Technical Specifications

When planning a digital microscopy system for diagnostic use, consider:

It is important to engage pathologists, laboratory staff and IT departments when assessing requirements to ensure the proposed system is scalable to meet future increases in workload.
The IT department should be engaged to determine a number of important requirements:

1. labour requirements – additional staff may be required as well as associated training costs, Isaacs and colleagues explained that their laboratory required a 0.5 FTE (full time equivalent) digital imaging technician and a 0.3 FTE information technology (IT) support person for a diagnostic workflow where 2.7% of the workflow was scanned.
   

   10. Isaacs, M., Lennerz, J. K., Yates, S., Clermont, W., Rossi, J. and Pfeifer, J. D. (2011). Implementation of whole slide imaging in surgical pathology: A value added approach. J Pathol Inform, 2, 39. doi: 10.4103/2153-3539.84232

2. additional costs that may be incurred for quality assurance, validation and equipment purchases;

3. analysis and costs for any additional time taken to diagnose using digital microscopy versus conventional microscopy

It is important to engage pathologists, laboratory staff and IT departments when assessing requirements to ensure the proposed system is scalable to meet future increases in workload. The IT department should be engaged to determine a number of important requirements:

1. assess interoperability with existing systems including the Laboratory Information (Management) System (LIS).
   

   11. Murray M. Laboratory Information Management Systems. http://logistics.about.com

   A LIS manages the specimen workflow from the entry of the specimen into the laboratory and creating a unique barcode label to the reporting and archiving of the specimen.

2. estimate file storage requirements and cost including short term storage and archiving

3. assess network and communications requirements including secure workstations and storage devices. Network bandwidth should be estimated and the impact of transmitting large size files and potential bottlenecks on the current network should also be assessed. Bandwidth represents the capacity of a network connection for supporting data transfers. Higher network bandwidth often translates to better performance, although overall performance also depends on other factors. Computer network bandwidth is measured in units of bits per second (bps). Most modern network devices support data rates of thousands and often millions or even billions of bps ( in units of Kbps, Mbps and Gbps). Network devices each possess a bandwidth rating according to the maximum data rate they are physically capable of supporting.
   

   12. Mitchell B. Bandwidth. http://compnetworking.about.com. 22 May 2015

4. The Digital Pathology Association recommends the Local Area Network (LAN) bandwidth to be 100 Mbps4 and the Wide Area Network (WAN) bandwidth to a minimum suitable for the screen resolution, image compression and tissue types so the pathologist can work productively with the displayed images.

5. assess privacy and security requirements to ensure the digital microscopy system meets jurisdictional legislation and the organisation’s privacy and security policies.

6. determine if additional software (such as interfaces, virus scanners and firewalls) need to be written or sourced.

7. assess and cost additional hardware that may need to be acquired, including monitors (consider screen size, graphics cards, graphic drivers), navigation devices and storage devices. To date there have been no studies that have examined the accuracy of diagnosis when a digital slide is viewed on small devices such as Smart phones and tablets, and so extensive validation is required before these devices are used for diagnostic purposes. Additionally, there is an issue with patient confidentiality that must also be considered before these mobile devices are used for diagnostic purposes.

8. ongoing IT expertise required for the support of interfaces, operating systems, databases, hardware and software configuration, network.

Physical characteristics should be considered in conjunction with the technical characteristics as these may affect the performance and usability of the system, and be mindful of the optimal operation conditions recommended by the manufacturer.
Physical factors to consider include:

• ensuring sufficient space for the digital microscopy system, workstations and other utilities; bench space to store slides before and after loading, distance between workstations, etc.
• optimal conditions as specified by the manufacturer i.e. room temperature, bench and floor stability, lighting, optimal size of the section on slide (typical size is 15mm x 15mm).
• noise, vibration and sounds when the equipment is in operation such as alarms, mechanical parts, system fans, etc.
• workstation requirements such as adequate chairs with back and neck support as well as glare from lights and sunlight onto monitors, glare filters for screen monitors
• system instructions and labelling including instructions that may be on the system, in operating manuals or via online help.

