Computer program based on GLDM can improve accuracy in diagnosis of prostate cancer

Scientists in France have developed a computer program that can analyse in detail different shades of grey in medical images. Writing in a recent issue of the International Journal of Signal and Imaging Systems Engineering, they explain how the program based on the Grey Level Difference Method can improve accuracy in the diagnosis of prostate cancer.

Salah Bourennane and colleagues at the Fresnel Institute in Marseille, France, explain how multi-band signal analysis has become a focus in remote sensing, colour analysis, and in industrial inspection. More recently, it has been used in medical imaging. The approach allows roughness, smoothness and other characteristics of a three-dimensional structure to be ascertained with greater precision than other methods. It works well for analysing urban landscapes and surface damage to engineering components, for instance.

The team has now extended the approach to the analysis of images obtained during medical imaging, which are commonly monochromatic. In the present case, the computer program was tested on some 624 images are from prostatic tissue needle biopsy where different forms of the disease were present. By using the extended GLDM method of multi-band analysis, the researchers were able to extract far more information from such a grey-scale image than is commonly possible simply by visual inspection or conventional image processing techniques.

The team analysed at an approximate average level of 50 shades of grey (either 32 or 64 grey levels). With the conventional GLDM method diagnostic accuracy was about 92%, with their extended approach, they boosted that to almost 96% on retrospective analysis of the 624-image database. This bodes well for the improvement of prostate cancer diagnosis using biopsy imaging.

"Extensive experiments have been carried out on many multi- spectral images for use in prostate cancer diagnosis and quantitative results showed the efficiency of this method compared to the [standard] Gray Level Difference Method," the team concludes. "The results indicate a significant improvement in terms of global accuracy rate."