NeuroMorphFusion: A Neuro-Inspired Hybrid Learning Framework for Interpretable Deep Lesion Detection in IoT-Enabled
Roseline Oluwaseun Ogundokun1,2, Rotimi-Williams Bello1, Pius Adewale Owolawi1
1Department of Computer Systems Engineering, Faculty of Information and Communication Technology, Tshwane University of Technology, Pretoria, South Africa.
Technology in Cancer Research & Treatment
|March 12, 2026
Summary
NeuroMorphFusion enhances medical imaging by combining deep learning with neuro-inspired models for efficient and interpretable lesion detection on edge devices. This framework achieves high accuracy while maintaining low latency for real-world clinical integration.
Area of Science:
- Artificial Intelligence
- Medical Imaging
- Computational Neuroscience
Background:
- Deep learning in the Internet of Medical Things (IoMT) improves automated lesion detection.
- Challenges include maintaining diagnostic accuracy, interpretability, and efficiency on resource-limited edge devices.
Purpose of the Study:
- To introduce NeuroMorphFusion, a hybrid framework for interpretable and efficient lesion detection.
- To address the limitations of current deep learning models in IoMT for edge computing.
Main Methods:
- NeuroMorphFusion integrates a ResNet18 backbone, a Spiking Neural Network (SNN), and a morphological attention mechanism.
- A semi-supervised reinforcement learning strategy and a genetic algorithm (GA) optimize the model for accuracy, explainability, and efficiency.
- Hyperparameter optimization was performed on a Jetson Nano, converging within eight minutes.
Main Results:
- NeuroMorphFusion achieved 98.18% classification accuracy on the IQ-OTHNCCD lung cancer CT dataset.
- The framework demonstrated superior transparency and efficiency compared to VGG16, SqueezeNet, MobileNetV3, and ResNet18.
- Multi-objective optimization balanced sensitivity, latency, and explainability, enhanced by SHAP and Grad-CAM visualizations.
Conclusions:
- NeuroMorphFusion effectively combines neuro-biological inspiration, mathematical interpretability, and edge-efficient computation for IoMT.
- The framework shows potential for real-world clinical integration and scalable IoMT deployment due to its accuracy, explainability, and low-latency optimization.
