{
  "abstract": "Background: Thymic epithelial tumours (TETs) are the most common primary neoplasms of the anterior mediastinum. Despite complete surgical resection (R0), a significant proportion of patients experience recurrence, highlighting the need for improved prognostic stratification. Morphological patterns recognised in clinical practice may have prognostic value, and this study applies deep learning (DL) to haematoxylin and eosin (H&E) whole slide images (WSIs) to predict recurrence in TETs, assisting pathologists in risk assessment and enabling precise treatment planning.\n\nMethods: A multicentre cohort of resected TET patients with complete (R0) resection and no distant metastasis was assembled from Erasmus Medical Centre (EMC), Maastricht University Medical Centre (MUMC+) and the TCGA public dataset, with an external validation cohort from Hospices Civils de Lyon (HCL). WSIs were annotated in QuPath and sampled at 20× magnification into 512×512 patches. HCL WSIs, stained with haematoxylin-eosin saffron, underwent Vahadane normalisation to harmonise staining across centres. Patch embeddings from a pretrained Phikon encoder were aggregated through a multi-instance learning for patient level recurrence prediction, using stratified three-fold cross-validation. GradCAM applied to the most contributing patches generated heatmaps of the areas of highest model activation.\n\nResults: Recurrence occurred in 19 of 298 patients (6.4%) internally and 5 of 95 (5.3%) externally. The model achieved an Area Under the Curve (AUC) of 0.85, a balanced accuracy (BAcc) of 0.71, and sensitivity of 79% internally. Performance was preserved externally (AUC 0.70, BAcc 0.62, sensitivity 80%) despite domain shift. In true positive cases, heatmaps consistently localised to tumour tissue adjacent to fibrous septa, capsular structures and vascular interfaces, with additional focal activation in areas of increased cellular heterogeneity in tumours lacking prominent separation.\n\nConclusions: The DL model can identify TET patients at risk of recurrence following R0 resection and demonstrates generalisability across institutions with differing tissue preparation protocols. Heatmaps corresponded to septal, capsular and perivascular structures, sites classically associated with tumour invasion and recurrence risk in TETs, suggesting the model may capture histological correlates of invasive behaviour. Future work will focus on larger datasets, multimodal architectures integrating clinical and imaging data, and further external validation.",
  "abstract_shown": true,
  "affiliations": [
    "Department of Pathology and Clinical Bioinformatics, Erasmus Medical Centre, Rotterdam, The Netherlands",
    "Department of Radiation Oncology (Maastro), GROW Research Institute for Oncology and Reproduction, Maastricht University Medical Centre+, Maastricht, The Netherlands",
    "Department of Respiratory Medicine, Erasmus Medical Centre, Rotterdam, The Netherlands",
    "Department of Pulmonology, Maastricht UMC+, Maastricht, the Netherlands",
    "Department of Pathology, Maastricht UMC+, Maastricht, the Netherlands",
    "Department of Radiotherapy, Erasmus MC Cancer Institute, Rotterdam, The Netherlands",
    "Department of Pathology, Centre Hospitalier Universitaire de Lyon, Lyon, France",
    "Department of Medical Oncology, Curie Institute, Paris, France"
  ],
  "article_type": "abstract",
  "arxiv": null,
  "author_total": 14,
  "authors": [
    {
      "affs": [
        0
      ],
      "name": "Anna Salut Esteve Domínguez"
    },
    {
      "affs": [
        1
      ],
      "name": "Stéphanie Peeters"
    },
    {
      "affs": [
        2
      ],
      "name": "Marlou Dimmers"
    },
    {
      "affs": [
        2
      ],
      "name": "Daphne Dumoulin"
    },
    {
      "affs": [
        3
      ],
      "name": "Florit Marcuse"
    },
    {
      "affs": [
        4
      ],
      "name": "Myrurgia Abdul-Hamid"
    },
    {
      "affs": [
        1
      ],
      "name": "Rianne Fijten"
    },
    {
      "affs": [
        1
      ],
      "name": "Yunlei Li"
    },
    {
      "affs": [
        1
      ],
      "name": "Andrew Stubbs"
    },
    {
      "affs": [
        1,
        5
      ],
      "name": "Dirk De Ruysscher"
    },
    {
      "affs": [
        6
      ],
      "name": "Lara Chalabreysse"
    },
    {
      "affs": [
        7
      ],
      "name": "Nicolas Girard"
    },
    {
      "affs": [
        0
      ],
      "name": "Jan von der Thüsen"
    },
    {
      "affs": [
        0
      ],
      "name": "Farhan Akram"
    }
  ],
  "bibtex": "https://research.draconitic.com/paper/fzcy2dsbjzgbr.bib",
  "cited_by": 0,
  "doi": "10.21037/med-26-ab025",
  "fulltext_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC13744058/",
  "id": "fzcy2dsbjzgbr",
  "issns": [
    "2522-6711"
  ],
  "landing_url": "https://doi.org/10.21037/med-26-ab025",
  "language": "en",
  "licence": "CC-BY-NC-ND-4.0",
  "licence_label": "CC BY-NC-ND 4.0",
  "licence_source": "article",
  "licence_url": "https://creativecommons.org/licenses/by-nc-nd/4.0/",
  "metadata_licence": null,
  "native_subjects": [
    [
      "heading",
      "Original Research"
    ],
    [
      "openalex",
      "1702|AI in cancer detection"
    ]
  ],
  "openalex": "W7225089561",
  "pdf_url": null,
  "pmcid": "PMC13744058",
  "pmid": null,
  "published": "2026-10-30",
  "publisher": "AME Publications",
  "ris": "https://research.draconitic.com/paper/fzcy2dsbjzgbr.ris",
  "sources": [
    {
      "first_seen": "2026-11-11T12:21:38Z",
      "native_id": "PMC13744058",
      "source": "pmc",
      "url": "https://doi.org/10.21037/med-26-ab025"
    }
  ],
  "subjects": [
    {
      "field": "medicine",
      "origin": "journal-openalex:F27",
      "subfield": "general"
    }
  ],
  "title": "AB025. Deep learning model predicts recurrence in thymic epithelial tumours: a multicentre study",
  "title_text": "AB025. Deep learning model predicts recurrence in thymic epithelial tumours: a multicentre study",
  "type": "peer-reviewed",
  "updated_at": "2026-11-11T12:21:38Z",
  "url": "https://research.draconitic.com/paper/fzcy2dsbjzgbr",
  "venue": "Mediastinum",
  "versions": []
}