gmail
com
Development and Validation of a Collaborative Digital Pathology Platform Integrating Human–AI Interaction
Background and Rationale
Digital Pathology and Artificial Intelligence (AI) are rapidly transforming diagnostic practice, education, and research. Whole Slide Images (WSIs) enable image sharing, collaborative annotation, development of educational resources, and computational analysis. However, one of the major limitations to the development of AI in pathology remains the availability of large, high-quality, expertly annotated datasets, as manual annotation is time-consuming and requires substantial pathology expertise.
This PhD project aims to develop and validate a collaborative Digital Pathology platform designed not only for WSI storage and sharing, but also for annotation, structured metadata management, education, and integration with AI algorithms.
A central innovative component will be the development of the “Human–AI Hyperloop”, an iterative system in which annotations and corrections performed by pathologists contribute to data and algorithm refinement, while AI-generated outputs assist pathologists in subsequent annotation steps. The resulting workflow consists of repeated cycles of:
Pathologist annotation → AI analysis → expert review and correction → dataset refinement → AI retraining.
The central hypothesis is that this iterative human–AI interaction can progressively improve dataset quality and AI performance while reducing the workload required for manual annotation.
Objectives and Methodology
The first objective will be the implementation and optimization of the Digital Pathology platform. Hardware and software infrastructure will support WSI storage and visualization, user management, collaborative annotation, comments, structured metadata, and integration of AI tools. Functional and stress tests will assess platform stability, scalability, simultaneous access, image management, and annotation performance.
The second and main experimental objective will be the development and validation of the Human–AI Hyperloop. Successive cycles of human annotation, AI analysis, expert correction, and algorithm retraining will be performed. Quantitative endpoints will include annotation time, number of corrections, interobserver and human–AI agreement, annotation quality, and AI performance across successive iterations.
The third objective will be the creation of structured histopathological datasets. Cases will be selected according to predefined criteria and accompanied by appropriate annotations and clinicopathological metadata. In parallel, a Digital Pathology teaching set will be developed, including diagnostically and educationally relevant cases with WSIs, annotations, diagnoses, differential diagnoses, immunohistochemical and molecular information when available, and educational comments.
A further component will address data governance, ethics, and GDPR compliance, including anonymization/pseudonymization, access control, data retention, image sharing, secondary use of data, and the use of histopathological images for AI development and validation.
Originality and Expected Results
The originality of the project lies in moving beyond the concept of a Digital Pathology platform as a simple image repository toward a dynamic ecosystem connecting pathologists, data, education, and AI.
The Human–AI Hyperloop represents the main experimental innovation. Rather than providing a static ground truth, pathologists will continuously interact with AI systems, progressively refining both datasets and algorithms.
The project is expected to generate a fully operational collaborative Digital Pathology platform, expertly annotated histopathological datasets, a structured teaching set, a validated Human–AI Hyperloop framework, and a GDPR-compliant data governance model. The resulting methodology may subsequently be extended to multicenter Computational Pathology projects.
Three-Year Research Plan
Year 1: platform implementation and optimization; functional and stress testing; definition of dataset criteria; development of the GDPR and data governance framework.
Year 2: creation of histopathological datasets and teaching sets; implementation of collaborative annotation workflows; integration of AI algorithms; development and initial testing of the Human–AI Hyperloop.
Year 3: systematic validation of the Human–AI Hyperloop; quantitative analysis of successive human–AI interaction cycles; dataset expansion; evaluation of multicenter scalability; scientific publications and dissemination of results.
Scientific Impact
The project addresses key challenges in Digital and Computational Pathology, particularly the creation of high-quality annotated datasets and the efficient integration of human expertise with AI. The platform may provide a permanent infrastructure for research, education, AI development and validation, and future national and international collaborations, establishing a model in which pathologists remain central to the continuous refinement of both data and Artificial Intelligence systems.
Curriculum Vitae - Luca Cima
(msword, it, 304 KB, 17/08/26)
I am a Staff Surgical Pathologist at the University and Hospital (AOUI) of Verona and a PhD student in "Inflammation, Immunity and Cancer" at the University of Verona. My clinical practice and research focus on breast, genitourinary and thoracic pathology, with a strong interest in Digital and Computational Pathology. I am particularly interested in the development and validation of AI-based tools, digital pathology implementation, biomarker assessment, and precision oncology. I have authored over 61 peer-reviewed publications and regularly serve as a reviewer for several international scientific journals. I am actively involved in international and multicenter research collaborations across Europe, North America and Asia, particularly in digital pathology, AI and validation studies. I have longstanding experience within the European Society for Digital and Integrative Pathology (ESDIP), including communication, education and congress activities. I am also a member of the Editorial Team of Pathologica and serve as its Social Media Editor. A further important area of my activity is pathology education and scientific communication, with invited lectures at national and international meetings.
Modules running in the period selected: 0.
Click on the module to see the timetable and course details.
Di seguito sono elencati gli eventi e gli insegnamenti di Terza Missione collegati al docente:
© 2026 | Verona University
******** CSS e script comuni siti DOL - frase 9957 ********
Test
