Digital pathology is now widely accepted as an essential part of the future for histopathology departments. The technology has improved massively in recent years, and advances in computational pathology are also coming to market, with advances in artificial intelligence applications not far behind. Both technologies rely on moving histopathology to digital imaging.
With the forecast shortfall in consultant histopathologist numbers in five years and ever increasing workloads, it is difficult to justify leaving consultant histopathologists with inefficient workflows that waste their time looking for slides, matching slides with previous samples and sending slides through the postal system and via couriers for second and third opinions.
Having recently constructed a business case for digital pathology across nine hospitals in the UK, here are my thoughts on the main areas to focus on when examining the business case.
Whole slide image file sizes typically range from 0.6Gb to 2Gb, and with most cases generating multiple slides, storage will be the largest area of cost. Issues to consider are:
How many months/years of images need to be kept ‘on-line’ for immediate and fast access?
Typically 3 months to 1 year’s worth of images are stored on-line.
Scanning at 20x equivalent magnification results in many re-scans, when pathologists need to view at greater magnification with no loss of image quality. A 20x magnification scan typically has a file size that is around two thirds the size of a 40x magnification scan, which is around two thirds the size of a 60x magnification scan. The norm is now to routinely scan at 40x equivalent magnification, as this provides the best balance between storage costs and re-scan costs. A 40x scan can be digitally magnified to 60x if required, and this is usually of acceptable quality.
Once the on-line access storage threshold has been reached, will images be deleted, archived to lower cost storage media such as tape, or compressed?
Deleting images that have been used to make clinical decisions feels wrong, but in the UK the Royal College of Pathologists does not support the destruction of slides after whole slide imaging. Therefore, if images are deleted, the original slides can be retrieved and re-imaged. Any annotations or metadata associated with the original images will of course be lost.
Archiving to tape can reduce the cost of storage by 85% compared to replicated SATA storage.
Compression reduces image size significantly, but permanently reduces the quality of the image. There are a number of ways to compress, including removing the 40x layer.
After storage, whole slide image scanners are likely to be the next largest area of cost. Here there are economies of scale if a single slide production and scanning centre can be set up to serve multiple laboratories e.g. if 12 whole slide image scanners are required across multiple hospitals, only around 7 would be required if centralised (whole slide image scanners cost around £125k to £175k each). Other benefits of a single slide processing centre are:
All formalin fixed paraffin embedded (FFPE) tissue can be stored at the centre to produce slides for pathologist requested follow on investigations e.g. immunohistochemistry stains.
All FFPE tissue and slides are produced in a standardised way, supporting site agnostic reporting. If images are shared for viewing at different hospitals using the same slide processing centre, they will not find the haematoxylin and eosin stain to be a bit pinker than they are used to, for example. Standardisation at a single centre therefore supports site agnostic reporting better than each site preparing, staining and scanning its own slides.
What is to be scanned?
At this point in time, only histopathology slides can be scanned. Non-gynaecological cytology slides are usually too thick, and would require scanning on multiple focal planes (although technology is developing to handle this). Gynaecological cytology in the UK is rapidly moving over to human papilloma virus (HPV) testing, so smear slides will become a thing of the past. Although only histopathology slides can currently be scanned, there is still a choice to be made, as follows:
- All histopathology slides. This is the standard approach.
- Biopsy slides only. In a typical UK district general hospital, these will be about 20% of all histopathology samples. These are the time critical samples, used for diagnosis and disease staging, and are required for cancer MDT discussion before starting treatment. They are also the samples that most commonly have follow on slides for special stains such as immunohistochemistry. Biopsy slide images tend to contain about a fifth of the digital volume of a resection sample. It is therefore possible to get around 80% of the patient benefit for around 20% of the cost of scanning all sample slides. Conversely, resection samples are rarely urgent, as diagnosis and staging have been completed and treatment started, and the histopathologist is mainly looking at the margins and confirming the grade of the tumour.
Typically around 20% of consultant histopathologist time can be released if all sample slides are scanned at 40x equivalent magnification and stored on-line for 1 year (obviously less if only biopsy samples are scanned). For the business case, it is important to agree how this time will be used, to forecast the cash releasing benefit. For example, if a proportion of the released time is used to report additional cases, forecasts can be made on the reduction in locum costs, outsourced reporting costs and more controversially reduction in programmed activities above 10 (i.e. a reduction in excess hours).
Written by Mark Magrath - Associate, The HCI Group
For help with your business case on digital pathology, please contact HCI. We can also advise on many other aspects of digital pathology, such as LIMS based reporting compared to PACS based reporting, procurement options and qualitative benefits.