Soori, Ashkan
(2026)
Quantitative image analysis of H&E and HER2 IHC whole-slide images for HER2 prediction in breast cancer.
[Laurea magistrale], Università di Bologna, Corso di Studio in
Physics [LM-DM270], Documento full-text non disponibile
Il full-text non è disponibile per scelta dell'autore.
(
Contatta l'autore)
Abstract
The histopathological diagnosis of breast cancer routinely relies on hematoxylin and eosin (H&E) staining to assess tissue morphology and immunohistochemistry (IHC) to evaluate biomarker expression. HER2 is an important biomarker in breast cancer, playing a key role in prognosis and treatment selection. This study investigated whether quantitative image features extracted from paired H&E and HER2 IHC whole-slide images could predict HER2 status and provide insight into the relationship between tissue morphology and HER2 expression. A feature-based machine learning framework was developed and evaluated using both a public breast cancer dataset and a clinical dataset. Quantitative image features describing tissue texture, cellular morphology, and staining intensity were extracted from image patches and used for classification and correlation analysis.
The extracted features enabled accurate discrimination between HER2-positive and HER2-negative tumours in the public dataset, although performance decreased in the clinical dataset because of the greater variability of routine clinical material. Correlation analysis revealed a moderate but significant association between quantitative features extracted from H&E and HER2 IHC images, indicating that the two modalities provide complementary rather than redundant information. Feature importance analysis identified tissue texture, cellular morphology, and staining intensity as the most informative feature categories for HER2 prediction. Overall, the findings demonstrate that routine H&E images contain biologically meaningful quantitative information related to HER2 status. Although H&E cannot replace HER2 IHC for definitive assessment of HER2 status, the proposed framework highlights the potential of quantitative image analysis to provide objective and reproducible decision support for pathologists and supports the development of computational pathology approaches for breast cancer diagnosis.
Abstract
The histopathological diagnosis of breast cancer routinely relies on hematoxylin and eosin (H&E) staining to assess tissue morphology and immunohistochemistry (IHC) to evaluate biomarker expression. HER2 is an important biomarker in breast cancer, playing a key role in prognosis and treatment selection. This study investigated whether quantitative image features extracted from paired H&E and HER2 IHC whole-slide images could predict HER2 status and provide insight into the relationship between tissue morphology and HER2 expression. A feature-based machine learning framework was developed and evaluated using both a public breast cancer dataset and a clinical dataset. Quantitative image features describing tissue texture, cellular morphology, and staining intensity were extracted from image patches and used for classification and correlation analysis.
The extracted features enabled accurate discrimination between HER2-positive and HER2-negative tumours in the public dataset, although performance decreased in the clinical dataset because of the greater variability of routine clinical material. Correlation analysis revealed a moderate but significant association between quantitative features extracted from H&E and HER2 IHC images, indicating that the two modalities provide complementary rather than redundant information. Feature importance analysis identified tissue texture, cellular morphology, and staining intensity as the most informative feature categories for HER2 prediction. Overall, the findings demonstrate that routine H&E images contain biologically meaningful quantitative information related to HER2 status. Although H&E cannot replace HER2 IHC for definitive assessment of HER2 status, the proposed framework highlights the potential of quantitative image analysis to provide objective and reproducible decision support for pathologists and supports the development of computational pathology approaches for breast cancer diagnosis.
Tipologia del documento
Tesi di laurea
(Laurea magistrale)
Autore della tesi
Soori, Ashkan
Relatore della tesi
Correlatore della tesi
Scuola
Corso di studio
Indirizzo
Applied Physics
Ordinamento Cds
DM270
Parole chiave
HER2 IHC,HER2 prediction,Digital pathology,Quantitative image analysis,Machine learning,H&E,Feature extraction,Histopathology,Image texture analysis,Breast cancer
Data di discussione della Tesi
29 Ottobre 2026
URI
Altri metadati
Tipologia del documento
Tesi di laurea
(NON SPECIFICATO)
Autore della tesi
Soori, Ashkan
Relatore della tesi
Correlatore della tesi
Scuola
Corso di studio
Indirizzo
Applied Physics
Ordinamento Cds
DM270
Parole chiave
HER2 IHC,HER2 prediction,Digital pathology,Quantitative image analysis,Machine learning,H&E,Feature extraction,Histopathology,Image texture analysis,Breast cancer
Data di discussione della Tesi
29 Ottobre 2026
URI
Gestione del documento: