Using deep learning to detect digitally encoded DNA trigger for Trojan malware in Bio-Cyber attacks.

This article uses Deep Learning technologies to safeguard DNA sequencing against Bio-Cyber attacks. We consider a hybrid attack scenario where the payload is encoded into a DNA sequence to activate a Trojan malware implanted in a software tool used in the sequencing pipeline in order to allow the perpetrators to gain control over the resources used in that pipeline during sequence analysis. The scenario considered in the paper is based on perpetrators submitting synthetically engineered DNA samples that contain digitally encoded IP address and port number of the perpetrator's machine in the DNA. Genetic analysis of the sample's DNA will decode the address that is used by the software Trojan malware to activate and trigger a remote connection. This approach can open up to multiple perpetrators to create connections to hijack the DNA sequencing pipeline. As a way of hiding the data, the perpetrators can avoid detection by encoding the address to maximise similarity with genuine DNAs, which we showed previously. However, in this paper we show how Deep Learning can be used to successfully detect and identify the trigger encoded data, in order to protect a DNA sequencing pipeline from Trojan attacks. The result shows nearly up to 100% accuracy in detection in such a novel Trojan attack scenario even after applying fragmentation encryption and steganography on the encoded trigger data. In addition, feasibility of designing and synthesizing encoded DNA for such Trojan payloads is validated by a wet lab experiment.

MACHINE LEARNING AND RADIOMICS IDENTIFY NOVEL BIOMARKERS OF BONE LOSS

Purpose: Altered bone turnover is a factor in many diseases including osteoarthritis (OA), osteoporosis, inflammation, and viral infection. The absence of obvious symptoms and insufficiently sensitive biomarkers in the early stages of bone loss limits early diagnosis and treatment. Therefore, it is urgent to identify novel, more sensitive, and easy-to-detect biomarkers which can be used in the diagnosis and prognosis of bone health. Our previous data using standard micro-computed tomography (μCT) measurements showed that SARS-CoV-2 infection in mice significantly decreased trabecular bone volume at the lumbar spine, suggesting that decreased bone mass, increased fracture risk, and OA may be underappreciated long-haul comorbidities for COVID patients. In this study, we applied integrated state-of-the-art radiomics and machine learning models to identify more sensitive image-based biomarkers of SARS-CoV-2-induced bone loss from μCT images. These radiomic biomarkers can potentially provide a non-invasive way of quantifying and monitoring systemic bone loss and evaluating treatment efficacy in both research and clinical practices. Methods: All animal use was performed with approval of the Institutional Animal Care and Use Committee. To quantify SARS-CoV-2-induced bone loss, 6-week-old transgenic mice (16 male, 16 female) expressing humanized ACE2 receptors were inoculated with a 2020 strain of SARS-CoV-2 or phosphate-buffered saline (Control) [Fig. A]. Viral infection was confirmed by detection of infectious SARS-CoV-2 in throat swabs and histological identification of SARS-CoV-2 labeled cells. At 6-14 days post-infection, lumbar vertebral bodies (L5) were scanned with μCT (μCT 35, SCANCO Medical AG;6 μm nominal voxel size). The open-source research platform 3D Slicer v2020 with a built-in Python console v3.8 was used for medical image computing and fully automated segmentation of cortical and trabecular bone. Standard μCT assessment of bone microstructure was performed. Radiomic feature extraction and data processing were performed using python based PyRadiomics v3.0.1. A total of 120 radiographic features were extracted from the segmented images [Fig. B]. Principle component analysis (PCA) for feature selection, a support vector machine learning (SVML) predictive model for classification, holdback method for model validation, and all statistical analyses (significance at p<0.05) were performed using JMP Pro v15 (SAS). Results: Using standard μCT methods, SARS-CoV-2 infection significantly reduced the bone volume fraction (BV/TV) by 10 and 10.5% (p= 0.04) and trabecular thickness (Tb.Th) by 8 and 9% (p= 0.02) in male and female mice, respectively, compared to PBS control mice [Fig. C]. Radiomics detected a 20-fold greater magnitude in change over standard methods. SARS-CoV-2 infection significantly changed radiographic parameters with the largest change being a 300% increase in the second-order parameter: cluster shade [Fig. D]. The 45 radiomic features comprising the first 3 principal components were selected for inclusion in the SVML model. The SVML Model (radial basis function kernel;cost = 4.8;gamma = 0.46) produced an area under the receiver operating characteristic curve (AUC) of 1.0 which reflects a perfectly accurate test [Fig. E]. Conclusions: SARS-CoV-2 infection of humanized ACE2 expressing mice caused significant bone changes, suggesting that decreased bone mass, increased fracture risk, OA, and other musculoskeletal complications could be long-term comorbidities for people infected with COVID-19. We developed an open-source, fully automated segmentation and radiomics system to assess systemic bone loss using μCT images. When coupled with machine learning, this system was able to identify novel radiographic biomarkers of bone loss that better discriminate differences in bone microstructure between SARS-CoV-2 infected and control mice than standard bone morphometric indices. The high accuracy of the SVML model in classifying SARS-CoV-2 infected mice opens the possibility of translating these biom rkers to the clinical setting for early detection of skeletal changes associated with long-haul COVID. The methods presented here were demonstrated using SARS-CoV-2 as a model system and can also be adapted to other diseases associated with altered bone turnover. Development of machine-learning methods for radiomic applications is a crucial step toward clinically relevant radiomic biomarkers of bone health and provides a non-invasive way of quantifying and monitoring systemic bone loss and evaluating treatment efficacy. [Formula presented]