Spatiotemporal-Hopfield neural cube for diagnosing recurrent nasal papilloma

Gadolinium-enhanced MRI is widely used in detection of recurrent nasal tumors. In this paper, a specifically designed two-layer Hopfield neural network called spatiotemporal-Hopfield-neural-cube (SHNC) is presented for detecting the recurrent nasal papilloma. With the extended 3D architecture, the network is capable of taking each pixel's contextual information into pixels' labeling procedure. As SHNC takes pixel's contextual information into its consideration, the effect of tiny details or noises is effectively removed. Furthermore, due to the incorporation of competitive learning rule to update the neuron states to avoid the trouble of having to satisfy strong constraints, the network is facilitated to converge fast. In addition, a more accurate signal-time curve called relative intensity change (RIC) for dynamic MR images is proposed as a representation of Gadolinium-enhanced MRI temporal information. The RIC curves of recurrent nasal papilloma are embedded into the SHNC. Our experimental results show that the SHNC can obtain more appropriate, more precise position of recurrent nasal papilloma than K-means, PCA and Eigenimage-filtering methods.