Clinical Assessment of Watermarked Medical Images

Abstract

Problem statement: Digital watermarking provides security to medical images. Watermarking in Region Of Interest (ROI) however distorts medical images but it is known that the resulting loss of fidelity is visually imperceptible. Approach: Clinical assessment will objectively evaluate the distortion on medical images to see whether or not medical diagnosis is altered. We used 75 medical images consisting of x-rays, ultrasound and CT scans. Digital watermarking was inserted in ROI and ROI/Region Of Non Interest (RONI) in all of them. Three assessors were randomly assigned 225 images, each receiving 75, a mixture of watermarked and non watermarked images. Results: Chi square test was used and p<0.05 was considered significant. There was no significant difference between original images and those watermarked in ROI or ROI/RONI. There was no comment on image quality in all the images assessed. Conclusion/Recommendations: Digital watermarking does not alter medical diagnosis when assessed by clinical radiologists. The quality of the watermarked images was also unchanged.