Subscribe to Newsletter
Inside the Lab Digital and computational pathology, Technology and innovation, Profession

A Boost to Brainpower

At a Glance

  • Image-based medical specialties – for example, dermatopathology – are uniquely poised to benefit from advances in augmented Intelligence (AuI)
  • AuI tools must be carefully designed and validated to minimize disruption or unintended outcomes
  • The tools must also be thoughtfully integrated into existing workflows to achieve increased efficiency and ease laboratory burden
  • Only through human-centered design around the pathologist can AuI succeed

Augmented Intelligence (AuI) has the potential to transform the clinical practice of physicians – and specialties with significant image-based components, such as radiology, dermatology, ophthalmology, and pathology, are uniquely poised to benefit from advances in deep learning applications. The concept of AuI focuses on the role of artificial intelligence (AI) in assisting clinicians. The goal? Not to replace human intelligence, but to enhance it.

Recognizing this potential, the American Medical Association (AMA) published its first policy to guide engagement and growth in this field in June 2018 (1). To ensure safe, effective, and equitable use of (and access to) AuI, the AMA seeks to:

  • ensure that improved patient outcomes and provider satisfaction are priorities,
  • integrate practicing physicians into the development of AuI,
  • promote development of thoughtfully designed, high-quality, clinically validated AuI,
  • encourage education for all stakeholders to promote greater understanding of the promise and limitations of AuI, and
  • explore the legal implications of AuI.

Recently, the American Academy of Dermatology approved a position statement affirming that the key to realizing AuI’s promise is twofold: one, to ensure that the technology is collaboratively developed and designed for the benefit of our patients, physicians, and the healthcare system at large; and two, to minimize the risk of potentially disruptive effects and unintended consequences (2). In dermatopathology, AuI tools and systems have the potential to enhance workflows and efficiency, support the ability to spread expertise and collaboration to those with limited access, and improve the predictive and prognostic power of traditional pathology approaches – all essentially supporting the specialist’s capabilities.

Enhancing efficiency, workflows, and access

The movement toward whole-slide imaging and slide digitization forms the foundation that empowers AuI-based solutions. It can foster collaboration, communication, and access through dissemination of expertise. Expanding our diagnostic capabilities to medically underserved or resource-constrained areas can reduce traditional barriers and facilitate consultations. Additionally, AuI tools can potentially help identify similar slides in an archive or repository of previously diagnosed cases as a means of extending expertise. Currently, telepathology overcomes some obstacles through store-and-forward imaging, robotic telepathology, or slide scanning; however, these methods are often limited by local bandwidth, tissue processing, manpower, local expertise, and cost (3).

AuI has the potential to improve efficiency and reduce dermatopathologists’ workloads.

In addition to improving access to pathology services, AuI has the potential to improve efficiency and reduce dermatopathologists’ workloads. An AuI virtual assistant could sort work based on complexity, highlight and identify features of interest on a slide, and allow pathologists to spend more time on cases requiring increased attention while simultaneously expediting assessment of more routine cases (4). Through deep learning approaches, AuI can also evaluate tumors based on diversity of nuclear features to help risk-stratify patients (5). Emerging efforts have demonstrated potential in evaluating deep learning methodologies on whole-slide images to augment decision-making (6)(7).

Future capabilities with more robust data and computational pathology approaches may enable AuI-based platforms to perform quantitative image analysis, identify features “hidden” to human perception, and make them available for consideration (8)(9). One practical result of these capabilities may be the reduction of inter- and intra-pathologist variation.

Uncovering novel relationships

The foundation and gold standard for dermatopathology diagnosis has always been the pathologist’s evaluation of tissue under the microscope. However, AuI’s ability to integrate and analyze complex streams of multimodal data (combining tissue histology, molecular outputs from diagnostics and next-generation sequencing, clinical images, and electronic health record data) may uncover clinical relationships to enhance predictive and prognostic power (10). For instance, molecular profiling techniques have proliferated in melanoma, and the incorporation of validated and reliable tumor-level genomic information with clinical and outcomes data may hold the potential to predict future disease behavior and treatment response (11).

Future approaches may even allow more fine-grained insight into protein–genomic associations and tumoral and stromal heterogeneity by tying molecular classification to immunohistochemical features (12). A dermatopathologist’s augmented ability to analyze, synthesize, and interpret additional streams of information may support timely, actionable, and precise diagnoses for patients.

