Journal Article

DKFZ-2026-01659

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png AI-based selection of tumor regions for genomic profiling in neuropathology.

Ghaffari Laleh, N. ; Friedrich, L.DKFZ* ; Aras, F. K.DKFZ* ; Hewitt, K. J. ; Schweizer, L.DKFZ* ; Carrero, Z. I. ; Savran, D.DKFZ* ; Haag, D.DKFZ* ; Barbosa, S.DKFZ* ; Sahm, F.DKFZ* ; Kather, J. N.DKFZ*

2026
Oxford University Press Oxford

Abstract: Automating pathology workflows with deep learning is increasingly feasible and clinically relevant. We present an AI-based method that identifies diagnostically relevant areas directly from H&E-stained slides, trained on 250 glioma cases using sparse, incomplete annotations. First, we show that attention-based multiple instance learning achieves accurate predictions despite noisy labels, easing the annotation burden. Second, the model highlights tumor regions with high cellularity or grade, offering reproducible guidance for tissue selection. In a prospective evaluation, AI-selected regions achieved a mean Dice score of 0.743 [±0.077], supporting integration into neuropathology workflows as reliable guidance for molecular diagnostics.

Keyword(s): glioma ; molecular diagnostics ; multiple instance learning ; tumor region segmentation ; weakly supervised learning

Note: #DKTKZFB26# / #NCTZFB26#

---

Contributing Institute(s):

1. KKE Neuropathologie (B300)
2. DKTK HD zentral (HD01)
3. DKTK Koordinierungsstelle Frankfurt (FM01)
4. Koordinierungsstelle NCT Heidelberg (HD02)

Research Program(s):

1. 312 - Funktionelle und strukturelle Genomforschung (POF4-312) (POF4-312)

Appears in the scientific report 2026

---

Database coverage:
; ; Article Processing Charges ; Clarivate Analytics Master Journal List ; DOAJ Seal ; Emerging Sources Citation Index ; Fees ; IF < 5 ; JCR ; PubMed Central ; SCOPUS ; Web of Science Core Collection

---