Histopathology in focus: a review on explainable multi-modal approaches for breast cancer diagnosis

Frontiers in Medicine, Journal Year: 2024, Volume and Issue: 11

Published: Sept. 30, 2024

Precision and timeliness in breast cancer detection are paramount for improving patient outcomes. Traditional diagnostic methods have predominantly relied on unimodal approaches, but recent advancements medical data analytics enabled the integration of diverse sources beyond conventional imaging techniques. This review critically examines transformative potential integrating histopathology images with genomic data, clinical records, histories to enhance accuracy comprehensiveness multi-modal It explores early, intermediate, late fusion methods, as well advanced deep multimodal techniques, including encoder-decoder architectures, attention-based mechanisms, graph neural networks. An overview tasks such Visual Question Answering (VQA), report generation, semantic segmentation, cross-modal retrieval is provided, highlighting utilization generative AI visual language models. Additionally, delves into role Explainable Artificial Intelligence (XAI) elucidating decision-making processes sophisticated algorithms, emphasizing critical need transparency interpretability. By showcasing importance explainability, we demonstrate how XAI Grad-CAM, SHAP, LIME, trainable attention, image captioning, precision, strengthen clinician confidence, foster engagement. The also discusses latest developments, X-VARs, LeGrad, LangXAI, LVLM-Interpret, ex-ILP, their utility detection, while identifying key research gaps proposing future directions advancing field.

Language: Английский

Role of Machine Learning in Acute Stroke Imaging: A Technical Review

Published: Aug. 23, 2024

Language: Английский