Unleashing the power of digital image analysis
1 Mar 2021
The practice of image analysis has been evolving for years, but the UK’s lockdown has hit fast-forward on its shift into the digital world. Here, Andreas Theodosi discusses the impact that COVID-19 has had on the industry and the benefits of digitalisation.
Image analysis is the extraction of meaningful information from images; it’s the quantification of pathology data. Many laboratories within the pathology field have been gradually moving towards more digital setups for some time now, as technological advances offer new and exciting routes to useful data. And naturally, as one component in a laboratory is digitalised, overlapping aspects of workflow are subsequently identified as also being able to benefit from digitalisation. But the closure of many life sciences facilities during the UK’s Coronavirus lockdown saw the rate of adoption of digital practices snowball, as online systems suddenly became key to enabling scientists to continue working from home.
Image analysis continues through lockdown
"It’s important not to underestimate the value of pathologists’ experience, however; they are still very much a part of our everyday work. And for some studies, it can simply make more sense to consult a pathologist"
Lockdowns may have put the brakes on progress in wet labs, but image analysis work was able to continue largely unaffected – it was one of the unplanned benefits of going digital; that teams could access resources and data on file sharing systems and quickly pick up work where they left off in the lab.
There is so much more to digital image analysis than being able to view slides on a computer system and put a number to how much staining there is in a sample. Continued and impressive innovations in technology are to thank for both incremental and significant developments to the practice - and clients are increasingly recognising the benefits.
It’s important not to underestimate the value of pathologists’ experience, however; they are still very much a part of our everyday work. And for some studies, it can simply make more sense to consult a pathologist. Having both routes of analysis provides good balance, but it is important that the life sciences industry as a whole reaps the rewards that digital image analysis has to offer.
Strength in objectivity and efficiency
One of the greatest strengths of an optimised image analysis algorithm is its objectivity. Looking down a microscope and trying to consistently interpret qualitative visual data into quantitative results is a subjective endeavour. But with the use of an algorithm that has been carefully optimised for that unique sample set, the qualitative to quantitative translation is the same every time. We can also look beyond the raw data - for example, using tools such as H-scoring to account for staining intensity - so that, where appropriate, a low distribution of staining may ultimately score the same or higher than a sample with a higher distribution once intensity has been accounted for.
Yet, objectivity is not an image analysis algorithm’s only strength – its efficiency is another great benefit. The algorithms need only be set up once before they can be applied to large batches of samples and left to run, freeing up time for us to ensure we can deliver a fast turnaround for our clients. Digitalisation of our workflow has really helped us to this end, by allowing us to offer clients instant accessibility to data, whether it be images, annotations, numbers, or documents. Close collaboration with our clients has become an inevitability, which has seen the time spent packaging and shipping drastically reduced.
Digital image analysis offers transparency
"Clients are witnessing first-hand that the applications of image analysis are wide-ranging, and that its role is simplification: instead of seeing colours on a slide, they can see numbers"
Something that clients appreciate about digital image analysis, however, is the transparency that it offers. When samples are sent to a pathologist, clients do not have the opportunity to see how they achieve their findings. With image analysis, on the other hand, we can give clients a live demonstration and show them how and why we tune the algorithm the way we do through screen sharing. Through this collaboration we can also examine the possible approaches for annotation of regions of interest and help a client decide whether automated tissue classification or manual annotation using an interactive tablet is appropriate for them and their samples.
Closer client relationships
The opportunity to work more closely than ever with our clients exposes them to what is possible, whether that be analysis of individual objects in a sample (such as TMA cores or individual vacuoles), registration of serial sections or working with fluorescent images to study colocalisation of markers.
Clients are witnessing first-hand that the applications of image analysis are wide-ranging, and that its role is simplification: instead of seeing colours on a slide, they can see numbers. Ultimately, quantitative data is what the industry wants; it simplifies decisions and can speed up the approval process for clients wanting to move to clinical trials.
Whilst our clients do have the choice of whether their samples are sent to a pathologist or are run through image analysis software, we expect to see the demand for digital analysis increase going forward. We often find that many perceive image analysis as constrained and limited, but with the progress that has been made in the last decade the power of this technology, in conjunction with an ever more digitalising world, is now being unleashed.
Author: Andreas Theodosi is a Research Scientist at HistologiX
