Mismatch Repair Deficiency as a Predictive and Prognostic Biomarker in Endometrial Cancer: A Review on Immunohistochemistry Staining Patterns and Clinical Implications

International Journal of Molecular Sciences, Journal Year: 2024, Volume and Issue: 25(2), P. 1056 - 1056

Published: Jan. 15, 2024

Among the four endometrial cancer (EC) TCGA molecular groups, MSI/hypermutated group represents an important percentage of tumors (30%), including different histotypes, and generally confers intermediate prognosis for affected women, also providing new immunotherapeutic strategies. Immunohistochemistry MMR proteins (MLH1, MSH2, MSH6 PMS2) has become optimal diagnostic MSI surrogate worldwide. This review aims to provide state-of-the-art knowledge on deficiency/MSI in EC clarify pathological assessment, interpretation pitfalls reporting status.

Language: Английский

---

Immune subtyping of melanoma whole slide images using multiple instance learning

Medical Image Analysis, Journal Year: 2024, Volume and Issue: 93, P. 103097 - 103097

Published: Feb. 1, 2024

Determining early-stage prognostic markers and stratifying patients for effective treatment are two key challenges improving outcomes melanoma patients. Previous studies have used tumour transcriptome data to stratify into immune subgroups, which were associated with differential specific survival potential predictive biomarkers. However, acquiring is a time-consuming costly process. Moreover, it not routinely in the current clinical workflow. Here, we attempt overcome this by developing deep learning models classify gigapixel haematoxylin eosin (H&E) stained pathology slides, well established workflows, these subgroups. We systematically assess six different multiple instance (MIL) frameworks, using five image resolutions three feature extraction methods. show that pathology-specific self-supervised 10x resolution patches generate superior representations classification of subtypes. In addition, primary dataset, achieve mean area under receiver operating characteristic curve (AUC) 0.80 classifying histopathology images 'high' or 'low immune' subgroups AUC 0.82 an independent TCGA dataset. Furthermore, able significantly (log rank test, P< 0.005). anticipate MIL methods will allow us find new biomarkers high importance, act as tool clinicians infer landscape tumours patients, without needing carry out additional expensive genetic tests.

Language: Английский

---

Recent Application of Artificial Intelligence in Non-Gynecological Cancer Cytopathology: A Systematic Review

Cancers, Journal Year: 2022, Volume and Issue: 14(14), P. 3529 - 3529

Published: July 20, 2022

State-of-the-art artificial intelligence (AI) has recently gained considerable interest in the healthcare sector and provided solutions to problems through automated diagnosis. Cytological examination is a crucial step initial diagnosis of cancer, although it shows limited diagnostic efficacy. Recently, AI applications processing cytopathological images have shown promising results despite elementary level technology. Here, we performed systematic review with quantitative analysis recent non-gynecological (non-GYN) cancer cytology understand current technical status. We searched major online databases, including MEDLINE, Cochrane Library, EMBASE, for relevant English articles published from January 2010 2021. The query terms were: “artificial intelligence”, “image processing”, “deep learning”, “cytopathology”, “fine-needle aspiration cytology.” Out 17,000 studies, only 26 studies (26 models) were included full-text review, whereas 13 analysis. There eight classes models treated according target organs: thyroid (n = 11, 39%), urinary bladder 6, 21%), lung 4, 14%), breast 2, 7%), pleural effusion ovary 1, 4%), pancreas prostate 4). Most focused on classification segmentation tasks. Although most showed impressive results, sizes training validation datasets limited. Overall, also non-GYN cytopathology analysis, such as pathology or gynecological cytology. However, lack well-annotated, large-scale Z-stacking external cross-validation was limitation found across all studies. Future larger high-quality annotations are required.

Language: Английский

---

Deep Learning-Based Computational Cytopathologic Diagnosis of Metastatic Breast Carcinoma in Pleural Fluid

Cells, Journal Year: 2023, Volume and Issue: 12(14), P. 1847 - 1847

Published: July 13, 2023

A Pleural effusion cytology is vital for treating metastatic breast cancer; however, concerns have arisen regarding the low accuracy and inter-observer variability in cytologic diagnosis. Although artificial intelligence-based image analysis has shown promise cytopathology research, its application diagnosing cancer pleural fluid remains unexplored. To overcome these limitations, we evaluate diagnostic of an model using a large collection cytopathological slides, to detect malignant associated with cancer. This study includes total 569 cytological slides from various institutions. We extracted 34,221 augmented patches whole-slide images trained validated deep convolutional neural network (DCNN) (Inception-ResNet-V2) images. Using this model, classified 845 randomly selected patches, which were reviewed by three pathologists compare their accuracy. The DCNN outperforms demonstrating higher accuracy, sensitivity, specificity compared (81.1% vs. 68.7%, 95.0% 72.5%, 98.6% 88.9%, respectively). discordant cases DCNN. After re-examination, average improved 87.9, 80.2, 95.7%, respectively. shows that can accurately diagnose potential support pathologists.

