Pulmonary nodule detector uses Gaussian intensity to improve CT scans
GRIPNet: Gaussian Radial Intensity Prior Guided Architecture for Pulmonary Nodule Detection in CT
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
Detecting very small lung nodules in CT scans is important for early lung cancer diagnosis, but these tiny nodules are hard to find because they look similar to other objects. The authors studied many annotated scans and found that nodules have a predictable pattern: their image intensity peaks at the center and fades outward like a Gaussian curve. They designed a new detection system, GRIPNet, which mimics this pattern through specialized filters and attention mechanisms, enabling more accurate and faster detection of these small nodules.
What this means in practice
- •For medical image teams: Improve detection of very small lung nodules in CT scans using intensity-based filters inspired by physics of nodule appearance.
- •For radiology software developers: Incorporate architecture elements that model nodule intensity decay patterns to boost accuracy and speed of lung cancer screening tools.
Authors
Haojie Yang, Ran Su
Abstract
Lung cancer causes more deaths than any other malignancy, and low-dose CT screening is the main pathway to early diagnosis. That pathway hinges on the smallest lesions, yet nodules below six millimeters remain hard to detect, because most methods treat a nodule as a generic object and ignore the imaging physics behind its appearance. We show that this appearance is highly regular. Intensity peaks at the geometric center of a nodule and decays radially in a Gaussian pattern, and a fit to 18,218 annotated lesions from three public benchmarks yields a mean radial coefficient of determination above 0.86 in every dataset and size stratum. A square convolution samples both axes uniformly and is mismatched to this radial signal, most severely for small nodules. Guided by this evidence, we propose GRIPNet (Gaussian Radial Intensity Prior Network), a detector in which every module maps to a measurable property of the intensity distribution. Pinwheel convolutions decompose radial gradients, a dual-frequency module separates boundary detail from structural context, dilated masked attention matches the decay extent, and an adaptive loss reweights samples by conspicuity. GRIPNet raises mAP@0.5 to 95.3, 91.6 and 97.9 percent on KanserSet, LUNA16 and Lung-PET-CT-Dx while sharpening high-IoU localization at real-time speed.