For more details see RCPA Position statement, Ergonomics

• The glass slide is mounted into a frame
• The frame is positioned under an objective lens
• By means of a digital camera, the first tile is focussed automatically
• The camera saves the image
• The frame is positioned to the next field for saving the next image tile
• This way hundreds to thousands of images are generated, which together represent the whole tissue in highest magnification
• All image tiles are stitched together to form a single image
• Finally, the image is saved in a suitable, mostly compressed, image format

It is recommended to consult with the scanner manufacturer for the optimal size of the tissue section but typically it is 15mm x 15mm. It is important to remember that the larger the tissue section on the histological slide or smear on the cytology slide to be scanned, the larger the final file size of the digital slide.
Scanners vary in the number of slides they can scan simultaneously. Scanners may have automatic specimen recognition which ensures a more efficient and faster scanning process. Manual focal points can be added to assist in the quality of the scan.

Figure: Scanning areas can be selected with the addition of focal points (green marks), here using an Aperio ScanScopeTM XT Digital Slide Scanning System (Leica Biosystems North Ryde, Sydney, Australia)
Source: Courtesy RCPAQAP, 2017

Figure : Tile-based -based scanning method of a glass cytopathology slide is shown in progress.
Source: Courtesy RCPAQAP, 2017

Laboratories should be encouraged to formulate a table of specifications for all scanners available. The scanners available will vary from year to year.
Specifications that should be covered include:

• platform characteristics
• capacity of the platform in terms of time for one throughput cycle
• number of slides scanned in each cycle
• average slide size of slides produced
• studies that assess clarity and resolution of WSI produced.

 
The digital microscopy system must capture a high quality and resolution image of the glass slide. Poorly prepared slides can result in poor WSI quality but as for conventional microscopy factors that may affect the quality of the slide include:

• thickness and size of the section,
• slide staining and drying time,
• artefacts on the glass slides (such as dust, fingerprints), and
• air bubbles under coverslip.

 
The resolution and quality of the captured image can be affected by a number of other factors:

1. Method of scanning the slide i.e. tile or linear scanning pattern;

2. Number of axes scanned, that is multiple-axis (X and Y axis) or multiple stacked axes (X, Y and Z-axis) or Z-stacking. Z-stacking involves scanning at multiple horizontal focal planes at a nominal inter-plane distance and some machines can scan up to 31 planes; Z-stacking could be useful for slides for cytology, especially viewing cell clusters, as well as small biopsy histopathology, lymphoma pathology or FISH. Z-stacking may provide depth of focus but will lead to larger file sizes.

3. Distribution of focal points across the scanned tissue area.

4. Number and dimensions of pixels of the scanner – Images are created by pixels or individual dots, and each individual pixel will contain a different colour. As the dots are placed next to each other and aligned, the combination of the pixels forms the picture on the screen.
   

   03. Farahani N et al. Whole slide imaging in pathology: advantages, limitations, and emerging perspectives. Pathology and Laboratory Medicine International. 2015. 7: 23-33.

5. Scanning magnification – the scanner should at least be able to scan at x20 and x40 magnification. Most optical microscopes have an eyepiece which provides x10 magnification and with a x40 objective, provide x400 magnification. Digitising scanners do not use an eyepiece and may not use a microscope objective lens; by convention images captured with a resolution of 0.25mpp are referred to as x40 images and likewise, images captured with a resolution of 0.5mpp are referred to as x20.
   

   15. Silver R. Pixel. http://www.buenosearch.com. December 2014

6. Scanning magnification options – the magnification will be set to the lowest magnification that provides all information needed for accurate diagnosis. This may be determined by the type of the case, pathologist preference or laboratory procedures. For example, the RCPA Invasive Breast Cancer structured pathology reporting of cancer protocol states to use x40 objective to perform the mitotic score. Magnifications of x60 may be useful in cytology and lymphoma pathology.

7. The viewable resolution is limited by the scanned resolution. Typical scanned resolutions are 0.5 micron/pixel (x20) and 0.25 micron/pixel (x40); the latter can generate 900 million image pixels per square centimetre of tissue section.

