DigiPatICS: Projecte d'optimització del diagnòstic anatomopatòleg en xarxa als hospitals de l'Institut Català de la Salut a través de la digitalització i eines d'intel·ligència Artificial

Type Start End
Other Oct 2020 Dec 2025
Responsible URL
Ferran Marqués DigiPatICS notice page @ICS

Reference

Institut Català de la Salut (ICS). Palex Medical. Programa Operatiu FEDER de Catalunya 2014-2020.

Description

GPI collaborates with the Institut Català de la Salut (ICS) in the context of the DigiPatICS project. The first phase of DigiPatICS (10/2020-12/2023) was conducted under the Framework of the Catalonia FEDER Program where UPC is an LTP of Palex Medical. The second phase (08/2024-12/2025) is directly funded by ICS with a potential extension until 2027.

DigiPatICS aims to enhance patient safety and quality of care by improving the diagnosis of histopathological samples and optimizing ICS digital pathology (DP) services and Artificial Intelligence (AI) tools.

These tools have been deployed in 8 ICS hospitals: HU Bellvitge, HU Doctor Trueta, H. Verge de la Cinta, HU Germans Trias i Pujol, HU Joan XXIII, Viladecans Hospital, HU Arnau de Vilanova, and HU Vall d'Hebron. It involves over 170 pathologists, 24 scanners connected to 13 macropath stations, generating a million scanned images annually (Whole Slide Images-WSI). Possibly one of the largests DP networks in the world.

DigiPatICS Scanners in situ DigiPatICS Scanners in situ
DigiPatICS scanners installed in ICS Hospitals.
Source: ICS news 03/2021
Pathologist and resident diagnosing in the office.
Source: Temprana-Salvador et al (2022)

GPI has developed algorithms to assist pathologists’ diagnosis by automatically analyzing WSI images, enabling rapid and consistent biomarker quantification and statistically robust results. Initially, ICS focused the project on breast cancer due to its high prevalence (it is estimated that one in eight women has had or will have breast cancer). Currently, we develop new algorithms to support doctors in other pathologies: analysis of Hematoxylin & Eosin (H&E) images for breast and lung cancer, in-situ hybridization (ISH) for gastric cancer, cytology for bronchoalveolar lavage (BAL) studies, and radiographic images for knee osteoarthritis assessment.

In a biopsy for suspected breast cancer, a tissue section is first analyzed using H&E staining to detect cancer presence. Other sections are immunohistochemically stained (IHC) to identify the cancer type by computing cells reacting to biomarkers and determining the optimal treatment.

Notably, WSI images are enormous, in the order of tens of GigaPixels (100Kpix x 200Kpix) and contain tens of thousands cells. Manual counting by pathologists on small tissue areas (about a hundred cells each) is assumed to be representative, but may lose statistical significance. 

GPI has created deep learning models to study, detect, and classify cells in IHC stains for pathologists' diagnoses. Semantic and instance segmentation algorithms with convolutional and graph-based neural networks are tailored to each medical problem. These algorithms process all cells in WSI images offline (e.g., overnight), and results are stored for quick access. Pathologists visually supervise the result and retrieve calculated values to quantify biomarkers over selected areas, often much larger than those chosen for manual calculation, improving statistical precision of diagnosis. 

In the first stage of the DigiPatICS project, four algorithms were developed for analyzing stains based on the presence of HER2 and Ki67 proteins, and estrogen and progesterone receptors.

 

Students' contributions to DigiPatICS

By project type and in descending order of submission date.