AuI’s ability to impact dermatopathology will depend as much on the thoughtful development and integration of such tools and systems as on the underlying technology itself.

Although a future in which dermatopathology transforms from a tissue-centered clinical science into an informatics science incorporating non-tissue data streams may look different, the core of the discipline will not change. It will still be our charge to support clinicians’ ability to make diagnoses and choose effective treatments for patients. However, AuI’s ability to impact dermatopathology will depend as much on the thoughtful development and integration of such tools and systems as on the underlying technology itself. At the moment, dermatopathology AuI diagnostics have been subject to only the early stages of research, and they must be rigorously tested and validated in a wide variety of diseases before we can move forward. However, the promise they show is worth the effort.

Overall, pathologists have generally positive attitudes towards AuI, with nearly 75 percent of respondents in a recent large global study reporting interest or excitement in using it as a diagnostic tool (13).However, even within this optimistic cohort, there was concern for potential job displacement and replacement. Therefore, it is essential for human-centered design, with the pathologist at the core of AuI-assisted diagnostics, to lead the way to better care and outcomes and perhaps greater fulfillment and satisfaction with pathology practice.

Receive content, products, events as well as relevant industry updates from The Pathologist and its sponsors.
Stay up to date with our other newsletters and sponsors information, tailored specifically to the fields you are interested in

When you click “Subscribe” we will email you a link, which you must click to verify the email address above and activate your subscription. If you do not receive this email, please contact us at [email protected].
If you wish to unsubscribe, you can update your preferences at any point.

  1. American Medical Association, “Augmented intelligence in health care H-480.90” (2018). Available at: Accessed September 25, 2019.
  2. C Kovarik et al., “Commentary: Position statement on augmented intelligence (AuI)”, J Am Acad Dermatol, 81, 998 (2019). PMID: 31247221.
  3. RG Micheletti et al., “Robotic teledermatopathology from an African dermatology clinic”, J Am Acad Dermatol, 70, 952 (2014). PMID: 24742843.
  4. Y Zheng et al., “Histopathological whole slide image analysis using context-based CBIR”, IEEE Trans Med Imaging, 37, 1641 (2018). PMID: 29969415.
  5. C Lu et al., “Nuclear shape and orientation features from H&E images predict survival in early-stage estrogen receptor-positive breast cancers”, Lab Invest, 98, 1438 (2018). PMID: 29959421.
  6. BE Bejnordi et al., “Context-aware stacked convolutional neural networks for classification of breast carcinomas in whole-slide histopathology images”, J Med Imaging, 4, 044504 (2017). PMID: 29285517.
  7. SN Hart et al., “Classification of melanocytic lesions in selected and whole-slide images via convolutional neural networks”, J Pathol Inform, 10, 5 (2019). PMID: 30972224.
  8. J van der Laak et al., “The Promise of Computational Pathology: Part I”, The Pathologist (2018). Available at:
  9. M Salto-Tellez et al., “The Promise of Computational Pathology: Part II”, The Pathologist (2018). Available at:
  10. U Djuric et al., “Precision histology: how deep learning is poised to revitalize histomorphology for personalized cancer care”, NPJ Precis Oncol, 1, 22 (2017). PMID: 29872706.
  11. AL Ji et al., “Molecular profiling in cutaneous melanoma”, J Natl Compr Canc Netw, 14, 475 (2016). PMID: 27059194.
  12. MKK Niazi et al., “Digital pathology and artificial intelligence”, Lancet Oncol, 20, e253 (2019). PMID: 31044723.
  13. S Sarwar et al., “Physician perspectives on integration of artificial intelligence into diagnostic pathology”, NPJ Digit Med, 2, 28 (2019). PMID: 31304375.
About the Authors
Justin Ko

Clinical Associate Professor, Director, and Chief of Medical Dermatology at Stanford University, Stanford, USA.

Carrie Kovarik

Associate Professor of Medicine and Associate Professor of Dermatology at the Hospital of the University of Pennsylvania, Philadelphia, USA.

Register to The Pathologist

Register to access our FREE online portfolio, request the magazine in print and manage your preferences.

You will benefit from:
  • Unlimited access to ALL articles
  • News, interviews & opinions from leading industry experts
  • Receive print (and PDF) copies of The Pathologist magazine