Medical diagnostic image data fusion based on wavelet transformation and self-organising features mapping neural networks

In recent years, the collection of various data coming from anatomical and functional imagery is becoming very common for the study of a given pathology, and their aggregation generally allows for a better medical decision in clinical studies. However, it is difficult to simulate the human ability of image fusion when algorithms of image processing are piled up merely. On the basis of the review of researches on psychophysics and physiology of human vision, this paper presents an effective multi-resolution image data fusion methodology, which is based on discrete wavelet transform theory and self-organizing features mapping neural network (SOFMNN), to simulate the processes of images recognition and understanding implemented in the human vision system. Through the two-dimensional wavelet transform, original images can be decomposed into different types of details and levels. The integration rule can be built using self-organizing neural networks, just like the automatic work in human brain. As an example, the fusion process is applied in the clinical case: the study of some particular disease by MR/SPECT fusion. Results are presented and evaluated, and a preliminary clinical validation is achieved. The assessment of the method is encouraging, allowing its application on several clinical diagnostic problems.