Dog's Brain for Human's Problem

       It was the dawn of my third year of working at Ramathibodi Hospital after I came back from the United States that an intriguing research opportunity arises as it was definitely rare that we would get to visualize rabies-induced changes in 3 Tesla MRI-observable characteristics of the dogs' brain. This opportunity was brought to our center's (Advanced Diagnostic Imaging Center) doorstep by various established individuals such as Assoc. Prof. Jiraporn Laothtamatas, Prof. Thiravat Hemachudha, Miss Boonlert Lumlertdacha, Dr. Supaporn Wacharapluesadee, and other members from AIMC, Chulalongkorn University, and Queen Saovabha Memorial Institute.

       It was apparent from the beginning that traditional scans of MRI would brought little evidences of the anomaly into the light and it was this awareness of this very fact that Assoc. Prof. Jiraporn decides to include various advanced imaging modalities, one of which being Diffusion Tensor Imaging. This is because many evidences have surfaced in regards to the diagnostic revelation brought on by the study of water diffusion in the brain.

       It does not take a trained eye to tell that each brain, be it human's or dogs, is different from one individual to another and it is due to this fact that another challenge arises - one that is statistical in nature. In order to perform group analysis of the acquired data, the anatomical areas of the brain from each dog must be aligned to the same anatomical space. This is a problem I have confronted during my time in the United States when the research team was trying to perform group analysis of the brains of individuals with Alzheimer's disease, most of which can be considered elderly. Due to the fact that their brain matters have shrunken to various degree as well as their ventricles enlarged. This posed an immense challenge in the alignment of their brains. Even though there were conventional methods available, doing so have led to much errors. We overcame this problem by dividing the alignment procedure into two steps, first being the alignment of just the gray matter followed by altering the other structures based on the aligned gray matter. This method has allowed the team to perform subsequent statistical analysis although errors in the position of various anatomical structures were still prevalent although can be considered an improvement compared to other conventional methods. This method was then brought to be performed on our cases of dog brains and it was a complete failure as the brains were just too different from one another.

       In an avid attempt to circumvent this problem, we then employed probabilistic fiber tracking of the whole brain, a task that took more than seven consecutive days of computation, in order to provide a basis for the alignment. It almost took us by surprise that doing so yield very accurate alignment of anatomical structures even that of the proximal portion of the spinal cord. 

       It is due to this very success that this technique was brought to be implemented on human brain especially in individuals with Alzheimer's disease. The accuracy and validity of this method is again confirmed. This method then form a very solid basis for the Thai Brain Mapping Project. 

       From this success, coupled with the wealth of data available at our center, an ongoing ambition to establish ourselves as a very capable Computational Radiology Laboratory is becoming more and more realistic each day.