8. Colour depth, which is typically 24-bit colour. Colour depth, also known as bit depth, is either the number of bits used to indicate the colour of a single pixel, in a bitmapped image or video frame buffer, or the number of bits used for each colour component of a single pixel. The stored value is typically a number representing the index into a colour map or palette. 24-bit represents a 224 or 16 million colour palette. At 24 bit, this is 2.7 GB of data per square centimetre of tissue section. Also consider the availability of software tools for colour, white balance, contrast and exposure. It should be possible to also adjust these in the view, like adjusting a microscope. Image colour can be influenced by other factors such as staining and software manipulation. Measures should be in place to safeguard against intentional or unintentional corruption or manipulation of the digital images and data (by way of rotation, and adjusting brightness and colour of an image) as such alterations have been found to significantly alter the final diagnosis.
   

   16. DICOM Standards Committee, Working Group 26, Pathology. DICOM Supplement 145: Whole Slide Microscopic Image IOD and SOP Classes. Virgina USA. 24 August 2010. 

1. Slide feeder should be reliable and be able to feed the scanner without damaging the slide.

2. Scanning speed should be adequate for the expected laboratory workload – this may be dependent on the size of the tissue area to scan and magnification required. For example, using x20 objectives for a single slide may take anywhere between 30 seconds and 5 minutes.

3. The scanner should support common barcode formats. In its simplest form, the barcode is a series of black lines of varying widths on a white background. The bar code itself does not contain any information. The identifying information is stored in a computer and is matched to the bar code based on the configuration of the individual lines. The bar code is read by an optical scanner that measures the relative widths of the spaces and bars. The computer interprets the scanned code and matches it to the distinct item in its database. There are international and industry standards for bar codes. Two-dimensional (2D) barcodes are gaining popularity for the ability to encode a greater amount of information in a smaller space. These barcodes can have up to 2,000 characters, as opposed to about 20 to 25 in a standard bar code. 2D barcodes are capable of encoding more than one identifier as data density is higher. Data integrity checks are more rigorous and error correction algorithms can be used. These barcodes also have a higher tolerance for printer failures and damage compared with linear bar codes.
   

   18. Parker M. Barcodes Explained. http://smallbusiness.chron.com/bar-codes-explained-39539.html

For the scanning system to function reliably it is recommended that the read rates for the slide and barcode/metadata are greater than 95%. The barcode read rate may be affected by the print quality, format, placement of barcode label etc and it is recommended that the slide is labelled with legible information (unique number identification and patient name).

• The network connection between the scanner and the storage device/s should be capable of transmitting as fast as the scanner can send the image file.
• The scanning system should be able to operate off-line if there is a network or LIS outage.

The user should have the ability to manage and monitor the mechanics of the scanning system and have adequate industry support.

The file size of the captured image is dependent on:

• dimensions of the tissue or cytological material on the glass slide,
• magnification of the scanning lens and number of Z-planes.

DICOM Supplement 145: Whole Slide Microscopic Image IOD and SOP Classes produced in 2010 by the DICOM Standards Committee, Working Group 26, Pathology. The document describes image characteristics and storage, which are outlined below.

1. To assist rapid panning, image data should be stored in a “tiled” fashion. This allows access to any subregion of the image without loading large amounts of data. To assist rapid zooming, the image may be stored at several preconfigured resolutions. If multiple Z-planes are captured, these may be stored as separate images.

2. The simplest way to store the image data is a single frame organisation where data are stored as pixels in rows like text running across a page.

3. In order to view an area, a much larger subset must be loaded. This has an important limitation for large images. A more sophisticated way of storage is a tiled organisation.

4. Pixels are stored in quadrilateral tiles which are in turn stored in rows. This has an important advantage for large images as only a small subset of the image needs to be loaded to access an area.

5. This organisation facilitates rapid panning, however there is still an issue with rapid zooming. The problem occurs at lower resolutions whereby progressively larger image areas must be accessed and down sampled.

