Journal:DigiPatICS: Digital pathology transformation of the Catalan Health Institute network of eight hospitals - Planning, implementation, and preliminary results

Full article title	DigiPatICS: Digital pathology transformation of the Catalan Health Institute network of eight hospitals - Planning, implementation, and preliminary results
Journal	Diagnostics
Author(s)	Temprana-Salvado, Jordi; López-García, Pablo; Vives, Josep C.; de Haro, Lluís; Ballesta, Eudald; Abusleme, Matias R.; Arrufat, Miquel; Marques, Ferran; Casas, Josep R.; Gallego, Carlos; Pons, Laura; Mate, José L.; Fernández, Pedro L.; López-Bonet, Eugeni; Bosch, Ramon; Martínez, Salomé; Ramón y Cajal, Santiago; Matias-Guiu, Xavier
Author affiliation(s)	Institut Català de la Salut, Centre de Telecomunicacions i Tecnologies de la Informació, Vall d’Hebron University Hospital, TIC Salut Social, Technical University of Catalonia, Germans Trias i Pujol University Hospital, Doctor Josep Trueta Hospital of Girona, Verge de la Cinta Hospital of Tortosa, Joan XXIII University Hospital of Tarragona, Arnau de Vilanova University Hospital, Bellvitge University Hospital
Primary contact	Email: jtemprana at vhebron dot net
Editors	Eloy, Catarina
Year published	2022
Volume and issue	12(4)
Article #	852
DOI	10.3390/diagnostics12040852
ISSN	2075-4418
Distribution license	Creative Commons Attribution 4.0 International
Website	https://www.mdpi.com/2075-4418/12/4/852/html
Download	https://www.mdpi.com/2075-4418/12/4/852/pdf (PDF)

Abstract

Complete digital pathology transformation for primary histopathological diagnosis is a challenging yet rewarding endeavor. Its advantages are clear with more efficient workflows, but there are many technical and functional difficulties to be faced. The Catalan Health Institute (Institut Català de la Salut or ICS) has started its DigiPatICS project, aiming to deploy digital pathology in an integrative, holistic, and comprehensive way within a network of eight hospitals, over 168 pathologists, and over one million slides each year. We describe the bidding process and the careful planning that was required, followed by swift implementation in stages. The purpose of the DigiPatICS project is to increase patient safety and quality of care, improving diagnosis and the efficiency of processes in the pathological anatomy departments of the ICS through process improvement, digital pathology, and artificial intelligence (AI) tools.

Keywords: digital pathology, computational pathology, artificial intelligence, deep learning, implementation, workflow, primary diagnosis, LIS, telepathology, network

Introduction

The Catalan Health Institute (Institut Català de la Salut or ICS) is the largest provider for the Catalan Health Service, the insurer of universal health coverage in Catalonia. It is the company with the most employees in Catalonia and the largest public company in Spain, with almost 39,000 professionals who provide services to almost six million people throughout the territory. [1] The ICS manages 283 primary care teams, three large high-tech tertiary hospitals (Vall d’Hebron, Bellvitge, and Germans Trias), four regional reference hospitals (Arnau de Vilanova in Lleida, Joan XXIII in Tarragona, Josep Trueta in Girona, and Verge de la Cinta in Tortosa), and a regional hospital (Viladecans) (Figure 1). The ICS accounts for seven percent of the Catalonian government budget, accounting for over 40 million primary visits and over 100,000 surgical interventions yearly.

Figure 1. Map of Catalonia and the eight ICS hospitals.

Some laboratories are beginning to successfully deploy digital pathology solutions for routine diagnosis, which we believe will be a growing trend in the next few years. [2,3,4,5,6,7,8,9,10] In Catalonia, the DigiPatICS project plans to accomplish a complete digital pathology transformation of primary histopathological diagnosis for over 168 pathologists. Many groups have reported equivalency between digital pathology and conventional pathology. [11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27]

With the DigiPatICS project, we aim to increase patient safety and quality of care, improving diagnosis and the efficiency of processes in anatomical pathology departments of the ICS through digital pathology and artificial intelligence (AI) tools. [28] With digital pathology, we aim for a network of eight hospitals to work as one in terms of case sharing and teaching, putting all our patients on equal footing. This transversal digital transformation will have an impact on the care of patients treated by all medical and surgical specialists.

First, we created a network between ICS centers. This helped us increase the reproducibility and quality of diagnoses, as well as offered greater equity and safety to patients. In turn, this network approach facilitated remote diagnosis, case sharing, sub-specialization, and teaching for pathologists. In addition, we aimed for better working conditions, impacting the optimization of workflows, productivity, and, finally, turnaround times. We also intended to improve ergonomics and postural health, as well as to facilitate morphometric tools and the quantification of diagnostic and prognostic biomarkers to involve the optimization of time and a higher quality in diagnosis.

From a more technical point of view, the aim was to achieve a central digital repository of images on the network, thereby reducing the burden of slide file management and integrating medical imaging with SIMDCAT, a digital medical imaging system used in Catalonia. It was also intended as a subproject to establish bidirectional communications with other locations, such as operating rooms.

