For example, a machine learning model trained with normal chest X-ray and common lung abnormalities, is expected to discover and flag idiopathic pulmonary fibrosis which a rare lung disease and unseen by the model during training. The nuances from intra-class variations and lack of relevant training data in medical image analysis pose great challenges for existing anomaly detection methods. To tackle the challenges, we propose a hybrid model - Transformation-based Embedding learning for Novelty Detection (TEND) which without any out-of-distribution training data, performs novelty identification by combining both autoencoder-based and classifier-based method. With a pre-trained autoencoder as image feature extractor, TEND learns to discriminate the feature embeddings of in-distribution data from the transformed counterparts as fake out-of-distribution inputs. To enhance the separation, a distance objective is optimized to enforce a margin between the two classes. Extensive experimental results on both natural image datasets and medical image datasets are presented and our method out-performs state-of-the-art approaches.