2nd Generation Brightfield versus 1st Generation Fluorescence Scanners

This post is submitted by Steve Potts, CEO of Flagship Biosciences, a companion diagnostics company. 

“A truly great book should be read in youth, again in maturity and once more in old age, as a fine building should be seen by morning light, at noon and by moonlight.” — Robertson Davies

These are exciting times to be consumers of digital pathology hardware. We are at the end of the 2nd generation brightfield scanners (1st Generation were the automated microscopes of CAS / VIAS / ACIS, and 2nd generation are the Nanozoomers / ScanScopes / Mirax scanners with great 20x whole slide scanning performance). Our in-house definition of a 3rd Generation brightfield scanner is the speed, quality, and error rates we expect from 2nd Generation scanners at 20x, but performed at 40x. This represents a four-fold improvement (20x to 40x is four times longer) in the last three or four years –Moore’s law marches on in brightfield scanning. To evaluate a 3rd Generation scanner, take a set of five slides to USCAP or other pathology show, and with a stopwatch evaluate the time, quality and error rate of the scans at 40x among the various new products. We are just beginning 3rd generation devices, and there are a lot of new and existing companies with innovative offerings in 2011. USCAP and CAP will likely demonstrate some fabulous choices for scanner consumers!

Now let’s look at fluorescence. At January’s Applied Immunohistochemistry and Molecular Pathology Retreat in Miami (IMHO the best and most serious conference on practical IHC year after year) there was an interesting comment, “Fluorescence microscopy is far ahead of brightfield microscopy in quantitation, but fluorescence scanning is far behind brightfield scanning”. Fluorescence scanning is simply a LOT harder than brightfield scanning. The four major fluorescent scanners on the market (3DHistech Pannoramic, Olympus VS110, Hamamatsu Nanozoomer, and the Aperio ScanScope FL) all have different advantages, and this blog could go on for pages discussing specific design tradeoffs we've evaluated. We run both brightfield and fluorescent scanning in high volume in our digital pathology lab, and here are some differences between the two scanning modalities that users should be aware of, regardless of which system is chosen:

Scanning Time:

• BRIGHTFIELD: For 15mm x 15 mm tissue sections, expect brightfield scanning times of 3 minutes at 20x, and 8-10 minutes at 40x.
• FLUORESCENCE: For fluorescence (our normal operation is three or four colors), expect 2-3 slides per hour at 20x. Forget about 40x fluorescence on whole slide images. 

QA and failure rate:

• BRIGHTFIELD: Expect to spend at least 1 minute per slide for a technician to QA a brightfield slide manually, this is required for good brightfield scanning operations, and far too many people don’t do this. Failure rates of 5% for 20x and 20% for 40x are normal, but these numbers are largely dependent on histology and slide preparation.
• FLUORESCENCE: Expect rescan rates of 20-40% at 20x, this is very much dependent on the given sample and the skill of the operator and how well tuned the fluorescent scanner is. The errors will come primarily from the tissue sample more than from the fluorescent scanning. Things like debris on the cover slip, Rocky Mountain-like tissue samples, dye over stains, are not as problematic for a brightfield scan as a fluorescent scan. 

Technician Skill:

• BRIGHTFIELD: Brightfield scanning requires a very methodical person, who is consistent and pays attention to detail in QA of each slide. There is very little biology discussion required, and scanning settings are not changed that much between studies. In other words, the scanning is generally the same regardless of histology, for most projects (TMA scanning and Z-stack do require specialized settings).
• FLUORESCENCE:  Every study requires specific knowledge of the biology that then must be used in deciding on scanning settings. One has to know the amount of autofluorescence, the differences between autofluorescence and signal in the tissue, the goal of the study. If quantitation of protein expression is required, a number of steps need to be in place standardizing the scan times and comparisons to controls, and control preparation. In short, the scanning settings are entirely dependent on the biology of each study. In our lab, a skilled Master’ level biologist runs the scanning, and we get the best results when he knows (before the study) the biological endpoint the client is trying to achieve. 

Fluorescence whole slide scanning technology is improving, and offers powerful multiplexing applications, but everything mentioned previously helps explain why there are 16 FDA protein expression clearances in brightfield digital imaging, and none in fluorescence. The pathologist's adoption of fluorescent scanning is another factor.