Here is one possible future: Some day, whole slide imagers will be ubiquitous in pathology labs throughout the U.S. They will scan our slides quickly and in multiple contiguous focal planes, thus retaining all the morphologic data that exists in “analog” form. The quality of the whole slide imaging will be so good that pathologists will have no need to review the original slides. Our workflow will be entirely digital and physical proximity of the histology lab and the pathologist will no longer matter. Artificial intelligence algorithms will interpret the scanned slides and will flag the urgent diagnoses. At first computer assisted diagnosis will make pathologists’ jobs easier and we will be able to do more work faster and extract more information from morphology than we could before. However, eventually pathologists will have less to do as the computers become smarter and faster. At some point slide interpretation will no longer be done by human beings, since AI can do everything they do faster and are less prone to error. Furthermore, they will learn to identify complex patterns with diagnostic and prognostic relevance which are beyond human perception. Patterns such as chromatin distribution, reconstructed 3D tissue architecture, and quantification of nuclear contour irregularity.
Here is another possible future: In vivo microscopy will improve to the point where it matches the resolution of histology, thus precluding the need for ex vivo microscopic interpretation. In conjunction with improvements in metabolic imaging and exosomal DNA analysis, histology will have nothing to contribute beyond the information which can be obtained in vivo. Whole slide imaging will become a redundant concept since there will no longer be slides to scan. Pathologists, radiologists, and clinicians who do endoscopy will compete for the professional right to interpret these in vivo images which will render most biopsies unnecessary.
Will one of these scenarios prevail? Or will the future be different?
[Photo is of Nexi, a robot developed at MIT]
2 thoughts on “2 possible futures for pathology”
Just this week I tried the whole slide image scanner for the first time even though I was introduced and given instructions on how to use it in July. I loved the ease of use and how I could put 3 autopsy cases and a surgical case in the scanner overnight and could come back the next day to snap jgp photos for presentations. One true weakness I found in whole slide image scanning was that mitotic figures were not readily visible. They were easily seen by light microscopy and I ended up having to take extra photos the ‘old fashioned’ way with the camera associated with the microscope to prove to people this was an anaplastic ependymoma.
I ended my week attending a grand rounds given by our new chair who comes from MGH. He discussed the past present and future of microscopy and introduced me to the work that is being done with in vivo microscopy.
I agree that maybe one day your predictions may become reality but I believe that we won’t see this world for another 50-100 years – well beyond our career. You have to prove that whole slide imaging and in vivo microscopy are evidentially better than histology. These two technologies are still in their infancy. Once you’ve finally proven their efficacy, you need to train generations of pathologists to interpret these modalities and move away from histopathology.
I think we will see these modalities become ancillary in our practice. But I don’t think we will see histology become obsolete in our lifetime. And one more thing – pathologists better start getting more assertive because there is no way radiologists can win at interpreting in vivo microscopy. That is our turf.
LikeLiked by 1 person
Turf issues should come in second after patient care quality. We are building magnetic resonance imaging systems with a current resolution of 20 microns, comparable to optical microscopy. These systems will be intended for breast and prostate in vivo imaging, where the entire glands will be sampled in less than five minutes without intravenous contrast. At this point, the contrast is not as good as can be achieved with H&E preps, but there are many MRI pulse sequences that are yet to be optimized. Imaging the entire gland would be too much for any human practitioner to handle by themselves, whether they were trained in radiology or pathology, so computer-aided diagnosis will be a critical step. I would be glad to correspond with individuals with interest in this field.