Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 6 appraisals
Head and neck pathology
A Deep Learning AI Model for Histopathological Diagnosis and Grading of Mucoepidermoid Carcinoma of Salivary Glands
PURPOSE: This study aims to evaluate the effectiveness of deep learning AI model in the diagnosis and grading of mucoepidermoid carcinoma. METHODS: We developed a multiple instance learning based-deep learning model which consisted of two modules. One to perform slide-level classification of whole slide images into mucoepidermoid carcinoma and non-mucoepidermoid carcinoma classes and the second one to perform grading into low, intermediate, and high grades. A total of 194 whole slide images were included in the study. Of these, 92 corresponded to mucoepidermoid carcinoma, while the remaining 102 represented non-mucoepidermoid carcinoma cases. RESULTS: Our results showed that the classification model achieved 90.62% accuracy, 90.62% macro F1 score, 90.60% weighted F1 score, 91.67% macro precision, 91.18% macro recall. The grading model achieved an accuracy of 73.3%, weighted F1-score of 73.9%, weighted precision of 76.1% and weighted recall of 73.3%. CONCLUSION: The proposed deep learning classification model achieved promising performance in differentiating mucoepidermoid carcinoma cases. However, the grading model achieved an overall accuracy of 73.3%. Thus, depending on AI solely is still questionable, however it can be used to augment the work of the pathologist.
22 July 2026
Read appraisal →The American journal of pathology
Deep Learning-Enhanced Biomarker Interpretation on Cytology Cell Blocks: Foundations and Emerging Opportunities in Spatial Pathobiology
Cytology cell block specimens are essential for diagnosing patients with advanced-stage malignancy and often represents the only available strategy for therapeutic biomarker evaluation. The use of cell blocks preserves tumor cells, captures high-grade or metastatic populations, and retains meaningful microenvironmental context, making them well suited for IHC analysis. With the rapid expansion of computational pathology, deep learning-assisted biomarker interpretation in cell blocks is emerging as a promising frontier for improving reproducibility, reducing interobserver variability, and enabling quantitative assessment of spatial tumor-immune interactions. Because many treatment-defining biomarkers are routinely assessed on cytology cell blocks, this mini-review highlights artificial intelligence-based applications for PD-L1, HER2, ER/PR, Ki-67, ALK/ROS1, BRAF V600E, and p16 markers that directly inform decisions about immunotherapy, targeted therapy, and hormone therapy. Also reviewed are emerging predictive models that convey biomarker status directly from morphology, extending the utility of artificial intelligence beyond conventional IHC interpretation. Finally, spatial pathobiology-related opportunities afforded by cell block preparations are discussed, and future directions for integrating artificial intelligence-enabled analysis into cytology workflows are outlined. Together, these advances position cytology cell blocks as an important platform for computational biomarker interpretation and morphology-driven precision oncology.
2 July 2026
Read appraisal →Progress in biomedical engineering (Bristol, England)
Advancement of deep learning models with whole slide image in diagnosis, subtyping and prognosis for glioma.
Gliomas are heterogeneous primary central nervous system tumors with diverse molecular subtypes and variable prognosis. A paradigm shift in histopathology is underway as samples are digitized into whole-slide images (WSIs). Deep learning (DL) shows great potential for glioma diagnosis, subtyping, grading, and prognosis. This review summarizes recent advances in WSI-based DL models, covering data preprocessing, model architectures, and translational performance. Model evolution has progressed from CNNs (e.g. ResNet, capturing local features via convolution) to Transformers (e.g. Vision transformer, modeling global dependencies via self-attention), hybrid architectures, and large language model. Typical pipelines include quality control, stain normalization, patch extraction, feature integration, and prediction. Public datasets such as The Cancer Genome Atlas and Clinical Proteomic Tumor Analysis Consortium serve as key resources. On various datasets (public and institutional), a glioma-specific model ResNet-50 achieved AUC 0.983 for subtype classification; a Vision Transformer model reached AUC 0.960 for molecular typing; a hybrid model, ROAM, attained AUC 0.990; and the pan-cancer hybrid model called CHIEF achieved AUC 0.9397 for diagnosis. Attention heatmaps and Shapley Additive exPlanations provide interpretability by linking model outputs to histologic regions. Novel multimodal fusion integrates genomic, proteomic, radiologic, and clinical data to enhance prediction and uncover biological relevance. Future research may focus on enhancing model generalizability and predictive accuracy via advanced architectures, developing lightweight models for resource-limited settings to facilitate clinical translation.