6. A solution is to pre-compute lower resolution versions of the image and thus the image may be considered a “pyramid” of image data.

7. WSI consists of multiple images at different resolutions where the height of the pyramid corresponds to the zoom level. The thumbnail image at low resolution and intermediate resolution images assist retrieval of image data at these arbitrary resolutions.

8. Where multiple resolution images are required, as in Z planes, each “level” is stored separately in the series.
   

   19. Snyder et al. Patient misidentifications caused by errors in standard bar code technology. Clin Chem. 2010. 56(10): 1554-1560.

Digital slides have image dimensions of up to 7.2 million pixels. Uncompressed, these images are up to 40 GB in file size, compressed up to 2 GB. Many digital microscopy systems will use file compression for storage and transmission of a file. Smaller biopsies form a smaller file size than excision specimens for example, however on average one could produce 1 GB per slide. A typical digital slide is 1.6 million pixels and would require 4.6GB of memory.

Efficient file compression should be applied, ensuring the digital slide can be reconstructed on a standard image viewer without compromising image quality. Also, as the sizes of the captured image are very large an efficient compression method should be used for storing and transmitting the image.

1. Lossless compression reduces the size of the file without loss of quality when the image is uncompressed. There is typically x3-x5 reduction in file size.

2. Lossy compression reduces the size of the file by eliminating some data and when the image is recreated on a viewer there is often a degree of loss of quality. There is typically x10-x50 reduction in file size. Lossy compression may be used only if the quality of the image is suitable to perform a reliable diagnosis. Studies by Foran, et al and Krupinksi, et al show that varying degrees of lossy compression can be applied without significant decrease in quality relative to light microscopy. Examples of lossy compression are JPEG and JPEG2000; JPEG2000 yields higher compression and fewer image artefacts but is compute-intensive.
   

   20. Thorstenson S et al. Implementation of large-scale routine diagnostics using whole slide imaging in Sweden: Digital pathology experiences 2006-2013. J Pathol Inform. 2014. 5: 15

   
   

   21. Foran, D., Meer, P., Papthomas T.,Marsic, I. Compression Guidelines for Diagnostic Telepathology. Accessed from http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.4.1522&rep=rep1&type=pdf

Diagnostic performance has been shown to be a function of level of compression and degree of compression ratios, ideal pixel resolution, monitor resolution and contrast ratios have been determined for maximal diagnostic accuracy when using WSI.

Examples of file compression:

1. A x20 0.5micron/pixel, 15mm x 15mm section has an uncompressed file size of 2.51GB. With compression (e.g. lossy compression JPEG) with 1:20 compression the file size becomes 128 MB; or with 1:10 compression, 256 MB.

2. A x40 0.25micron/pixel, 15mm x 15mm section has an uncompressed file size of 10.1GB. With compression (e.g. lossy compression JPEG) with 1:20 compression the file size becomes 502 MB; or with 1:10 compression, 1GB.

3. The size increases with the number of linear planes scanned (or Z-stacking) so if the slide is scanned with 9 planes, the file is 9 times the size.
   

   02. Kalinski, T., Zwönitzer, R., Grabellus, F., Sheu, S.-Y., Sel, S., Hofmann, H., . . . Roessner, A. (2009). Lossy compression in diagnostic virtual 3-dimensional microscopy—where is the limit? Hum Pathol, 40(7), 998-1005. doi: 10.1016/j.humpath.2008.12.010

The verification and validation process in section 4.1 of the RCPA Guidelines for Digital Microscopy in Anatomical Pathology and Cytology October 2015, can be used to determine the degree of compression that will still allow reliable diagnosis taking into account the local specimen mix and diagnostic tasks required. Compression algorithms and resultant degree of acceptable loss should be validated to be diagnostically accurate.

All images used for diagnostic purposes must be stored on a secured device in a secure location.

All storage devices (including database servers and backup devices) used for storing and archiving images and related case information, must also be kept in a secure location.

There must be an efficient and reliable method for storing and retrieving images (including from online and archive file systems).

