Portal de Periódicos da CAPES

Keeping up appearances — ultrasound imaging in a digital age

2006; Wiley; Volume: 27; Issue: 3 Linguagem: Inglês

10.1002/uog.2739

1469-0705

Alan D. Kaye, Sarah Newton,

Pancreatic and Hepatic Oncology Research

Further adding to this trend has been the dawn of the age of electronic imaging and high-speed broadband Internet. Due to the low cost and ease of their transmission, digital photos and movies have almost totally replaced ‘hard’ media for the purposes of sonographic observation and diagnosis, with real-time video and data communication even allowing for the advent of ‘tele-echographic’ remote ultrasound scans as reported previously in this Journal1. Given the pace of this progress, it is no surprise that ultrasound today is at the height of its use as a cost-effective, non-invasive diagnostic tool. Yet while this breadth of use and ease of image transmission spell good times ahead for ultrasound in the clinical environment, they present certain pertinent challenges for those whose role it is to publish the latest developments in the field. Ultrasound examination is unique in the type of image presented to the observer, inasmuch as it is contrast-based in nature and dynamic in form. While ideally suited to their purpose when viewed in situ, in real time and high definition on a suitable display screen, ultrasound images risk losing their impact when reproduced, frozen, on the pages of a journal. Over the last few years, it has been apparent that, despite authors' best intentions, a number of the images being submitted for publication in the Journal are falling far short of their capacity to act as exemplary snapshots of novel and interesting clinical situations. In this Editorial, we hope to present some of the dangers to look out for when preparing digital images for submission and, in so doing, better equip our readers with the skills needed to ensure that their material is reproduced to the best of its potential. There are several factors that dictate how an image will appear when printed, each one of which must be optimized if the result is to give an accurate and true representation of the original source material. These factors will be dealt with in turn, with examples of images given to illustrate each point. Following this, we will lay out some specific recommendations to bear in mind when capturing and processing images from ultrasound scanners to ensure that the maximum possible image fidelity is achieved on publication. ‘Halftone’ images (i.e. photographs and ultrasound scans, in monochrome or color) and, sometimes, line drawings, are stored as grids of dots, whose size per unit area determine the resolution of the image. The higher the resolution of the image and the larger the size of the picture, the greater the chances are that the image can be reproduced sharply. The Journal's guidelines state a preference of 300 dots per inch (dpi) for halftone images, with greater resolutions (500 dpi) for combinations of line drawings and halftones, and greater still (preferably 800 dpi, but a minimum of 600 dpi) for line drawings alone. What must be taken into account, however, is that these recommended resolutions apply to the image at the size at which it is finally reproduced; if an image has to be scaled up for reproduction, the resolution effectively decreases, often detrimental to the final result. Differences in image resolution can be seen in Figure 1. The resolution of image (a) (75dpi; common ‘screen’ resolution) is a quarter that of image (b) (300dpi), showing a noticeable loss of clarity, definition and detail. Reproduced from Mari et al.2. Several file formats exist for the transmission of digital images. However, for the purposes of this Editorial, only two will be discussed in any detail. The first is TIFF (Tagged Image File Format, or ‘.tif’). This is recommended for the submission of all images as it is a ‘lossless’ format; the algorithm (LZW) usually used to compress the raw picture data does not discard any of the information contained within the image. Given this, it provides an established, safe and flexible method for image capture and manipulation and caters for all of the production requirements of the Journal. The second is JPEG (Joint Photographic Experts Group, or ‘.jpg’). This too employs compression but, unlike TIFF, it is a ‘lossy’ format, greater compression rates and smaller file sizes being achieved by discarding image information. The greater the extent to which the image is compressed, the more obvious will be the picture degradation. While often one is able to select the degree of compression of a JPEG at the point at which the image is saved, this is not always the case. In addition, care must be taken when repeatedly saving an image in JPEG format, as each save can progressively worsen the quality of the image (Figure 2). There are cases (such as naturalistic images and images without hard edges and/or text) in which JPEG may be acceptable; however, if