Uncertainty aware models for the diagnosis of primary and secondary skin tumours

Introduction

Project PID2022-140189OB-C22 funded by

The research project Uncertainty aware models for the diagnosis of primary and secondary skin tumours (Modelos conscientes de la incertidumbre para el diagnóstico de tumores cutáneos primarios y secundarios) is part of the project Bridging the gap between computational pathology and clinical practice: an AI system for the diagnoSiS of prImary and Secondary skin Tumours (ASSIST) (Acercando la patología computacional a la práctica clínica: un sistema de IA para el diagnóstico de tumores cutáneos primarios y secundarios)m funded by MCIN/AEI/10.13039/501100011033 and by ``ERDF A way of making Europe", by the European Union from 2023 to 2026.

The Uncertainty aware models for the diagnosis of primary and secondary skin tumours research team consists of doctors from the Visual Information Processing Group (VIP) at Universidad de Granada and experts pathologist from the Anatomic Pathology Unit from Hospital Universitario Clínico San Cecilio (HUSC) and Hospital Universitario Virgen de las Nieves (HUVN), undergraduates with extensive experience in the project theme, a doctor from the Northwestern University (Evanston, Illinois, USA), a doctor from the Institute for Artificial Intelligence in Medicine - Center for Computational Imaging and Signal Analytics in Medicine (Chicago, Illinois, USA), a doctor from the University of Cambridge (UK).

This page will provide information on the project results and publications.

---

Summary

The workload of Pathology departments is exponentially growing due to the increasing number of biopsies, cancer cases and screening programs. Despite the expected growth in biopsy analysis in the coming years, the approved applications based on artificial intelligence (AI) intended to speed up this process are almost non-existent. The FDA authorised 521 AI- or machine-learning-enabled medical devices in the period 1995 to 2022 but only 4 of those applications are intended for use by pathologists, thus the market for the introduction of AI based tools for pathology is still very much in its infancy.

ASSIST constitutes a coordinated effort between research groups at the Polytechnic University of Valencia, the University of Granada, the Hospital Clínico Universitario of Valencia, and the Hospital Universitario San Cecilio of Granada aimed at bridging the translation gap between research and clinical practice by developing a robust, fair, and explainable computer aid diagnosis system that mimics the whole process carried out by pathologists when determining the absence/presence of cancer and its type, which goes beyond the diagnosis of primary tumours. ASSIST will be able to recognise rare entities which probably correspond to secondary tumours, retrieve similar cases from an intelligent atlas of histopathological images, and provide pathologists with a classification of the tumour under study. Although the approach proposed in ASSIST can be utilised on other tumours, ASSIST will focus on skin tumour detection, starting from the results of a previous coordinated project of the applicant team.

As a whole, ASSIST will jointly investigate: (1) how to efficiently standardise histological whole slide images, (2) how to extract their most relevant regions of interest (ROIs), (3) how to obtain a robust embedding to characterize those ROIs, (4) how to detect out of distribution (OoD) cases, (5) how to find the best model to diagnose known cases through crowdsourced multiple instance learning and update this model using online learning strategies; and (6) how to retrieve the most similar samples from an extensive histopathological image atlas for the diagnosis of OoD cases. Subproject 2 will lead the tasks related to (1), (4) and (5) and collaborate with Subproject 1 in the creation of the ASSIST database as well as the validation of fair and explainable models. The areas of research of ASSIST will help to address challenges in the field of artificial intelligence, one of the key areas of State policy, and in precision health, a national R&D&I strategic line in the thematic group Health with which the project is aligned.

ASSIST is expected to have social and economic impact at different levels. The hospitals in the proposal will benefit from an AI solution that will help to provide accurate and faster diagnosis and therefore contribute to the reduction of pathologists' workload. The system could be installed or used remotely by other Spanish hospitals in the future. This would open the door to a new form of collaboration between pathology departments at national level. At worldwide level, ASSIST will allow Spain to be a pioneer in the use of computational techniques applied to the diagnosis of primary and secondary skin tumours.