Language: Английский

---

Recent application of artificial intelligence on histopathologic image-based prediction of gene mutation in solid cancers

Briefings in Bioinformatics, Journal Year: 2023, Volume and Issue: 24(3)

Published: April 27, 2023

Abstract Purpose Evaluation of genetic mutations in cancers is important because distinct mutational profiles help determine individualized drug therapy. However, molecular analyses are not routinely performed all they expensive, time-consuming and universally available. Artificial intelligence (AI) has shown the potential to a wide range on histologic image analysis. Here, we assessed status mutation prediction AI models images by systematic review. Methods A literature search using MEDLINE, Embase Cochrane databases was conducted August 2021. The articles were shortlisted titles abstracts. After full-text review, publication trends, study characteristic analysis comparison performance metrics performed. Results Twenty-four studies found mostly from developed countries, their number increasing. major targets gastrointestinal, genitourinary, gynecological, lung head neck cancers. Most used Cancer Genome Atlas, with few an in-house dataset. area under curve some cancer driver gene particular organs satisfactory, such as 0.92 BRAF thyroid 0.79 EGFR cancers, whereas average 0.64, which still suboptimal. Conclusion predict appropriate caution. Further validation larger datasets required before can be clinical practice mutations.

Language: Английский

---

Diagnostic Challenges during Inflammation and Cancer: Current Biomarkers and Future Perspectives in Navigating through the Minefield of Reactive versus Dysplastic and Cancerous Lesions in the Digestive System

International Journal of Molecular Sciences, Journal Year: 2024, Volume and Issue: 25(2), P. 1251 - 1251

Published: Jan. 19, 2024

In the setting of pronounced inflammation, changes in epithelium may overlap with neoplasia, often rendering it impossible to establish a diagnosis certainty daily clinical practice. Here, we discuss underlying molecular mechanisms driving tissue response during persistent inflammatory signaling along potential association cancer gastrointestinal tract, pancreas, extrahepatic bile ducts, and liver. We highlight histopathological challenges encountered chronic inflammation routine practice pinpoint tissue-based biomarkers that could complement morphology differentiate reactive from dysplastic or cancerous lesions. refer advantages limitations existing employing immunohistochemistry point promising new markers, including generation novel antibodies targeting mutant proteins, miRNAs, array assays. Advancements experimental models, mouse 3D have improved our understanding response. The integration digital pathology artificial intelligence also visual inspections. Navigating through responses various contexts will help us develop reliable improve diagnostic decisions ultimately patient treatment.

Language: Английский

---

Whole Slide Image-Level Classification of Malignant Effusion Cytology Using Clustering-Constrained Attention Multiple Instance Learning

Published: Jan. 1, 2025

Background: Cytological diagnosis of pleural effusion plays an important role in the early detection and lung cancers. Recently, attempts have been made to overcome low diagnostic accuracy interobserver variability using artificial intelligence-based image analysis. However, such analysis is primarily performed at image-patch level not whole-slide (WSI) level. This study aims develop a WSI-level classification malignant effusions metastatic cancer based on fluid cytology quality-controlled, nationwide dataset.Methods: The dataset was collected by consortium research group that included three major university hospitals Committee Quality Assurance Program Korean Society Cytopathology. It contains 566 normal 271 WSIs from fluids. A clustering-constrained attention multiple-instance learning (CLAM) model used for classification.Results: CLAM achieved high 97%, with area under curve 0.97, representing 13% improvement over patch model-based WSI classification. also significantly reduced time computing resources compared those required during patch-level heat map generation WSIs.Conclusion: successfully demonstrated performance differentiating large, dataset. Further external validation ensure generalizability.

Language: Английский

---

Effectiveness of encoder-decoder deep learning approach for colorectal polyp segmentation in colonoscopy images

Applied Intelligence, Journal Year: 2025, Volume and Issue: 55(4)

Published: Jan. 10, 2025

Language: Английский

---

Artificial Intelligence and Whole Slide Imaging, a new tool for the Microsatellite Instability prediction in Colorectal Cancer: friend or foe?

Critical Reviews in Oncology/Hematology, Journal Year: 2025, Volume and Issue: unknown, P. 104694 - 104694

Published: March 1, 2025

Language: Английский

---

Diagnosis Test Accuracy of Artificial Intelligence for Endometrial Cancer: Systematic Review and Meta-Analysis

Journal of Medical Internet Research, Journal Year: 2025, Volume and Issue: 27, P. e66530 - e66530

Published: April 18, 2025

Background Endometrial cancer is one of the most common gynecological tumors, and early screening diagnosis are crucial for its treatment. Research on application artificial intelligence (AI) in endometrial increasing, but there currently no comprehensive meta-analysis to evaluate diagnostic accuracy AI cancer. Objective This paper presents a systematic review AI-based screening, which needed clarify provide evidence technology Methods A search was conducted across PubMed, Embase, Cochrane Library, Web Science, Scopus databases include studies published English, evaluated performance screening. total 2 independent reviewers screened titles abstracts, quality selected assessed using Quality Assessment Diagnostic Accuracy Studies—2 (QUADAS-2) tool. The certainty test Grading Recommendations Assessment, Development, Evaluation (GRADE) system. Results 13 were included, hierarchical summary receiver operating characteristic model used showed that overall sensitivity 86% (95% CI 79%-90%) specificity 92% 87%-95%). Subgroup analysis revealed similar results type, study region, publication year, low. Conclusions can effectively detect patients with cancer, large-scale population future further Trial Registration PROSPERO CRD42024519835; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024519835

Language: Английский

---