possible, the minimum amount of compression should be applied to achieve the maximum picture quality and reduce the risk of undesirable effects. After multiple saves as a JPEG (b), the original TIFF image (a) has degraded to the point at which much of its detail has been lost. Reproduced from Simchen et al.3. Other formats, such as GIF (Graphics Interchange Format), PNG (Portable Network Graphic) and EPS (Encapsulated PostScript) are less suitable for the submission of images and should be avoided. All color images in the Journal are printed using four inks: cyan, magenta, yellow, and black—the so-called four-color process (usually referred to as CMYK). By applying varying amounts of these four inks to the paper, a wide but finite palette of other colors can be created; this range of colors is called the gamut. In contrast, most capture devices used today (e.g. digital cameras/scanners) operate in a different ‘color space’, one using red, green, and blue (RGB). The RGB gamut (range of colors) is not the same as that of the CMYK process; some of the colors captured in the RGB color space cannot be reproduced in the CMYK color space and vice versa (Figure 3). Diagram indicating how the RGB and CMYK gamuts compare. Therefore, at some point, RGB-captured images will be converted into the CMYK color space. The conversion is performed on the basis that all points in the RGB gamut are mapped to a point in the CMYK gamut. Inevitably, some alteration in certain colors occurs. In particular, the more intense reds, greens, and blues appear somewhat duller (as can be seen when comparing the online and print versions of this article). Authors should bear this in mind when submitting images containing these colors, or colors that fall outside both gamuts (as may be produced by ‘computer-enhanced’ or ‘artificial-color’ techniques such as color Doppler). A pitfall to be avoided is the submission of black and white (‘monochromatic’) images in the RGB or CMYK color spaces (which often goes unnoticed when images are viewed on-screen). Images inadvertently submitted in this way may contain hidden color information which, when converted to grayscale for monochromatic printing, may result in unexpectedly poor reproduction (Figure 4). Truly monochromatic images should, therefore, be converted to and previewed in grayscale, adjusting contrast where necessary, prior to their submission. The ‘black and white’ image (a) is, in fact, in the RGB gamut. When converted to and printed in grayscale (b), a noticeable loss of contrast occurs. Reproduced from Berg et al.4. At the point of image capture, ALWAYS ensure the image is saved as a TIFF. When this cannot be done, Windows Bitmap (‘.bmp’) acts as a good intermediate prior to editing on graphics software, though as an uncompressed format it does take up more memory space. AVOID saving files in JPEG format, unless additional options exist to specify the degree of image compression. In this case select the minimum compression/maximum quality possible and ensure all previous ‘working’ saves have been carried out using a lossless format (i.e. TIFF or Windows Bitmap). Subsequent to image capture, ensure that any manipulation of the image does not cause a reduction in resolution. We require single (84-mm) and double (175-mm) column-width color and gray-scale images at a minimum of 300 dpi. Be aware that scaling-up smaller images will cause a reduction in resolution. Remember to save a copy of the original image in an uncompressed format as a back-up in case of potential problems with the manipulated version. Always preview images in the correct ‘color space’. For RGB images containing vibrant or artificially-generated colors, check to see that detail is not lost when the image is previewed in CMYK. For images not destined to be printed in color, especially sepia-toned 3D ultrasound images, please convert the image to grayscale before submitting to the Journal. If necessary, use the ‘Auto-contrast/levels’ feature found on many graphics editing programs to help restore lost clarity (Figure 5). Be wary of embedding images in other files (eg. Word, Powerpoint, Excel). Inserting/extracting images into/from different programs subjects the files to unpredictable changes in quality. NEVER copy-and-paste an embedded image from one file format to another prior to submission. While embedded images are useful guidelines to the intended layout of the manuscript, always attach the original TIFF file when submitting papers. The author supplied this image in the correct color space and at the correct resolution. However, when published, the contrast of the image was poor (a); the histogram below (b) shows its narrow tonal range. Applying Adobe Photoshop™'s (Adobe Systems Incorporated, San Jose, CA, USA) ‘Auto-contrast’ function significantly improved the diagnostic quality of the image (c); the histogram below (d) now shows the full tonal range of the image, from ‘100% black’ to ‘100% white’. Reproduced from Guerriero et al.5.