Image Compression and File Formats
Image compression reduces the size of images, saving disk storage space and reducing the time required for sending the images over a network. The large file sizes necessary for colour images has encouraged the use of compression for retinal images.
Image compression methods may be divided into two categories: loss-less and lossy.
Loss-Less Compression
Loss-less compression is error free; in other words, the uncompressed image is completely identical to the original image. Unfortunately the size reductions possible using loss-less compression methods with retinal images is relatively modest, typically reducing images to only half of their original size. Loss-less compression is found in a number of graphic file formats including TIFF (Tagged Image File Format), TGA (Targa), and PNG (Portable Network Graphics).
Lossy Compression
Lossy compression sacrifices image information, which it considers will not be noticed visually. The uncompressed image is therefore different from the original image. Such methods should therefore be used with care for medical images. The most popular lossy compression method is JPEG (named after the Joint Photographic Experts Group who developed it). JPEG allows the amount of image information which is discarded to be controlled using a “quality setting” – a parameter which is commonly misunderstood. Although often expressed as a percentage, the value does not represent the proportion of image data which is discarded; neither does it represent the size of the compressed image. Furthermore, different software packages use completely different scales: some produce the best quality images at quality setting 0%, and others produce the highest quality at 100%. Specifying the settings used to produce JPEG files is therefore useless unless the software package used is also specified.
Much greater file size reductions of retinal images are possible using JPEG compression, depending on how much distortion can be tolerated. JPEG compression can introduce unsightly artefacts in the image, usually appearing as 8×8 blocks of pixels due to the way the algorithm operates.
Display Monitors
There are two types of monitor in common use. The first is based on a cathode ray tube (CRT) similar to that in a television set. It is capable of producing bright and clear images with good colour rendition. The main disadvantage with CRT monitors is that they are large and heavy. An alternative is a flat screen technology made from thin film transistors (TFT). These are much smaller and lighter making them ideal for mobile screening units. However, they currently do not have the contrast range or accurate colour rendition of the CRT. They are particularly poor for viewing greyscale images.
Power Failure and Digital Imaging
Patient images can be lost if there is a power failure, an uninterruptible power supply (UPS) will keep the camera and computer system powered in the event of an electrical failure. The UPS has a battery, which, if the correct duration model is selected, should last long enough to allow the computer to operate for at least 5-10 minutes following a power failure. With an UPS, if the power were turned off, all recently produced images can still be saved into the hard disc.
Like all battery powered equipment a UPS requires maintenance so that it will function correctly when needed. It is necessary to check the battery function at regular intervals and the storage batteries require changing approximately every four years.
Storage, Archiving and Backup
There is currently a medical-legal requirement to store all digital images, which have been taken during the screening programme in their original uncompressed capture format. As large numbers of photographs will be captured, these will quickly fill up the hard drive of the computer and consequently it will become difficult to use the computer, as the computer will slow down. Images therefore, need to be archived to optical disc media, tape or CD-ROM. The images can then be deleted from the capture computer’s hard disc.
Conventional or film based photographic images are analogue. An analogue image always has an original form to which it relates; this can be in either negative or positive form. Copies can be made from the original and normally there will be a slight degradation in quality, but as further copies are made from these copies then the image quality deteriorates further.
Digital imaging uses numbers to display data or information. As the original image consists only of numerical data, exact copies can be made and images can be made from these copies without degradation. Each stage of copying is precise and there is no loss of image quality between generations and, if stored, there should be no deterioration over time. Furthermore, as the images are represented as numeric data, they can be transported over phone lines, satellite or computer networks.
Once the images have been acquired and stored, they must be easily accessible by the graders. The use of a central dedicated computer or server, which can be accessed via a networked system, is recommended. The management of these images is via a database. The patients can be accessed via the database and the images reviewed at grading stations, which have access to the server and through a networked computer system.
Downloading
Once images are in the computer’s storage device, they will lie there until they have been transferred or downloaded to another computer or device. Uploading and downloading refer to the direction of transferring files between computers. Uploading involves sending a file to another system and downloading involves transferring a file from another computer to your own computer. The most common place to download retinal screening images is to a ‘main frame’ or central computer system where the images can be managed.
Downloading can either be done through special devices such as flash card readers or card adapters or can be done through cable connections such as parallel ports, SCSI (fast), or USB (faster), Firewire-IEE1934 (fastest).
Databases
A database is computer software, which tracks images and patient information. Important information such as patient name, hospital number, visual acuity and other relevant information can be entered into the system. The software should also be capable of entering the date of the procedure, the location of the patients’ images (disc and path), the date when the photograph was taken and any other relating data. The system must also be capable of adding further imaging episodes that might take place during subsequent visits. The patient database is an ‘inroad’ to recalling images. For example, selecting a required file should be possible by querying a date or hospital number and selected images should be able to be pulled up to a full-screen image with textual information being displayed. In addition to viewing a single image, groups of images may be required for comparison purposes. This should be done using the choices from the software menus.