20 June 2026
Read appraisal →Biomedical physics & engineering express
SAM2HIPT: a hybrid deep learning framework integrating SAM2 and HIPT with joint loss optimization for immunohistochemical cell nucleus segmentation
Nucleus segmentation in immunohistochemistry images plays a critical role in cancer diagnosis and treatment assessment. However, existing methods remain limited in segmentation accuracy and boundary delineation due to staining heterogeneity, densely packed cell distributions, and complex background interference. To address these challenges, this paper proposes a two-stage nucleus segmentation framework, termed SAM2HIPT. In the first stage, the pre-trained Segment Anything Model 2 (SAM2) is employed to generate initial segmentation predictions for input images, wherein the image encoder is kept frozen to preserve the pre-trained visual representation capacity while the mask decoder is fine-tuned to adapt to the characteristics of the pathological image domain; local texture, morphological, and boundary information are extracted through visual feature encoding to produce initial nucleus segmentation masks and spatial prior representations. In the second stage, the Hierarchical Image Pyramid Transformer is introduced to refine the initial segmentation results, performing multi-scale, multi-level feature representation and fusion of morphological, textural, and spatial structural information through a hierarchical vision Transformer architecture, thereby enhancing nuclear structural representation and boundary consistency. To enable collaborative optimization across both stages, a joint loss function is designed to impose unified constraints on segmentation accuracy and feature representation. Evaluated on two public histopathological benchmark datasets, BCData and DeepLIIF, the proposed method achieves Dice coefficients of 0.92 and 0.91, respectively, and HD95 boundary error values of 1.05 pixels and 1.10 pixels, demonstrating superior segmentation performance and robustness over multiple state-of-the-art baseline methods.
11 May 2026
Read appraisal →Journal for immunotherapy of cancer
Deep learning-driven recognition of panoramic tumor microenvironment features in H&E sections and its application
The tumor microenvironment (TME), composed of tumor cells together with stromal cells, immune cells, vascular networks, and other components, constitutes a complex ecosystem that plays a decisive role in tumor initiation, progression, metastasis and therapeutic response. Traditional pathological diagnosis mainly relies on pathologists manually examining H&E-stained tissue sections under the microscope, a method that not only suffers from substantial interobserver variability but also has relatively low analytical efficiency. With the rapid development of computational pathology, the integration of whole-slide imaging technology and deep learning algorithms has provided powerful tools for characterizing tumor microenvironment. These techniques enable automated characterization of cellular, spatial and molecular heterogeneity within the tumor microenvironment, providing integrated insights that advance precision diagnostics and improve prediction of therapeutic response and patient outcomes. Based on a comprehensive review of existing research, this paper highlights recent advances in deep learning-driven recognition of panoramic TME features from H&E slides and their clinical applications and further discusses both the translational potential and current limitations of this technology in oncology research and clinical applications.
11 Apr 2026
Read appraisal →Cancer research
Multicenter Histology Image Integration and Multiscale Deep Learning Support Machine Learning-Enabled Pediatric Sarcoma Classification
UNLABELLED: Pediatric sarcomas present diagnostic challenges due to their rarity and diverse subtypes, often requiring specialized pathology expertise and costly genetic tests. To overcome these barriers, we developed a computational pipeline leveraging deep learning methods to accurately classify pediatric sarcoma subtypes from digitized histology slides. To ensure classifier generalizability and minimize center-specific artifacts, a dataset comprising 867 whole-slide images (WSI) from three medical centers and the Children's Oncology Group was collected and harmonized. Multiple convolutional neural network and vision transformer (ViT) architectures were systematically evaluated as feature extractors for SAMPLER-based WSI representations, and input parameters, such as tile size combinations and resolutions, were tested and optimized. The analysis showed that advanced ViT foundation models (UNI and CONCH) significantly outperformed earlier approaches, and incorporating multiscale features enhanced classification accuracy. The optimized models achieved high performance, distinguishing rhabdomyosarcoma (RMS) from non-RMS soft-tissue sarcomas (NRSTS) with an AUC of 0.969 and differentiating RMS subtypes (alveolar vs. embryonal) with an AUC of 0.961. Additionally, a two-stage pipeline effectively identified scarce Ewing sarcoma images from other NRSTS (AUC = 0.929). Compared with conventional transformer-encoder architectures used for WSI representations, these SAMPLER-based classifiers were three orders of magnitude faster to train, despite operating entirely without a graphical processing unit. This study highlights that digital histopathology paired with rigorous image harmonization provides a powerful solution for pediatric sarcoma classification. SIGNIFICANCE: An approach pairing a multi-institutional dataset with a published pipeline extends imaging-based diagnostic capabilities and creates the potential for accurate, rapid cancer diagnoses across resource-limited and remote settings. This article is part of a special series: Driving Cancer Discoveries with Computational Research, Data Science, and Machine Learning/AI .
5 Apr 2026
Read appraisal →