Multicellular 3D tumor models are becoming a powerful tool for testing of novel drug products and personalized anticancer therapy. Tumor spheroids, a commonly used 3D multicellular tumor model, more closely reproduce the tumor microenvironment than conventional 2D cell cultures. It should be noted that spheroids can be produced using different techniques, which can be subdivided into scaffold-free (SF) and scaffold-based (SB) methods. However, it remains unclear, to what extent spheroid properties depend on the method of their generation. In this study, we aimed to carry out a head-to-head comparison of drug sensitivity and molecular expression profile in SF and SB spheroids along with a monolayer (2D) cell culture. Here, we produced non-small cell lung cancer (NSCLC) spheroids based on human lung adenocarcinoma cell line A549. Drug sensitivity analysis of the tested cell cultures to five different chemotherapeutics resulted in IC₅₀ (A549-SB) > IC₅₀ (A549-SF) > IC₅₀ (A549-2D) trend. It was found that SF and SB A549 spheroids displayed elevated expression levels of epithelial-to-mesenchymal transition (EMT) markers and proteins associated with drug resistance compared with the monolayer A549 cell culture. Enhanced drug resistance of A549-SB spheroids can be a result of larger diameters and elevated deposition of extracellular matrix (ECM) that impairs drug penetration into spheroids. Thus, the choice of the spheroid production method can influence the properties of the generated 3D cell culture and their drug resistance. This fact should be considered for correct interpretation of drug testing results.

The analysis of the microvasculature and the assessment of angiogenesis have significant prognostic value in various diseases, including cancer. The search for invasion into the blood and lymphatic vessels and the assessment of angiogenesis are important aspects of oncological diagnosis. These features determine the prognosis and aggressiveness of the tumor. Traditional manual evaluation methods are time consuming and subject to inter-observer variability. Blood vessel detection is a perfect task for artificial intelligence, which is capable of rapid analyzing thousands of tissue structures in whole slide images. The development of computer vision solutions requires the segmentation of tissue regions, the extraction of features and the training of machine learning models. In this review, we focus on the methodologies employed by researchers to identify blood vessels and vascular invasion across a range of tumor localizations, including breast, lung, colon, brain, renal, pancreatic, gastric and oral cavity cancers. Contemporary models herald a new era of computational pathology in morphological diagnostics.

Invasion and metastasis are important hallmarks of lung cancer, and affect patients’ survival. Early diagnostics of metastatic potential are important for treatment management. Recent findings suggest that the transition to an invasive phenotype causes changes in the expression of 700–800 genes. In this context, the biomarkers restricted to the specific type of cancer, like lung cancer, are often overlooked. Some well-known protein biomarkers correlate with the progression of the disease and the immunogenicity of the tumor. Most of these biomarkers are not exclusive to lung cancer because of their significant role in tumorigenesis. The dysregulation of others does not necessarily indicate cell invasiveness, as they play an active role in cell division. Clinical studies of lung cancer use protein biomarkers to assess the invasiveness of cancer cells for therapeutic purposes. However, there is still a need to discover new biomarkers for lung cancer. In the future, minimally invasive techniques, such as blood or saliva analyses, may be sufficient for this purpose. Many researchers suggest unconventional biomarkers, like circulating nucleic acids, exosomal proteins, and autoantibodies. This review paper aims to discuss the advantages and limitations of protein biomarkers of invasiveness in lung cancer, to assess their prognostic value, and propose novel biomarker candidates.

Mathematical models of non-small-cell lung cancer are powerful tools that use clinical and experimental data to describe various aspects of tumorigenesis. The developed algorithms capture phenotypic changes in the tumor and predict changes in tumor behavior, drug resistance, and clinical outcomes of anti-cancer therapy. The aim of this study was to propose a mathematical model that predicts the changes in the cellular composition of patient-derived tumor organoids over time with a perspective of translation of these results to the parental tumor, and therefore to possible clinical course and outcomes for the patient. Using the data on specific biomarkers of cancer cells (PD-L1), tumor-associated macrophages (CD206), natural killer cells (CD8), and fibroblasts (αSMA) as input, we proposed a model that accurately predicts the cellular composition of patient-derived tumor organoids at a desired time point. Combining the obtained results with “omics” approaches will improve our understanding of the nature of non-small-cell lung cancer. Moreover, their implementation into clinical practice will facilitate a decision-making process on treatment strategy and develop a new personalized approach in anti-cancer therapy.

Lymphovascular invasion (LVI) in lung cancer is a significant prognostic factor that influences treatment and outcomes, yet its reliable detection is challenging due to interobserver variability. This study aims to develop a deep learning model for LVI detection using whole slide images (WSIs) and evaluate its effectiveness within a pathologist's information system. Experienced pathologists annotated blood vessels and invading tumor cells in 162 WSIs of non-mucinous lung adenocarcinoma sourced from two external and one internal datasets. Two models were trained to segment vessels and identify images with LVI features. DeepLabV3+ model achieved an Intersection-over-Union of 0.8840 and an area under the receiver operating characteristic curve (AUC-ROC) of 0.9869 in vessel segmentation. For LVI classification, the ensemble model achieved a F1-score of 0.9683 and an AUC-ROC of 0.9987. The model demonstrated robustness and was unaffected by variations in staining and image quality. The pilot study showed that pathologists' evaluation time for LVI detecting decreased by an average of 16.95%, and by 21.5% in "hard cases". The model facilitated consistent diagnostic assessments, suggesting potential for broader applications in detecting pathological changes in blood vessels and other lung pathologies.

В настоящей работе представлены результаты применения сверточной нейронной сети для диагностики гипертрофий левых отделов сердца посредством анализа электрокардиограмм (ЭКГ) в 12 стандартных отведениях. В ходе исследования был собран и обработан новый уникальный набор данных, содержащий 64 тысячи записей ЭКГ. На основе сопутствующих записям заключений были сформированы метки принадлежности к двум рассматриваемым классам: гипертрофия левого желудочка и гипертрофия левого предсердия. Набор сигналов и выделенные метки были использованы для обучения глубокой сверточной нейронной сети с остаточными блоками, получившаяся модель способна детектировать гипертрофию левого желудочка с качеством по F-мере свыше 0.82 и гипертрофию левого предсердия с качеством свыше 0.78. Кроме того, был осуществлен поиск оптимальной архитектуры нейросети, произведена экспериментальная оценка эффекта от включения в модель метаданных пациентов и предобработки сигнала, а также сделан сравнительный анализ трудности детектирования гипертрофий левых отделов по отношению к двум другим часто встречающимся нарушениям сердечной активности – мерцательной аритмии и блокады левой ножки пучка Гиса.