Spiking neural network predicts cancer nerve invasion with less energy
SCINTILLA-SNN: A Spiking Multi-Scale Selective Aggregation Network for Perineural Invasion Prediction
Computer Vision and Pattern Recognition
Summary
Predicting whether certain cancer has spread along nerves before surgery is helpful but hard because signs on MRI scans are very small and hard to spot. The authors created a special type of neural network that processes MRI scans more efficiently by focusing on important small areas using a brain-inspired spiking method. Their method not only predicts this nerve invasion better than usual methods but also uses much less computing energy. They tested it on 10 years of patient data and showed good accuracy and energy savings.
What this means in practice
- •For hospital imaging teams: Implement MRI-based tools that identify nerve invasion in bile duct cancer before surgery with improved accuracy and lower computational costs.
- •For medical device developers: Create energy-efficient imaging analysis hardware or software that uses spiking neural networks to detect subtle cancer features in volumetric medical scans.$Commercial implications: Enables development of specialized low-power diagnostic imaging devices or software for hospitals and clinics using this paper's spiking network approach.
Authors
Youngung Han, Yului Jeong, Kyeonghun Kim, Dohyun Kweon, Suah Park, Hyunsu Go, Sungha Park, Anna Jung, Jinyong Jun, Yunho Choe, Yunjin Seo, Ken Ying-Kai Liao, Hyuk-Jae Lee, Nam-Joon Kim
Abstract
Preoperative prediction of perineural invasion (PNI) in cholangiocarcinoma (CCA) is clinically valuable but remains challenging because PNI-related cues on magnetic resonance imaging (MRI) are subtle, sparse, and spatially localized around the tumor boundary. Standard 3D CNN and transformer architectures process volumetric data in a dense or spatially uniform manner, which can dilute subtle PNI-related evidence while requiring a large number of multiply-accumulate operations over 3D feature grids. To address these limitations, we propose SCINTILLA-SNN, a 3D spiking network composed of a four-stage hierarchical backbone and a Multi-Scale Spike Aggregation (MSSA) module for PNI prediction. The backbone extracts hierarchical volumetric representations through spiking convolutional stages and local spike window modulation stages. Given the resulting stage-wise representations, MSSA maps each spatial token to a learnable content value and modulates it with a spike-dynamics gate derived from firing rate and timestep-wise membrane-potential variability. The resulting score, referred to as the diagnostic token score, is used to selectively aggregate sparse PNI-related evidence. Experiments on a 10-year retrospective cohort of 182 CCA patients show that SCINTILLA-SNN achieves an AUROC of 0.748 under 5-fold cross-validation, while reducing the estimated inference energy by 23.18$\times$ compared with dense MAC-only computation of the same network.