The project included the development of AI tools with machine learning (ML) and deep learning, taking advantage of the availability of whole slide images (WSIs) that were obtained after digitization. The objectives were to recognize tissue patterns, select tumor areas, and quantify them, among others. Hopefully, this use of AI tools will contribute to improving the quality of diagnosis and the efficiency of processes.

Materials and methods

DigiPatICS was created as a European Regional Development Fund (ERDF) project, with European funds for the optimization of anatomopathological diagnosis in a network of public ICS hospitals in Catalonia through digitalization and AI tools.

Subsequently, a market consultation was carried out, and, finally, it was tendered with the file code CSE/CC00/1101202869/20/AMUP [29].

Planning, scope, and tender process

A definition of needs was first carried out. We firmly believe meticulous planning is essential, taking into account all functional and technological requisites. Failing to detail such requirements can end in the failure of a digitization project, resulting in expensive scanners installed in pathology laboratories that are barely used. It is also important to highlight that going digital is not about acquiring pathology scanners; we focused our project on the purchase of a service with a shared risk with the bidder to achieve our objectives. Since this transformation was meant to be a one-way street with no possibility of going back to microscopes, all planning needed to include sufficient contingencies to avoid any kind of downtime for pathologists, as well as to ensure they benefit from the potential added value.

The purpose of the DigiPatICS project was to increase patient safety and quality of care, improving diagnosis and the efficiency of processes in pathological anatomy departments of the ICS using digital pathology and AI tools.

In defining the scope, several questions arose:

• Do we want to save the whole slide images (WSIs) forever? Who will store them?
• Is our laboratory information system (LIS) ready?
• Do we want to (or have to) address pre-analytics?
• Do we want to address dark-field microscopy (e.g., direct fluorescence, FISH)?
• Do we want to digitize the macroscopic images?
• Do we need to update the hospital network?
• Do we need to update our pathologists’ workstations?
• Do we want to share cases with the outside world?
• Is teaching important?
• Do we want AI algorithms?
• Do we want to do telepathology?
• What do we do with cytology?
• Do we have money for everything?

Those concerns and how they were resolved will be addressed shortly, but we can already answer some of these. We did want to store all the WSIs forever and to use that repository to train our own AI algorithms, which is clearly one of the great advantages of such a transformation. We also believed that this project must be an integral transformation, including routine histopathology, fluorescence, research, and macroscopic images. Tools for teaching and teleconsultation should be included, which meant having the option to share images outside our hospitals’ secured LAN. We realized that our LIS, preanalytics, network, and pathologist workstations all needed substantial upgrades to be able to undertake such a transformation. [9]

What about cytology? Digitizing cytology, even if feasible [30,31,32,33], has some particular concerns, including scanning times being much longer than in histology (due to the need for more resolution, larger scan area, zero tolerance of out-of-focus areas, and Z-stacking). That means needing to install more scanners to be able to take on the same activity, and it impacts storage needs. Dark-field scanning (FISH) has similar issues, but there was a significant difference in the volume of slides to scan. Cytology involves a large number of samples to digitize in our hospitals (over 400,000 each year), which is not affordable currently in this project due to budgetary constraints. (In other words, the activity addressed in our project included all bright-field, routine histopathology, histochemistry, immunohistochemistry, direct immunofluorescence, ISH, and FISH slides. Cytology was scanned on an as-needed basis.)

In Table 1, we summarize the total number of slides generated during 2019 at our eight hospitals, broken down by type. To that number of over one million slides, an expected growth in activity of 10% to 15% must be added each year. In addition, some resources were reserved for research and non-strictly routine samples and were not accounted for in these numbers.

Table 1. Number of slides in 2019. Slide type # of that slide type Routine histopathology 814,573 Immunohistochemistry 186,453 Histochemistry 64,209 Direct immunofluorescence 12,392 FISH 2,695 CISH 1,983 Total 1,082,305

Regarding the amount of personnel involved, DigiPatICS provided service to 107 pathologists, seven biologists, 40 residents, and 14 observers, adding up to a total of 168 professionals working with digital diagnosis.

In the tender process, all relevant aspects were taken into account for bidder evaluation, as shown in the following list:

▪ Automatic evaluation criteria (51 points)

○ Economic valuation (40 points)

▪ Evaluation of the financial offer (30 points)

▪ Evaluation of the maintenance offer (10 points)

○ Automatic technical evaluation (11 points)

▪ Quality management system. Certification of processes and algorithms (3 points)

▪ Process consulting (1 point)

▪ Image management platform adaptations (3 points)

▪ Storage for research slides (1 point)

▪ Short-term “hot” storage (3 points)

▪ Criteria subject to judgment value (49 points)

○ Scanners: deployment and image quality (17 points)

○ Diagnostic viewer (8 points)

○Image management platform (4 points)

○ Training module (2 points)

○ Built-in tools and algorithms (3 points)

○ Architecture and monitoring (1 point)

○ Definitive storage in SIMDCAT (4 points)

○ Integration of case information in a unified model (2 points)

○ Server infrastructure requirements and DPCs. Coherence, management model, virtualization (2 points)

○ Workstations (2 points)

○ Artificial intelligence (3 points)

○ Implementation and additional improvements (1 point)

Notes

This presentation is faithful to the original, with only a few minor changes to presentation, though grammar and word usage was substantially updated for improved readability. In some cases important information was missing from the references, and that information was added.