AI & ML Practical Magic

We've achieved 100% accurate fraud detection for high-value biologicals using 3D micro-CT 'biometrics.'

April 14, 2026

Original Paper

CNN-Modal Learning for Cordyceps sinensis Authentication in High-value Agricultural Products

Yuqi Li, Shuhan Guan, Wenbin Zhou, Hongtao Bi, Yuancan Xiao, Ming Zhang, Caixia Li, Tiancai Li, Lixin Wei, Lei Jiang

SSRN · 6569381

The Takeaway

By combining 3D micro-CT scans with CNNs, researchers can distinguish wild from cultivated Cordyceps with perfect specificity in 3 seconds. It eliminates the need for rare expert verification in high-stakes agricultural markets.

From the abstract

Authentication of wild Cordyceps sinensis—a high-value agricultural and medicinal fungus frequently adulterated with cultivated substitutes—is currently lacking in terms of reliable, non-destructive, and scalable identification methods. To address this critical gap, we developed a novel non-destructive identification platform by fusing 3D microcomputed tomography (CT) and convolutional neural network (CNN) modal learning. Wild and cultivated samples were imaged via micro-CT, stereomicroscopy, an