Detecting Outliers with Poisson Image Interpolation

Supervised learning of every possible pathology is unrealistic for many primary care applications like health screening. Image anomaly detection methods that learn normal appearance from only healthy data have shown promising results recently. We propose an alternative to image reconstruction-based and embedding-based a new self-supervised method tackle pathological detection. Our approach originates in the foreign patch interpolation (FPI) strategy has superior performance on brain MRI abdominal CT data. use better strategy, Poisson (PII), which makes our suitable challenging regimes. PII outperforms state-of-the-art by good margin when tested surrogate tasks identifying common lung anomalies chest X-rays or hypo-plastic left heart syndrome prenatal, fetal cardiac ultrasound images. Code available at https://github.com/jemtan/PII.