CT-3DFlow : Leveraging 3D Normalizing Flows for Unsupervised Detection of Pathological Pulmonary CT scans
This addresses the problem of automated pathology detection in medical imaging for clinicians, but it is incremental as it applies an existing method (normalizing flows) to a new domain (3D CT scans).
The paper tackled unsupervised detection of pathological pulmonary CT scans by training a 3D normalizing flow model on healthy data and using log-likelihood deviations for anomaly detection, achieving state-of-the-art performance on a chest CT test dataset.
Unsupervised pathology detection can be implemented by training a model on healthy data only and measuring the deviation from the training set upon inference, for example with CNN-based feature extraction and one-class classifiers, or reconstruction-score-based methods such as AEs, GANs and Diffusion models. Normalizing Flows (NF) have the ability to directly learn the probability distribution of training examples through an invertible architecture. We leverage this property in a novel 3D NF-based model named CT-3DFlow, specifically tailored for patient-level pulmonary pathology detection in chest CT data. Our model is trained unsupervised on healthy 3D pulmonary CT patches, and detects deviations from its log-likelihood distribution as anomalies. We aggregate patches-level likelihood values from a patient's CT scan to provide a patient-level 'normal'/'abnormal' prediction. Out-of-distribution detection performance is evaluated using expert annotations on a separate chest CT test dataset, outperforming other state-of-the-art methods.