Diagnostics, Год журнала: 2025, Номер 15(11), С. 1389 - 1389

Опубликована: Май 30, 2025

Background/Objectives: Breast cancer diagnosis heavily relies on histopathological assessment, which is prone to subjectivity and inefficiency, especially with whole-slide imaging (WSI). This study addressed these limitations by developing an automated breast cell classification algorithm using information-extreme machine learning approach universal cytological features, aiming for objective generalized diagnosis. Methods: Digitized histological images were processed identify hyperchromatic cells. A set of 21 features (10 geometric 11 textural), chosen their potential universality across cancers, extracted from individual These then used classify cells as normal or malignant algorithm. optimizes information criterion within a binary Hamming space achieve robust recognition minimal input features. The architectural innovation lies in the application this feature analysis classification. Results: algorithm’s functional efficiency was evaluated dataset 176 labeled images, yielding promising results: accuracy 89%, precision 85%, recall 84%, F1-score 88%. metrics demonstrate balanced effective model Conclusions: proposed utilizing offers potentially computationally efficient alternative traditional methods may mitigate some deep analysis. Future work will focus validating larger datasets exploring its applicability other types.

Язык: Английский