Medical Image Processing

Block diagram for our super-resolution image reconstruction is shown in Fig.1.

1. Total Vriation regularization decomposes an input image into structure and texture components.
2. Linear interpolation and edge enhancement for structure components and only linear interpolation for texture components are processed.
3. The structure and texture components are combined to obtain an output image.

The experimental results of super-resolution image reconstruction on a hand X-ray image is shown below.

(a) Original image	(b) Super-resolution image
Fig. 2: Experimental results for super-resolution image reconstruction.

Rheumatoid arthritis is a disease in which the distance between the joints is gradually shortened due to inflammation. As time passes after the onset of the disease, bones become deformed and it becomes difficult to carry out daily activities, so it is important to detect the disease in early stage and prescribe medication as soon as possible.
However, it is difficult for physicians to distinguish early symptoms of rheumatoid arthritis from other diseases visually. Therefore, we aim to develop an application that automatically measures and quantitatively evaluates Joint Space Distances from X-ray images.

1. Super-resolution image reconstruction is performed on an input image to obtain a high-resolution image.
2. An upper edge of a joint is selected by using a mouse.
3. A curve is fitted to the selected points and a curve is drawn.
4. Do the same process with a lower edge and a curve is drawn.
5. Measure the distance between the two curves to the joint.

(a) Original image	(b) Curve fitting image
Fig. 3: Application for measurement of JSD