Digital Diagnostic Imaging Recommendations

Recommendations on the Delivery of, Access to and
Viewing of Diagnostic Quality Digital Images for Clinicians

The move from an analogue, or screen based x-ray capture to Digital Diagnostic Imaging capture and distribution is inevitable, however many clinicians may not have had sufficient exposure or understanding to appreciate the complex technical specifications involved in such changes. 

Imaging can be divided into three (not necessarily mutually exclusive) broad categories:

1. Diagnostic imaging

2. Interventional imaging

3. Imaging for teaching and research

This web site predominantly addresses category 1, and more specifically, the process and form in which such images are made available to the clinician looking after the patient. It aims to provide guidance to clinicians who have a need to review diagnostic quality images in the course of patient care. A number of specific imaging needs related to certain medical specialties (e.g. cardiology, radiation oncology) are not specifically covered at this stage, however the common utilisation of the DICOM format does mean that the image data can be distributed and accessed using the same principles.

It is also recognised that significant technological advances in diagnostic imaging have led to a variety of image formats and functionality requiring innovative solutions to maximize the effectiveness of delivery, access, manipulation and archiving.  Variations in display matrices and bit depth, three dimensional display methods, multi planar reconstruction, Image fusion, dynamic imaging and colour rendering all support the progressive move from a film based delivery approach to digital format. 

However, when patient care decisions depend upon the review of the diagnostic images, those responsible for the ultimate clinical management of the patient must be involved in any decision to adopt a material change in the way images are provided.  The transition to digital imaging must not compromise the diagnostic and treatment capabilities of the clinician, and should facilitate, not hinder, appropriate image access at the patient care interface.

Similarly, it is important that referrers have realistic expectations in requesting “film” in all cases. Multislice CT studies often create over 1,000 images per patient study, and an element of selection must be applied judiciously when key images are filmed. When selected images are supplied on film (or other hard copy) they must be appropriate to the clinical setting.

Equally, it is not always possible to anticipate at the outset all the possible uses of, and requirements for, a diagnostic imaging examination. While precise communication between the referrer and the provider can minimise difficulties concerning examination type, processing and delivery, there should also be a mechanism whereby additional appropriate requirements (e.g. those arising from on-referral) can be met expeditiously.

Data Content of Diagnostic Imaging Examinations

In considering contemporary imaging, it is useful to differentiate between images such as plain radiographs (large matrix) and those generated by other modalities, including CT, MRI and ultrasound (small matrix). The diverse matrix sizes, and the variable numbers of images generated by the various modalities, pose different challenges in image display.

Plain (digital) radiography typically utilizes large matrix (e.g. 2000 x 3000) resolution images.  The recommended minimum resolution for diagnostic quality is 2.5lp/mm.  This equates to approximately 127ppi, and at 10 or more bits per pixel, the major challenge is the size of the image and the need for a large high quality monitor to allow full size display, and adequate network or transfer capacity as images range from 12 – 40MB in size.

Cross-sectional images obtained from CT and MRI is typically small matrix (e.g. 512 x 512) at 12 - 14 bit images, and is more easily shared digitally.  They are captured digitally, and individual image size is generally less than 1MB.  The major challenge from these modalities is the large numbers of images often obtained (~ 100 – 3000) and the collective size of the entire data set.

Imaging modalities classed by data content

Within Diagnostic Imaging there are three broad groups of image data sets (with some overlap between these):

1. Plain Radiography

(e.g. Plain x-rays, Mammography)

- typically large matrix with relatively few images

Full size, diagnostic quality (high resolution and measurable#) images are required for diagnostic and therapeutic measurement and templating, as the treating clinician may need to use these images for critical decisions regarding prosthesis positioning or loosening,  fracture patterns, sequential comparison of suspicious pulmonary lesions and so on.

