TOTOKU White Paper Grayscale Resolution Required Grayscale Accuracy Flat Display Systems for Medical Imaging

Table of Contents 1. Preface ---------------------------------------------------------------------------------------------------------------------------3 2. Requirement for Perceptually Linear Display Characteristics -----------------------------------------------------3 3. Composition of Imaging Display System--------------------------------------------------------------------------------5 4. Contrast Resolution of Display System----------------------------------------------------------------------------------5 5. Functions and Operations of LUT-----------------------------------------------------------------------------------------6...

An image monitor used for filmless diagnosis is required to have high gray scale accuracy in displaying monochrome images. Generally, the bit width of the look-up table (LUT) of a monitor is a specification item that represents the capability of the monitor in its catalog. This white paper examines various errors in display systems to find out how the operations of LUTs and differences of input-output bit widths are related to the accuracy in displaying images. It aims to, establish the superiority of Totoku display systems and platform-independent approach to multi-shade* display and consequently...

Figure 2. GSDF (Grayscale Standard Display Function) Figure 3. Detectable minimum luminance modulation Calculated with GSDF Figure 4 Gamma characteristics and displayed grayscale images Figure 4 shows the comparison of 1000 different grayscale display images due to the difference in monitor display characteristics. This figure between any two shades are perceived as the same degree of change shows that, on a monitor with GSDF characteristics, differences grayscale displayed. On a monitor with gamma 2.2, however, there is a tendency that the shades are emphasized in a low luminance range and compressed...

Composition of Imaging Display System Figures 6a and 6b show a composition example of a standard imaging display system. A look-up table (LUT) used to convert the display-specific shading characteristics to the GSDF is provided in the computer's graphics card on some systems (Figure 6a) and in the monitor on other systems (Figure 6b), the latter of which are increasing due to a widespread use of medical imaging monitors in recent years. Images transferred from modality system are normally 12 bits wide but are output to the monitor with a bit width downscaled by the computer's operating system,...

LUT is the memory inserted between the display and the imaging signals transmitted as p-values and included in the graphics processing circuit. The LUT has a function of specifying the display's digital driving level (DDL) in order to output the shading data of pixels recorded in the video memory as luminance values. shows an example of a monitor with a built-in LUT and a conceptual illustration of operations for 10-bit LUT with 8-bit input-output (8-bit LUT with a 10-bit width). Figure 7 Conceptual illustration of LUT operations on a display system (monitor) with a built-in 10-bit LUT with 8-bit...

Accuracy in Conformance to GSDF Curve This chapter examines some items concerning the accuracy in conversion of the monitor-specific display characteristics to the GSDF characteristics. The display luminance accuracy is dependent on the bit width of the LUT, the gamma characteristics, and the luminance range specific to the display system. This section assumes that the display-specific gamma is 2.2 and that the luminance range is 0.7-410 cd/m . Deviation from GSDF Curve The GSDF is an analog, continuous curve but, on an actual system driven digitally, the luminance responses of the display are...

Errors in Comparison with GSDF (Luminance Errors per Step of p-values) The figure below shows how accurately the luminance values per step of input signals (e.g., 256-point luminance values for an 8-bit width signal) are displayed in comparison with the luminance values per step defined in the GSDF. The error, expressed as the "number of JNDs," is used as an indicator of the system accuracy performance. "Number of JNDs" per Interval of p-values and Deviation (Grayscale Continuity) On an ideal system, an increment of the display luminance should be one JND per increment of one step of input signal....

On monitors with a display luminance range from 0.7cd/m2 to 410cd/m2 as shown in Figure 11, for example, the "number of JNDs" existing in this luminance range is 676.4-57.8=618.6. number of steps is 256. 618.6÷255=2.43. If the input signal is 8 bits wide, the The mean value for "numbers of JNDs" per p-value interval (Jmean) is Thus, a discontinuity of luminance may be discernable in part of images with gradual luminance changes. If the input signal is 10 bits wide, Jmean is 618.6÷1024=0.60, less than one JND. Perceptually, therefore, it can be represented as a continuous luminance change. The...

Contrast Response Error (Contrast per JND associated with GSDF) The evaluation of a contrast response error is a method of quantitative and detailed inspection on whether increments in luminances in relation to increments in p-values demonstrate a perceptually linear relationship. This method is defined as the Advanced Evaluation in AAPM-TG18 (American Association of Medical Physicists of Medicine – Task Group 18). As shown in Figure 12, the luminance change rate per p-value interval (dL/L: contrast) is normalized to a value per JND and contrasted with that of the GSDF. Figures 17D to 23D in...

Comparison of GSDF Accuracies on Specific Systems 8-bit Input-Output System (LUT 8, 10, and 12bits) On this system, 12-bit image data stored on a computer is compressed by an 8-bit viewer and a graphics card into image signals with 8-bit, 256-step shades that drive the monitor. The monitor has a built-in LUT 8 bits wide or better. output bit width of 8, 10, or 12 bits. In generally 8-bit input-output systems have a LUT with an In this paper, we compare the characteristics of these three systems. Systems with a built-in LUT in the graphics card are handled in the same way as above. (1) 8-8-8-bit...