Master Thesis

Degree Project

i2Rs/CBIs/internships

  • Sáez Parés, Laura (Spring 2024). Development of biomedical analysis algorithms, Internship report, DSE degree, FIB/ETSETB-UPC, adv. F. Marqués
  • Beaus Iranzo, Pablo (Spring 2023). Tissue Detection in Whole Slide Images, i2R report, MATT master, ETSETB-UPC, adv. S. Rabanaque, J.R. Casas, F. Marqués
  • Le Boeuf Flo, Andres (Spring 2023). Detection and classification of cancerous cells, Internship, MATT master, ETSETB-UPC, adv. M. Pardàs, F. Marqués
  • Rodríguez Muñoz, Martín (Spring 2023). Analysis and segmentation of immunohistochemistry images for breast cancer diagnosis: HER2 Staining, I2RCED report, DSE degree, FIB/ETSETB-UPC, adv. P. Salembier, F. Marqués
  • Qiu, Yikai (Fall 2022). Ductal Carcinoma in situ (DCIS) segmentation on HE whole slides by... truth, I2RCED report, DSE degree, FIB/ETSETB-UPC, adv. F. Marqués, J.R. Casas, S. Rabanaque
  • Rubio Jornet, Víctor (Fall 2022). Improving WSI registration for cross-slide histopathological analysis of multiple stains, i2R report, MATT master, ETSETB-UPC, adv. J.R. Casas, S. Rabanaque
  • Formosa Marin, Feliu (Spring 2022). Cell detection and classification of lung cancer histological images using 2 single-purposed parallel neural networks, i2R report, MATT master, ETSETB-UPC, adv. P. Salembier.
  • González i Juclà, Daniel (Fall 2021). Cell nuclei detection and segmentation from registered HER2 and CK19 histological images, I2RCED report, DSE degree, FIB/ETSETB-UPC, adv. J.R. Casas, V. Vilaplana
  • Pereira Cánovas, Anxo-Lois (Fall 2021). Cancerous nuclei segmentation and classification in histological images models comparison, I2RCED report, DSE degree, FIB/ETSETB-UPC, adv. F. Marqués, P. Salembier, M. Pardàs
  • Turch Ferreres, Arnau (Fall 2021). Upgrading the breast tumor detection in Hematoxylin & Eosin stained biopsy images, I2RCED report, DSE degree, FIB/ETSETB-UPC, adv. J.R. Casas, V. Vilaplana
  • Pereira Dos Santos, Victor (Spring 2021). In-situ tumor detection in H&E images, Internship, MATT master, ETSETB-UPC, adv. M. Pardàs, F. Marqués
  • Pujol Salido, David (Spring 2021). Tumor detection in HER2 images, I2R report, MATT master, ETSETB-UPC, adv. V. Vilaplana
  • Rabanaque Rodríguez, Sonia (Spring 2021). Alignment of multi-stained digital histological images, CBI report, DSE degree, FIB/ETSETB-UPC, adv. J.R. Casas, V. Vilaplana
  • Ramírez Márquez, Alex (Spring 2021). Tumor detection in HER2 images, I2R report, MATT master, ETSETB-UPC, adv. V. Vilaplana
  • Aguilera González, Cristina (Fall 2020). Cell nuclei detection and segmentation in histological images based on a deep semantic segmentation network, CBI report, DSE degree, FIB/ETSETB-UPC, adv. F. Marqués, P. Salembier
  • Fernández López, Christian (Fall 2020). HER2 nuclei and membrane segmentation, I2R report, MATT master, ETSETB-UPC, adv. V. Vilaplana
  • Magariño Llavero, Pau (Fall 2020). Cancerous nuclei segmentation using object detection, CBI report, DSE degree, FIB/ETSETB-UPC, adv. F. Marqués, M. Pardàs
  • Rosell Murillo, Marina (Fall 2020). Use of Image 2 Image translation to handle variations between multicentric HER2 stains of breast cancer biopsy, CBI report, DSE degree, FIB/ETSETB-UPC, adv. F. Marqués, M. Pardàs

Current students (MsTh, DegProj, i2R/CBI/internships)

  • González Mestre, Adrià (exp. Spring 2025). Breast tumor detection in histological images with NNUnet, Master Thesis MATT, ETSETB-UPC, adv. V. Vilaplana
  • Alonso del Hoyo, Sara (exp. Spring 2025). Automatic quantization of Bronchoalveolar Lavage (BAL) images, Bachelor Thesis DSE, ETSETB/FIB-UPC, adv. M. Pardàs, F. Marqués
  • Caus Solé, Albert (exp. Spring 2025). Co-registration of histopathological images in the DigiPatICS project, internship and Bachelor Thesis GRETST, ETSETB-UPC, adv. J.R. Casas
  • Ramon Montmany, Laura (exp. Spring 2025). Breast tumor detection in histological images using deep learning, Bachelor Thesis DSE, ETSETB/FIB-UPC, adv. V. Vilaplana, L. Jiménez
  • Roca Humanes, Júlia (exp. Spring 2025). Functionalities of co-registration of histopathological images, Bachelor Thesis GRETST, ETSETB-UPC, adv. J.R. Casas
  • Esquerrà Giné, Paula (exp. Spring 2025). Automatic Knee Osteoarthritis Severity Grading, I2RCED report, DSE degree, FIB/ETSETB-UPC, adv. P. Salembier, F. Marqués
  • Cruz Carné, Nil (exp. Spring 2025). Quantification of In Situ Hybridization (ISH) images for HER2 assessment in gastric cancer, i2R and MET master, ETSETB-UPC, adv. D. Anglada, M. Pardàs