• Referring and treating doctors require access to the full set of images in diagnostic quality when in their judgment these are needed.
• Digital images should be of comparable diagnostic quality to analogue (screen) film.
• Computed Radiography (CR) / Digital Radiography (DR) have lower image spatial resolution than analogue film but advantages with respect to contrast resolution, magnification and distribution. However, the following quality issues have to be addressed depending on the means of image display:

1.       Film printing:

• must be to a “diagnostic quality” standard
  
• High resolution (not < 300dpi);
• Full size– (image displayed at same size as captured)
• Measurable (# see below)

2.       Digital Display:

• Image data in IHE DICOM compliant format (therefore will be lossless)
• Convenient and rapid data loading;
•  Viewed on an adequate monitor – high resolution and Grey Scale Display function     GSDF DICOM (Part 14) compliant;
• Presenting LossLess image resolution (DICOM protocol);
• Display software to provide a suitable interface for the clinician.

      # Note: Radiographs always magnify the imaged anatomy. With analogue methods, standardised techniques are employed (fixed focus-film distance and minimum object-film distance) so that the degree of magnification is fairly reproducible for a particular examination and patient.  An experienced observer can then estimate real size from image size fairly reliably.
Digital imaging allows arbitrary re-sizing of the displayed image after it has been recorded.  Unless re-sizing is avoided (images are displayed “full size”), or there is an unambiguous description of the extent of any re-sizing, measurement estimates can be critically compromised.

2. Dynamic and Complex Diagnostic Imaging

(e.g. scintigraphy, angiography, ultrasound, fluoroscopy)

- typically small matrix with relatively few images retained, and / or dynamic image series
                    (though some evolving applications generate large numbers of images)  

These image types may require direct imaging specialist involvement, vascular access, real time dynamic analysis, complicated quantitative analysis, and often time sensitive image capture. This document does not specifically address the transfer of these types of images, although, if the recorded data complies with the DICOM protocol, as will usually be the case for at least the static component, the images may be handled as for other small matrix examinations. 

3. Cross-sectional imaging (of a defined volume)

(e.g. CT, MRI, SPECT/CT, PET/CT)

- typically small matrix, but with many images

Large amounts of cross-sectional data are recorded and manipulated, and the Radiologist assessment and diagnosis are important. However, adequate assessment of the case by the clinician may require correlation of the clinical features with detailed assessment of the imaging findings, and thus access by the treating clinician to the complete image data in electronic form is frequently required. This is especially important in urgent situations when treatment is required before the formal radiological report is available.

*** Specific issues that promote a move to digital delivery include:

• The large number of source images (in CT typically thin-section transverse (“axial”) cross-sectional images) sometimes cannot be satisfactorily viewed on film
• Digital presentation has major advantages for image viewing and manipulation, for example the ability to scroll through the images and change presentation states.
• Image manipulation and 3D Image viewing by the treating doctor are required in certain situations, again only achievable with digital distribution and display techniques.

*** Where used, non-network digital media should be:

1. Fast loading - Loads quickly (on an appropriate end-user platform)

2. Reliable, non-volatile and with robust data storage;

3. Displayed on a simple intuitive software interface.

4. Uneditable

5. Externally labelled for content

The quality of the images supplied to treating doctors, where diagnostic information is required, should be equivalent to those viewed by the radiologist at the time of reporting.

Anatomical Reference Images for Cross-sectional imaging 

In the delivery of cross-sectional images, either digitally or by hard copy, it is critical that a spatial location guide , indicating the relationship of each cross-sectional image to standard anatomical landmarks, be provided (these may be known as ‘scout images’ or ‘pilot’ views, topograms, etc.)

This applies whether the images are delivered on film or on digital media, however for digital media this should be part of the functionality of the DICOM reading software.

Industry-standard format for digital image data

Diagnostic-quality images distributed in digital format must comply with DICOM standards and IHE PDI profiles if distributed on portable media. For both small and large-matrix digital image data, the data set should provide full resolution data for processing, manipulation and subsequent display if required depending on the clinical requirements.

Compression

The DICOM format includes provision for “lossless” compression of images, where the volume of image data is reduced (typically by a factor of 2 times), without affecting the appearance of the displayed image.

Higher degrees of data compression are usually “lossy”, i.e. the displayed image is of lower quality than the uncompressed original. Depending on the nature of the examination, and the nature and degree of compression employed, lossy compression may or may not affect the diagnostic value of an image.

Use of lossy compression may be considered where it has been shown that the degree of lossy compression employed has no effect on diagnostic quality.   However, in most situations, primary or secondary diagnosis should be on lossless images.