Correction for attenuation and visual response in digital radiography.

OBJECTIVES: To derive a method to correct for the exponential attenuation adding visual linearization for digital radiography and to perform an experimental test to study effects on observer performance. METHOD: A theoretical analysis was performed and expressions were derived to correct radiographic data for the attenuation. An experiment was performed exposing an aluminum step wedge with holes of increasing depths behind each step plus soft tissue simulation. Seven original images were created with a digital intraoral X-ray sensor and a further 14 transformed images produced. Ten observers examined the images in a randomized order reporting the number of holes seen per wedge step. The results were analysed by One Way ANOVA. RESULTS: It was possible to correct for attenuation and the response of the human visual system to light intensities from a computer monitor. Perception was significantly improved in all recalculated radiographs that took both attenuation and the response of the eye into account (P-values ranged from <0.0001 to <0.0286). CONCLUSIONS: The transforms may be useful for improved perception when viewing digital radiographs.

Visual linearization of the display of digital radiographs.

OBJECTIVES: To derive and test a method to linearize the visual response of the display of digital radiographs so that equal steps in gray levels will be perceived as equal steps in brightness. METHOD: A mathematical analysis was performed and expressions for visual linearization were derived. Twenty-four test images were computer generated to confirm that visual linearization may be achieved. Each image had three groups of square areas of different size placed in three rows. The left and right squares in each row were given different gray levels to simulate various contrast levels. The middle squares were initially given the same gray level value as one of the outer squares. The images were examined by ten observers who could change the gray levels of the middle squares so that the step in brightness between the middle square and the outer squares become subjectively equal. The test was performed three times employing two different monitors. RESULTS: The experimental test confirmed that visual linearization could be achieved. Linear regression analyses gave determination coefficients of 0.9926 amd 0.9942 for monitors with gamma-values of 1.93 and 2.50. respectively. The mean data from the ten observers perfectly fitted those theoretically calculated. CONCLUSION: Visual linearization of gray levels can be achieved but further clinical research is needed to determine if this improves diagnosis.