Radiology versus pathology: Not more quantitative, just more digital

Steve Potts, of Flagship Biosciences, a digital pathology services company, submitted the following blog comparing the perceptions of radiology with pathology.

At a molecular imaging meeting earlier this year, Flagship presented pharma regulatory compliance for digital pathology. Our talk was the only one on pathology. The others were all various presentations on the different modalities of radiology imaging applied to clinical trials -- X-ray, MRIs, microCT scans, PET scans, CAT scans, dog scans, and so on 🙂

A radiologist told me several years ago, "Pathology is where you see the details of real biology -- the highest resolution look at the local disease state." And yet why does the average person on the street know what an MRI is, what a CAT scan, what an X-ray is, and yet has never heard of an IHC, or much less an H&E? Not only much greater awareness among the general public, but there is also a perception that radiology is a quantitative science. It is not.

How many pattern recognition programs exist in the daily radiology workflow? How much quantitation occurs by the computer in radiology? Nothing wrong with this, but most clinical diagnoses in radiology are a trained professional making qualitative judgments on a computer screen. The three-dimensional nature of tissue heterogeneity makes quantitative evaluation in radiology as difficult as it is in pathology. Regarding pharmaceutical companion diagnostics development, standard radiology imaging approachs in clinical trials involve agreement on a study design with manual tiered scoring, similar in many ways to qualitative pathology approaches. Once could argue radiology is less quantitative than pathology if one considers the progress made in breast cancer HER2/ER/PR quantitative IHC clearances. There has been a noticable decrease in the volume of CAD (computer-aided diagnosis) messaging at this year's RSNA tradeshow, the number one show for the medical radiology technology community.

Radiology isn't more quantitative, it is just more digital.

Imagine two presenters, one using a blackboard and the other powerpoint. Which one seems more modern, more professional? Or two people writing a manuscript -- which appears better, the neat word document or the pages of legal pad scribbles? The irony is that no cute powerpoint presentation beats that one professor you had (maybe the teacher in high school or college who first got you interested in science -- it only takes one) who used every portion of the blackboard, and left you hanging on every word.

Similar in manuscripts. At Flagship we have been digging back through a century of histopathology literature looking for special stains to use in our new multiplexing IHC technology to help the computer identify subanatomic organelles. The quality of the manuscripts of some of the pioneering work in histology stains from decades ago is remarkable. Careful, detailed pen and ink drawings of which stains work best in which substructures. Thoughtful, comprehensive staining recipes from the days when quality beat quantity in publications. Biology rationales behind non-specific staining artifacts in other substructures, from the era of pen and ink drawings. But while digital does not mean better, it certainly appears more quantitative to the outsider.

Radiology isn't waiting for pathology to go digital. From their perspective, it is just one more digital mode to integrate with better resolution than they had before.

The topic of the RSNA meeting this year is instructive -- personalized medicine. The next time I go to a "molecular imaging" conference, I want people to think pathology first, and radiology second.

Radiology is not more quantitative than pathology, it is just more digital.