 

 

DigiPatICS in the Media


Sáez, C.Cancer Revolution: How AI is Improving Diagnosis and Treatment”. Ara.cat, Aug 2024

Catalan Public Hospitals Use Pioneering Project to Detect Tumors More Accurately and Safely

Jordi Temprana, pathologist at HU Vall d'Hebron, and Ferran Marqués, project coordinator at UPC.
Image crédits:
Cristina Calderer, Ara newspaper

 

Vall d’Hebron Barcelona Hospital Campus, “ICS hospitals use artificial intelligence to improve breast cancer diagnosis”. Blog Post, Jan 2024

The DigiPatICS project has created the largest digital network of pathological anatomy in Europe and facilitates sharing of diagnoses and knowledge in real time.

The Catalan Health Institute (ICS) has developed four AI algorithms of its own for improved breast cancer diagnosis based on the quantification of biomarkers HER2, Ki67, estrogen and progesterone receptors.
Source:
Vall d’Hebron Barcelona Hospital Campus

 

Publicaciones del Sur SA. “Four algorithms improve breast cancer diagnosis in seven Catalan hospitals”. Andalucía información, Jan 2024

Four algorithms improve breast cancer diagnosis in seven Catalan hospitals.

The algorithms work by quantifying four biomarkers that are indicators of breast cancer types.
Source:
Andalucía información

Cercle TecnològicDigiPATICS, the program for digitizing pathological anatomy samples”. Metadata magazine, July 2021

The project led by the ICS will begin to be implemented at the Vall d'Hebron Hospital this September and in the rest of Catalan centers over the next few months.

The DigiPATICS project will accelerate and optimize the management and diagnosis processes
of pathological anatomy samples in Catalan hospitals.
Source:
Metadata magazine

 

 

Publications

Pina O, Dorca E, Vilaplana V. Cell-DETR: Efficient cell detection and classification in WSIs with transformers. In: Medical Imaging with Deep Learning (MIDL 2024). Medical Imaging with Deep Learning (MIDL 2024). ; 2024.
Pina O, Vilaplana V. Unsupervised Domain Adaptation for Multi-Stain Cell Detection in Breast Cancer with Transformers. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (DEF-AI-MIA workshop). IEEE/CVF Conference on Computer Vision and Pattern Recognition (DEF-AI-MIA workshop). ; 2024.
Anglada D, Sala J, Marqués F, Salembier P, Pardàs M. Enhancing Ki-67 Cell Segmentation with Dual U-Net Models: A Step Towards Uncertainty-Informed Active Learning. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops. IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops. IEEE; 2024. (838.72 KB)
Perez-Cano J, Valero ISansano, Anglada D, Pina O, Salembier P, Marqués F. Combining graph neural networks and computer vision methods for cell nuclei classification in lung tissue. Heliyon. 2024 ;10(7). (4.05 MB)
Pardàs M, Anglada D, Espina M, Marques F, Salembier P. Stromal tissue segmentation in Ki67 histology images based on cytokeratin-19 stain translation. JOURNAL OF MEDICAL IMAGING. 2023 ;10(3). (13.78 MB)
Temprana-Salvador J, López-García P, Vives JCastellví, de Haro L, Ballesta E, Abusleme MRojas, Arrufat M, Marques F, Casas J, Gallego C, et al. DigiPatICS: Digital Pathology Transformation of the Catalan Health Institute Network of 8 Hospitals - Planification, Implementation and Preliminary Results. Diagnostics. 2022 ;12(4). (5.87 MB)