Counting Mitoses With Digital Pathology in Breast Phyllodes Tumors

Artificial intelligence's impact on breast cancer pathology: a literature review

This review discusses the profound impact of artificial intelligence (AI) on breast cancer (BC) diagnosis and management within the field of pathology. It examines the various applications of AI across diverse aspects of BC pathology, highlighting key findings from multiple studies. Integrating AI into routine pathology practice stands to improve diagnostic accuracy, thereby contributing to reducing avoidable errors. Additionally, AI has excelled in identifying invasive breast tumors and lymph node metastasis through its capacity to process large whole-slide images adeptly. Adaptive sampling techniques and powerful convolutional neural networks mark these achievements. The evaluation of hormonal status, which is imperative for BC treatment choices, has also been enhanced by AI quantitative analysis, aiding interobserver concordance and reliability. Breast cancer grading and mitotic count evaluation also benefit from AI intervention. AI-based frameworks effectively classify breast carcinomas, even for moderately graded cases that traditional methods struggle with. Moreover, AI-assisted mitotic figures quantification surpasses manual counting in precision and sensitivity, fostering improved prognosis. The assessment of tumor-infiltrating lymphocytes in triple-negative breast cancer using AI yields insights into patient survival prognosis. Furthermore, AI-powered predictions of neoadjuvant chemotherapy response demonstrate potential for streamlining treatment strategies. Addressing limitations, such as preanalytical variables, annotation demands, and differentiation challenges, is pivotal for realizing AI's full potential in BC pathology. Despite the existing hurdles, AI's multifaceted contributions to BC pathology hold great promise, providing enhanced accuracy, efficiency, and standardization. Continued research and innovation are crucial for overcoming obstacles and fully harnessing AI's transformative capabilities in breast cancer diagnosis and assessment.

Artificial intelligence modelling in differentiating core biopsies of fibroadenoma from phyllodes tumor

Breast fibroepithelial lesions (FEL) are biphasic tumors which consist of benign fibroadenomas (FAs) and the rarer phyllodes tumors (PTs). FAs and PTs have overlapping features, but have different clinical management, which makes correct core biopsy diagnosis important. This study used whole-slide images (WSIs) of 187 FA and 100 PT core biopsies, to investigate the potential role of artificial intelligence (AI) in FEL diagnosis. A total of 9228 FA patches and 6443 PT patches was generated from WSIs of the training subset, with each patch being 224 × 224 pixel in size. Our model employed a two-stage architecture comprising a convolutional neural network (CNN) component for feature extraction from the patches, and a recurrent neural network (RNN) component for whole-slide classification using activation values from the global average pooling layer in the CNN model. It achieved an overall slide-level accuracy of 87.5%, with accuracies of 80% and 95% for FA and PT slides respectively. This affirms the potential role of AI in diagnostic discrimination between FA and PT on core biopsies which may be further refined for use in routine practice.

Counting mitoses: SI(ze) matters!

Mitoses are often assessed by pathologists to assist the diagnosis of cancer, and to grade malignancy, informing prognosis. Historically, this has been done by expressing the number of mitoses per n high power fields (HPFs), ignoring the fact that microscope fields may differ substantially, even at the same high power (×400) magnification. Despite a requirement to define HPF size in scientific papers, many authors fail to address this issue adequately. The problem is compounded by the switch to digital pathology systems, where ×400 equivalent fields are rectangular and also vary in the area displayed. The potential for error is considerable, and at times this may affect patient care. This is easily solved by the use of standardized international (SI) units. We, therefore, recommend that features such as mitoses are always counted per mm2, with an indication of the area to be counted and the method used (usually "hotspot" or "average") to obtain the results.

Performance and efficiency of machine learning algorithms for analyzing rectangular biomedical data

Most biomedical datasets, including those of 'omics, population studies, and surveys, are rectangular in shape and have few missing data. Recently, their sample sizes have grown significantly. Rigorous analyses on these large datasets demand considerably more efficient and more accurate algorithms. Machine learning (ML) algorithms have been used to classify outcomes in biomedical datasets, including random forests (RF), decision tree (DT), artificial neural networks (ANN), and support vector machine (SVM). However, their performance and efficiency in classifying multi-category outcomes of rectangular data are poorly understood. Therefore, we compared these metrics among the 4 ML algorithms. As an example, we created a large rectangular dataset using the female breast cancers in the surveillance, epidemiology, and end results-18 database, which were diagnosed in 2004 and followed up until December 2016. The outcome was the five-category cause of death, namely alive, non-breast cancer, breast cancer, cardiovascular disease, and other cause. We analyzed the 54 dichotomized features from ~45,000 patients using MatLab (version 2018a) and the tenfold cross-validation approach. The accuracy in classifying five-category cause of death with DT, RF, ANN, and SVM was 69.21%, 70.23%, 70.16%, and 69.06%, respectively, which was higher than the accuracy of 68.12% with multinomial logistic regression. Based on the features' information entropy, we optimized dimension reduction (i.e., reduce the number of features in models). We found 32 or more features were required to maintain similar accuracy, while the running time decreased from 55.57 s for 54 features to 25.99 s for 32 features in RF, from 12.92 s to 10.48 s in ANN, and from 175.50 s to 67.81 s in SVM. In summary, we here show that RF, DT, ANN, and SVM had similar accuracy for classifying multi-category outcomes in this large rectangular dataset. Dimension reduction based on information gain will increase the model's efficiency while maintaining classification accuracy.