| The modeling of the response of biological systems under a variety of conditions plays a very important role in the field of biomedical engineering. Current models generally do not have the resolution and accuracy that is desired for effective research in this field. Therefore, there is a great need for anatomically-based, high resolution and accurate models of physiological systems. The emergence of the Visible Human Project database has provided the infrastructure necessary for such a model. However, in order to take advantage of this database, it is necessary to classify the pixels in the images in terms of the type of tissue they represent. An automated tissue classification technique using artificial fuzzy neural networks is proposed in this research. This technique may be used to develop a variety of anatomically-accurate models of human body such as electrical, electromagnetic, acoustic, thermal models based on the Visible Human dataset. As an example of the functionality of the technique, it was used to develop an electrical model of human head for a study on the effects of electrical stimulation on the vestibular system. The technique was able to correctly identify about 90% to 95% of the pixels. The technique also has a utility in characterizing the tissue transfer function which can be utilized to compensate the media artifact in ECG, EEG, MEG and EMG studies. |
