EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Tumor Bud Classification in Colorectal Cancer Using Attention-Based Deep Multiple Instance Learning and Domain-Specific Foundation Models

Mesut Şeker iD, M Khalid Khan Niazi, Wei Chen iD, Wendy L Frankel, Metin N Gurcan iD

Cancers · 07.04.2025 · 17 (7) · 1245

DOI: 10.3390/cancers17071245 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Cancers, online erschienen am 07.04.2025. Verfasst von Mesut Şeker und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Electrical and Electronics Engineering, Dicle University, Diyarbakir 21280, Turkey. Die Arbeit wurde bisher 16-mal zitiert (Crossref) und nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NCI NIH HHS, The National Cancer Institute.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background/Objectives: Identifying tumor budding (TB) in colorectal cancer (CRC) is vital for better prognostic assessment as it may signify the initial stage of metastasis. Despite its importance, TB detection remains challenging due to subjectivity in manual evaluations. Identifying TBs remains difficult, especially at high magnification levels, leading to inconsistencies in prognosis. To address these issues, we propose an automated system for TB classification using deep learning.

Methods

We trained a deep learning model to identify TBs through weakly supervised learning by aggregating positive and negative bags from the tumor invasive front. We assessed various foundation models for feature extraction and compared their performance. Attention heatmaps generated by attention-based multi-instance learning (ABMIL) were analyzed to verify alignment with TBs, providing insights into the interpretability of the features. The dataset includes 29 WSIs for training and 70 whole slide images (WSIs) for the hold-out test set.

Results

In six-fold cross-validation, Phikon-v2 achieved the highest average AUC (0.984 ± 0.003), precision (0.876 ± 0.004), and recall (0.947 ± 0.009). Phikon-v2 again achieved the highest AUC (0.979) and precision (0.980) on the external hold-out test set. Moreover, its recall rate (0.910) was still higher than that of UNI’s (0.879). UNI exhibited a balanced performance on the hold-out test set, with an AUC of 0.960 and a precision of 0.968. CtransPath showed strong precision on the external hold-out test set (0.947) but had a slightly lower recall (0.911).

Conclusions

The proposed technique enhances the accuracy of TB assessment, offering potential applications for CRC and other cancer types.

Schlagwörter

Stichwörter der Autor:innen

colorectal cancerfoundation modelsmulti-instance learningtumor budding

Autor:innen und Institutionen

  • Department of Electrical and Electronics Engineering, Dicle University, Diyarbakir 21280, Turkey.
  • Department of Pathology, The Ohio State University, Columbus, OH 43210, USA.
  • Center for Artificial Intelligence Research, Wake Forest University School of Medicine, Winston-Salem, NC 27101, USA.

ORCID-Profile: Mesut Şeker, Wei Chen, Metin N Gurcan

Förderung

  • NCI NIH HHS
  • The National Cancer Institute

Zitieren

APA

Şeker, M., Niazi, M. K. K., Chen, W., Frankel, W. L., & Gurcan, M. N. (2025). Tumor Bud Classification in Colorectal Cancer Using Attention-Based Deep Multiple Instance Learning and Domain-Specific Foundation Models. Cancers, 17(7), 1245. https://doi.org/10.3390/cancers17071245

Vancouver

Şeker M, Niazi MKK, Chen W, Frankel WL, Gurcan MN. Tumor Bud Classification in Colorectal Cancer Using Attention-Based Deep Multiple Instance Learning and Domain-Specific Foundation Models. Cancers. 2025;17(7):1245. doi:10.3390/cancers17071245

BibTeX
@article{eker2025,
  author = {Şeker, Mesut and Niazi, M Khalid Khan and Chen, Wei and Frankel, Wendy L and Gurcan, Metin N},
  title = {Tumor Bud Classification in Colorectal Cancer Using Attention-Based Deep Multiple Instance Learning and Domain-Specific Foundation Models},
  journal = {Cancers},
  year = {2025},
  volume = {17},
  number = {7},
  pages = {1245},
  doi = {10.3390/cancers17071245}
}
Kostenlos starten →