Huntington’s disease detection using the cascaded convolutional neural network with generative adversarial network based augmentation

Huntington’s disease is a hereditary neurodegenerative disorder requiring early diagnosis for effective clinical intervention. Traditional diagnostic methods are often invasive and inadequate for early-stage detection. This study presents a Chaotic Lyrebird Optimization–based Cascaded Convolutional Neural Network (ChLyO_CCNN) for accurate Huntington’s disease detection using MRI images. To address data scarcity and bias, Generative Adversarial Network (GAN)–based data augmentation is applied, followed by Gaussian filtering for noise reduction. The cascaded CNN architecture integrates P-Net, R-Net, and O-Net to progressively refine feature extraction and classification. Hyperparameters of the network are optimized using the ChLyO algorithm to enhance convergence and generalization. Experimental results demonstrate that the proposed method achieves high diagnostic performance with an accuracy of 99.76%, precision of 99.65%, and sensitivity of 99.81%, outperforming existing approaches. The proposed framework offers a reliable and efficient AI-assisted solution for Huntington’s disease detection in clinical neuroimaging applications.