How to do image segmentation using deep belief network

6 次查看（过去 30 天）

Aik Hong 2015-2-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/174931-how-to-do-image-segmentation-using-deep-belief-network

评论： Aik Hong 2015-2-4

Hi all, I’m currently doing a project on image segmentation. I have a few datasets of CT scan slices image of a few patients. One dataset consists of around 170 slices of CT scan image.

I’ve manually segmented the region of interest using Matlab imfreehand tool. Thus I have one set of the original image together with another set of segmented image (in binary form).

Original images: 512x512 uint16 (IM-0001-0001.dcm till IM-0001-0170.dcm)

Manually segmented images: 512x512 logical (AAAmanual1.mat till AAAmanual170.mat)

I’ve downloaded the deep belief network Matlab source codes in the deep learning toolbox. As I know, we have to setup and train the DBN, then we can use the trained DBN to automatically segment the image and get the output as the segmented region.

But I have no idea how should I proceed from here to setup and train the DBN, then use it to segment the region that I want, given the input image (or original image whichever more suitable)? What else parameters do I need to change/adjust? How should I write the matlab code for this?

Need your help to suggest / advice how can I proceed from here… your help is very much appreciated. Thank you.