Search

Liquid biopsy demonstrates excellent potential in patient management by providing a minimally invasive and cost‐effective approach to detecting and monitoring cancer, even at its early stages. Due to the complexity of liquid biopsy data, machine‐learning techniques are increasingly gaining attention in sample analysis, especially for multidimensional data such as RNA expression profiles. Yet, there is no agreement in the community on which methods are the most effective or how to process the data. To circumvent this, we performed a large‐scale study using various machine‐learning techniques. First, we took a closer look at existing datasets and filtered out some patients to assert data collection quality. The final data collection included platelet RNA samples acquired from 1397 cancer patients (17 types of cancer) and 354 asymptomatic, presumed healthy, donors. Then, we assessed an array of different machine‐learning models and techniques (e.g., feature selection of RNA transcripts) in pan‐cancer detection and multiclass classification. Our results show that simple logistic regression performs the best, reaching a 68% cancer detection rate at a 99% specificity level, and multiclass classification accuracy of 79.38% when distinguishing between five cancer types. In summary, by revisiting classical machine‐learning models, we have exceeded the previously used method by 5% and 9.65% in cancer detection and multiclass classification, respectively. To ease further research, we open‐source our code and data processing pipelines (https://gitlab.com/jopekmaksym/improving‐platelet‐rna‐based‐diagnostics), which we hope will serve the community as a strong baseline. [ABSTRACT FROM AUTHOR]

Liquid biopsies offer a minimally invasive sample collection, outperforming traditional biopsies employed for cancer evaluation. The widely used material is blood, which is the source of tumor-educated platelets. Here, we developed the imPlatelet classifier, which converts RNA-sequenced platelet data into images in which each pixel corresponds to the expression level of a certain gene. Biological knowledge from the Kyoto Encyclopedia of Genes and Genomes was also implemented to improve accuracy. Images obtained from samples can then be compared against standard images for specific cancers to determine a diagnosis. We tested imPlatelet on a cohort of 401 non-small cell lung cancer patients, 62 sarcoma patients, and 28 ovarian cancer patients. imPlatelet provided excellent discrimination between lung cancer cases and healthy controls, with accuracy equal to 1 in the independent dataset. When discriminating between noncancer cases and sarcoma or ovarian cancer patients, accuracy equaled 0.91 or 0.95, respectively, in the independent datasets. According to our knowledge, this is the first study implementing an image-based deep-learning approach combined with biological knowledge to classify human samples. The performance of imPlatelet considerably exceeds previously published methods and our own alternative attempts of sample discrimination. We show that the deep-learning image-based classifier accurately identifies cancer, even when a limited number of samples are available. [ABSTRACT FROM AUTHOR]