Taddaaaaa!!!!! This poster is not actually the finalised one. In fact I have the latest version but I didn't take a photo of it. So this is the best what I can show at the moment. Basically my PhD research is about detecting prostate cancer using image processing techniques based on diffusion MRI. The best part of my research is, it doesn't cover the detection part only but it will cover the localization and staging identification. Up to know I got some ideas on how to do the first two parts (detection and localization) but still a bit blur with the staging identification. To be quite honest, I haven't read anything related to the last part of my research yet but I will do soon.
Some keywords related to my research are texture segmentation, features extraction, texture analysis, diffusion MRI, etc. These are very common keywords in image processing and those who were/are in this filed they know the keywords very well. Been reading almost 50 journal, I found so many techniques, methods and approaches created by different researchers makes me think what is actually the best method to be used or applied in my research. The fact is, there are many different types of images which contain many different textures. This situation makes one particular method is suitable for only one type of texture. It's quite difficult to create a generic method which is able to do segmentation for many different images.
When I look at my first objective is to develop a generic texture based methodology which is one of the main challenges in my research. It doesn't sound difficult but out of almost 50 journals that I've read so far, only two of them have developed a generic texture based segmentation methodology. According to my Supervisor, the method should be flexible in doing segmentation. Not just for MR images but for other images such as landscape, portrait, objects, etc. That is actually very difficult because I need to develop a segmentation methodology which is able to deal with colours and grey-level (black and white images especially in medical imaging). Since coloured images and black and white images are quite different especially their boundaries, the segmentation methodology should be able to deal with noise on the image.
Although they have several methods/techniques that I can apply in my research but I need to do a lot of testing combining two different techniques (a technique deals with textures with colours and a technique deals with textures with black and white images). The best method should be a combination of those techniques which are compatible enough to deal with colours and grey-level.
Ha..I'm off now and see you in my next post!