LFMM‑SCC is a multimodal machine learning framework designed for robust skin cancer classification using 3D Total Body Photography (3D‑TBP). The system employs a Convolutional Neural Network (CNN) to analyze lesion images and a Random Forest Classifier (RFC) to process clinical and demographic features. These complementary predictions are fused through a Gradient Boosting Classifier, achieving superior performance compared to standalone models.
- Accuracy & F1‑Score: 98.11%
- Performance: Significantly outperforms individual CNN and RFC baselines
- Design Goal: Reliable early triage of skin cancer cases in telehealth and low‑resource settings
By integrating visual and non‑visual data streams, LFMM‑SCC demonstrates the power of late fusion in medical AI, enabling scalable and effective screening solutions for dermatology.