BACKGROUND:Colorectal cancer (CRC) is a leading cause of mortality worldwide, and early examination via colonoscopy is increasingly used to prevent CRC mortality. Recently, studies have attempted utilizing artificial intelligence for the classification of CRC. However, datasets were limited in these studies, and the limited number of findings resulting from these studies are not specific to cancerous or noncancerous findings. We present a full pipeline of ensemble deep learning approach to classify five diagnostic categories. Through this pipeline, whole slide images (WSIs) exhibiting low quality can be filtered out before being processed by the computer-aided diagnosis system. METHODS:A dataset of 18,922 CRC WSIs collected and labeled as non-tumor, hyperplastic polyp, adenoma, adenocarcinoma, and neuroendocrine tumor (NET, carcinoid). We developed two models: clustering-constrained attention multiple instance learning model was used to classify adenocarcinomas, adenomas, NET, and non-dysplastic lesion classes. EfficientNet then distinguished between non-tumor and hyperplastic polyp for WSIs classified as non-dysplastic lesion class in the first step. RESULTS:The proposed ensemble pipeline, which resembles the process of analysis from the pathologists, showed better performance even with multiple classes CRC. The micro-, macro-, and weighted-F1-scores were 86.57%, 83.83%, and 86.86%, respectively. Notably, the NET classification tasks scored an F1-score of 87.18%. CONCLUSION:The proposed ensemble model addresses common challenges in automated CRC diagnosis, offering reliable solutions with safeguards for exceptional cases. It enhances clinical practice by accurately classifying multiple CRC types. In conclusion, this method is effective where scanned colorectal WSIs can be of low quality for real-world diagnosis.

Background Brain metastasis is a common complication among patients with lung cancer, yet the underlying mechanisms remain unclear. In this study, we aimed to investigate the pathogenesis of brain metastasis in lung cancer. Methods We established highly colonizing metastatic lung cancer cells, A549-M2, through multiple implantations of A549 human lung cancer cells in the carotid artery of athymic nude mice. Results Compared to parental cells (M0), M2 cells demonstrated slower growth in culture plates and soft agar, as well as lower motility and higher adhesion, key characteristics of mesenchymal–epithelial transition (MET). Further analysis revealed that M2 cells exhibited decreased expression of epithelial–mesenchymal transition markers, including ZEB1 and Vimentin. M2 cells also demonstrated reduced invasiveness in co-culture systems. RNA sequencing and gene set enrichment analysis confirmed that M2 cells underwent MET. Intriguingly, depletion of Noggin, a BMP antagonist, was observed in M2 cells, and replenishment of Noggin restored suppressed migration and invasion of M2 cells. In addition, Noggin knockdown in control M0 cells promoted cell attachment and suppressed cell migration, suggesting that Noggin reduction during brain colonization causes inhibition of migration and invasion of metastatic lung cancer cells. Conclusions Our results suggest that lung cancer cells undergo MET and lose their motility and invasiveness during brain metastatic colonization, which is dependent on Noggin.

Automatic pattern recognition using deep learning techniques has become increasingly important. Unfortunately, due to limited system memory, general preprocessing methods for high-resolution images in the spatial domain can lose important data information such as high-frequency information and the region of interest. To overcome these limitations, we propose an image segmentation approach in the compressed domain based on principal component analysis (PCA) and discrete wavelet transform (DWT). After inference for each tile using neural networks, a whole prediction image was reconstructed by wavelet weighted ensemble (WWE) based on inverse discrete wavelet transform (IDWT). The training and validation were performed using 351 colorectal biopsy specimens, which were pathologically confirmed by two pathologists. For 39 test datasets, the average Dice score, the pixel accuracy, and the Jaccard score were 0.804 ± 0.125, 0.957 ± 0.025, and 0.690 ± 0.174, respectively. We can train the networks for the high-resolution image with the large region of interest compared to the result in the low-resolution and the small region of interest in the spatial domain. The average Dice score, pixel accuracy, and Jaccard score are significantly increased by 2.7%, 0.9%, and 2.7%, respectively. We believe that our approach has great potential for accurate diagnosis.

Little is known about the underlying metabolic alterations of gliomas. The objective of this study was to analyze metabolomic profiles of gliomas diagnosed according to revised WHO classification to demonstrate metabolic signatures beyond isocitrate dehydrogenase (IDH) 1/2 mutation. 1H NMR spectroscopy of tumor extracts was performed to analyze brain tumor metabolism. We detected 46 metabolites including 2-hydroxyglutarate from human brain tumors. Metabolic profiles obtained were analyzed using multivariate analysis and MetaboAnalyst 3.0, a pathway analysis tool. We found that lactate, glutamate, alanine, glutamine, 2-hydroxglutarate, serine, O-phosphocholine, glycine, glycerol, myo-inositol, aspartate, leucine, threonine, creatine, and valine had top-ranked VIP scores in metabolic pathway analyses of glioma. Major metabolism pathways perturbed in glioma included alanine/aspartate/glutamate metabolism, glycine/serine/threonine metabolism, pyruvate metabolism, taurine/hypotaurine metabolism, and d-glutamine/d-glutamate metabolism. Altered metabolites were defined between low-grade and high-grade gliomas. We identified metabolomics signatures of gliomas associated with 2-hydroxglutarate and glioma grade. Metabolic approach may lead to metabolomic cluster-precision strategy and development of metabolic anti-glioma therapy in the future.