Image Quality Overview

Lesion contrast.

The photograph of the Professor Roentgen has been modified by the addition of an artificial mole (just below his eye). In the left image, the mole is only slightly darker than the surrounding tissues; in the middle image, the mole is darker, and in the right hand image, the lesion is very dark. The mole (lesion) contrast thus increases as one moves from the left image to the right image. The blemish could equally be lighter, as shown in the images below, where again, the lesion contrast increases from left to right.

Image blur

The series of images depicted above illustrate the concept of blur (aka sharpness). The left image is very sharp, and all details are readily visible. In the middle image, there is loss of fine detail, and the image would be described as being slightly blurred (or unsharp). In the right image, there is blurring along the horizontal direction, as might occur if there was significant motion during the exposure of the film used to acquire this image.


The left image is uniform, where every pixel takes on the same value (127), and is displayed using a window width of 256 and a window level of 127, so that the image is uniformly gray. In the middle image, the value of each pixel has been randomly taken from a Gaussian distribution, where the mean value is 127, and the standard deviation is 10. In this case, 67% of the pixels take on values between 117 and 137, 95% of the pixels have values between 107 and 147, and 99% of the pixels have values between 97 and 157. The fluctuations depicted in the middle image are an example of random noise. The image on the right shows the appearance of a "noisier" uniform image, where the mean value remains 127, but the noise value (standard deviation) of the Gaussian distribution of these pixel values has been increased to 20.

The effect of random noise on lesion detection is illustrated in the three images below, which are identical but have the mole lesion shown above added. In the right image, the mole is high contrast, and is readily observable. In the left image, the mole contrast has been reduced to a level that makes detection impossible. In the middle image, the mole contrast is intermediate between these two extremes, but detection is still possible.

Artifacts in Images

The images shown below illustrate artifacts that can occur including a structured pattern (left) and geometric distortion (right).