Automated Small Kidney Cancer Detection in Non-Contrast Computed Tomography

TL;DR

This paper presents an automated pipeline for detecting small kidney cancers in non-contrast CT scans, demonstrating improved accuracy and potential for screening applications.

Contribution

The study introduces a novel automated detection pipeline using a 2D axial-sample model that outperforms previous methods in identifying small renal cancers in NCCT images.

Findings

2D axial-sample model achieved AUC of 0.804

Pipeline sensitivity of 61.9% and specificity of 92.7%

Outperforms previous automatic detection methods

Abstract

This study introduces an automated pipeline for renal cancer (RC) detection in non-contrast computed tomography (NCCT). In the development of our pipeline, we test three detections models: a shape model, a 2D-, and a 3D axial-sample model. Training (n=1348) and testing (n=64) data were gathered from open sources (KiTS23, Abdomen1k, CT-ORG) and Cambridge University Hospital (CUH). Results from cross-validation and testing revealed that the 2D axial sample model had the highest small ($\leq$40mm diameter) RC detection area under the curve (AUC) of 0.804. Our pipeline achieves 61.9\% sensitivity and 92.7\% specificity for small kidney cancers on unseen test data. Our results are much more accurate than previous attempts to automatically detect small renal cancers in NCCT, the most likely imaging modality for RC screening. This pipeline offers a promising advance that may enable screening for kidney cancers.