AAD
Author: Willhemenia Loremipsum Special session: Lorem ipsum dolor sit amet Category: Lorem ipsum View all results

Please login or create an account. If you do not have access to this content, you will be shown a 30 second preview and licensing options.

  • Presentation

Digital Pathology and AI in Dermatopathology: Adoption, Implementation, and Future Directions

Description

The speaker discusses the rapid rise of digital pathology and AI in dermatopathology, framing AI adoption as both exciting and inevitable given exponential growth. They review the relatively recent clinical adoption of whole slide imaging and AI-assisted diagnostic tools, noting key barriers such as equipment cost, IT support, storage, and workflow changes. The talk explains essential infrastructure and terminology, including slide scanner focus methods, the need for scalable storage, and the roles of LIS, IMS, and DICOM standards to support interoperability and avoid vendor lock-in. The speaker also outlines validation expectations, such as CAP guidelines for concordance between digital and glass slides, and describes a phased rollout strategy that began with retrospective scanning for education and tumor boards, then moved to triage, consultation, and routine sign-out. Examples show benefits in telepathology, remote consultation, same-day decision-making, teaching, and faster multi-site collaboration. The talk then turns to AI applications, including quality control, prescreening Mohs cases, triaging difficult or easily missed diagnoses, recognizing case misassignments, and helping with tasks like mitotic figure detection, tumor-infiltrating lymphocyte quantification, PRAME standardization, report generation, and prognostication. The speaker highlights studies where AI matched or exceeded human performance in melanoma, SCC, Merkel cell carcinoma, and mutation prediction, emphasizing that the greatest future opportunity may be multimodal systems combining digital pathology with genomics for more precise, patient-specific care.

View more

Conclusions

  • Digital pathology is moving from early adoption toward routine clinical use, but successful implementation depends on investing in scanners, storage, IT support, and workflow infrastructure.
  • Vendor-neutral standards such as DICOM and scalable image management systems are important to avoid lock-in and support future AI integration.
  • Validation can be achieved safely when digital and glass slide diagnoses reach high concordance under established guidelines and washout periods.
  • Telepathology and remote image sharing can make consultation, second opinions, education, and global collaboration far more efficient than traditional slide shipping.
  • AI can meaningfully improve pathology workflows by helping with triage, case organization, quality control, and detection of missed or difficult cases.
  • For several diagnostic tasks, such as melanoma classification and Mohs section review, AI can match or even outperform expert pathologists in accuracy.
  • Computer-assisted scoring of features like mitotic figures, tumor-infiltrating lymphocytes, tumor area, and PRAME expression can improve precision and standardization compared with manual review.
  • AI-based quantification may reduce interobserver variability and bias, especially in tasks where human readers show poor reproducibility.
  • Digital and AI-derived prognostic features can predict recurrence, metastasis, and survival better than traditional staging alone in some skin cancers.
  • The strongest future direction is multimodal models that combine histology, clinical data, and genomics to produce more patient-specific prognostic and diagnostic support.
  • Current AI tools are best viewed as decision-support systems that augment pathologists rather than replace them, with humans remaining in the loop for interpretation and oversight.
  • Overall, the presentation argues that digital pathology is a foundation for broader AI-enabled pathology, and institutions that prepare now will be better positioned for the next wave of diagnostic innovation.
  • Tanaka K, Lami K, Odate T, Hori T, Sato T, Okoshi EN, Fukuoka J. Cytology Screening Using Z-Stack Digital Slides: A Validation Study. Diagn Cytopathol. 2025 Nov;53(11):568-575.#10.1002/dc.70015
  • Romero Lauro G et al. J Digit Imaging 2013, 26(4):668-77.
  • Kumar, Gupta, and Gupta. Journal of Digital Imaging, 2020.
  • Stathonikos, Veta, Huisman, and van Diest. Going fully digital: Perspective of a Dutch academic pathology lab. J Pathol Inform. 2013.#10.4103/2153-3539.114206
  • Evans, Brown, Bui, et al. College of American Pathologists guideline update on validation.
  • Pantanowitz L, Dickinson K, Evans AJ, Hassell LA, et al. Journal of Pathology Informatics, 2014.
  • Sullivan CM, DiMaio D, Lauer S, Pradhan D, Youngs J, Samson K, Georgesen CJ. The Need for Standardization of PRAME Immunohistochemistry in Melanocytic Neoplasms. Dermatopathology (Basel). 2025 Dec 31;13(1):5.#10.3390/dermatopathology13010005
  • Ahsan M, Jahangir F, Rathore S, Mumtaz M, Zaman S. Evaluation of whole-slide imaging for diagnosing frozen sections. Ann Diagn Pathol. 2025 Apr;75:152431.#10.1016/j.anndiagpath.2024.152431
  • Sturm B, Creytens D, Smits J, Ooms AHAG, Eijken E, Kurpershoek E, Küsters-Vandevelde HVN, Wauters C, Blox WAM, van der Laak JAWM. Computer-Aided Assessment of Melanocytic Lesions by Means of a Mitosis Algorithm. Diagnostics (Basel). 2022 Feb 8;12(2):436.#10.3390/diagnostics12020436
  • Rosenbaum BE, Schafer CN, Han SW, Osman I, Zhong H, Brinster N. Computer-assisted measurement of primary tumor area is prognostic of recurrence-free survival in stage IB melanoma patients. Mod Pathol. 2017 Oct;30(10):1402-1410.#10.1038/modpathol.2017.64
  • Enevoldsen et al. Digital quantification of PRAME. 2024.
  • Tran, Schmidle, Guo, et al. Nature Communications. 2025.
  • Vibha R. Rao, Vy Nguyen, Thuy L. Phung, and Shrey S. Sukhadia. Interpreting Deep Learning-Based Prediction of the BRAF V600E Mutation Using Diagnostic Whole Slide Images in Skin Cutaneous Melanoma. The American Journal of Pathology.#10.1016/j.ajpath.2025.11.004
  • Self supervised artificial intelligence predicts poor outcome from primary cutaneous squamous cell carcinoma at diagnosis. npj Digital Medicine.#10.1038/s41746-025-01496-3
  • A Novel Artificial Intelligence-Based Parameterization Approach of the Stromal Landscape in Merkel Cell Carcinoma: A Multi-Institutional Study. Laboratory Investigation.#10.1016/j.labinv.2024.102123
  • American Medical Association. CPT Category III Codes source.#10.32388/9yif04