Data integrity must be maintained, so the storage device must ensure that the data is stored reliably, and there is no data loss or corruption if the storage device fails. An appropriate archive and backup process should also be put in place to protect image data and associated metadata according to regulated preservation guidelines and durations.

The storage device should be able to store as fast as the scanner can transmit the image file. If the storage system is connected to an LIS, then the storage device should also be able to store as fast as the LIS can transmit data. Archived data must be stored on robust storage devices and not a portable media format (such as laptop, smartphone, tablet, or USB stick).

The case data (including digital slide images, patient and clinical information) can only be modified with appropriate tracking.

A secure transmission protocol should be used for image and data transfer such as Hypertext Transfer Protocol (HTTP) across networks. This is a protocol used to request and transmit files, especially webpages and webpage components, over the Internet or other computer network. (REF 24)

The stored digital slide image must be associated with the machine-readable identifier.

Storage must be able to associate the scanned machine-readable identifier to the scanned image, and depending on the communication with the LIS the patient metadata may also be linked to the record.

If the digital microscopy system is integrated with a LIS, the minimum information to be stored in the digital microscopy system storage for each case includes:

• Patient identifiers (minimum of two) as per NPAAC requirements in Australia,
• Patient data,
• Case information including accession number, block number, stain types,
• Appropriate timestamps of transmission records,
• Tracking of details for audit functionality and pathologist comments.

Each scanned image should contain a record of the following information at a minimum:

• file dimensions;
• colour depth;
• image resolution;
• creation date and time.

Each image captured for a given identifier should be stored and be able to be viewed or deleted by the user, as required, with appropriate tracking.

The stored image and related case data must be available for the minimum retention time.

When WSI are used for diagnostic purposes, the storage of the WSI and associated diagnostic material must satisfy the existing retention times for glass slides; for Australia, as stated in the NPAAC publication Requirements for The Retention of Laboratory Records and Diagnostic Material.

Glass slides must also be kept for the minimum retention time; for Australia, as stated in the NPAAC publication Requirements for The Retention of Laboratory Records and Diagnostic Material.

1. Objective

   To understand the technical specifications to consider when implementing digital microscopy in the workplace for diagnostic use in histopathology and cytology.

2. Knowledge

   Section1: Set Up Requirements

   Outcomes: Understand setup requirements to consider when implementing digital microscopy for diagnostic use, such as:

   1. Labour requirements;

   2. Additional costs of time;

   3. Additional costs for quality assurance;

   4. Interoperability with existing LIS;

   5. File storage requirements;

   6. Network and communication requirements;

   7. Privacy and Security;

   8. Hardware and Software.

   Section 2: Pathologist Workstation

   Outcomes:

   1. Understand physical characteristics of the pathologist workstation;

   2. Understands the importance of safety and ergonomics of the pathologist workstation.

   Section 3: Image Acquisition

   1. Outcomes: Understand the principles of image acquisition when implementing digital microscopy for diagnostic use, including:

   2. Correct tissue size;

   3. Quality of glass slides;

   4. Scanning resolution and quality;

   5. Scanning reliability.

   Section 4: Image File Characteristics

   Outcomes: Understand image file characteristics when implementing digital microscopy for diagnostic us, including:

   1. Pyramid of image data;

   2. Multiple Z-planes and colour planes.

   Section 5: Image Compression

   Outcomes: Understand principles of image compression when implementing digital microscopy for diagnostic use, such as:

   1. Acceptable file format;

   2. Acceptable file size;

   3. Compression algorithms validated.

   Section 6: File Storage and Archive

   Outcomes: Understand the principles of file storage and archiving to consider when implementing digital microscopy for diagnostic use, including:

   1. Images and storage devices in secure location;

   2. Reliable and fast storage and retrieval;

   3. Scanned image data and integrity maintained;

   4. Case data changes can be tracked;

   5. Integration into LIS confirmed;

   6. Storage complies with NPAAC guidelines (Australia).

3. Behaviours

   Practices the fundamental principles of digital microscopy.