Search

Glioblastoma multiforme (GBM) is a WHO grade IV tumor and its heterogeneity pushes oncologists to focus on more personalized treatments for individual patients. This challenge is aided by radiomics, which involves the extraction of valuable features from Magnetic Resonance images, the most common non invasive method to identify Glioblastoma multiforme. Analyzing radiomic features through machine learning has been a robust way to predict overall survival (OS) days of GBM patients which significantly affects the personalized treatments and wellbeing.Despite the promise of machine learning for predicting GBM patient overall survival (OS), the limited interpretability of complex models hinders their clinical adoption and the translation of valuable insights gained from radiomic features into actionable treatment plans. Therefore, this study is focused on designing a highly accurate Random Forest (RF) model to predict the OS days of GBM patients from radiomic features and utilize SHapely Additive exPlanation (SHAP) values to provide a comprehensive interpritable analysis of the model, enabling clinicians to understand the role of radiomic features in patient survival prediction. The designed RF model achieved a validation accuracy of 62.5% surpassing previous work done on similar studies. The interpritable analysis revealed critical insights for personalized treatment, including the identification of outlier patients with unique radiomic features, risk inflection points for key features that may guide treatment decisions and pairwise interactions of radiomic features that may be novel biomarkers that affect the OS of GBM patients. This study’s findings pave the way for the development of more clinically-usable machine learning models for personalized GBM treatment planning, ultimately improving patient outcomes.

The term “Swarm intelligence” outlines a broad scope, which is generally defined as the collective behaviour of many individuals towards a certain task, each operating autonomously in a decentralised manner. Swarms are inspired by the type of biological behaviours of animals and this technology involves decentralised local control and communication, which ensures the problem-solving process is more efficient through modification and infusion into swarm robotic technology. Swarm robotics related experiments and applications are relatively expensive compared to other types of robotics experiments. The most common solution is to use computer based virtual simulations and accordingly, this study introduces a reliable and cost-effective hybrid solution using Mixed Reality (MR) technologies to perform swarm experiments on real robots and virtually deployed robots simultaneously. MR bridges the gap between the virtual and physical realms through Augmented Reality (AR) and Augmented Virtuality (AV) and addressing the weaknesses of simulators and providing a more accurate representation of real-world performance. This approach provides a platform for researchers to explore swarm robotics with enhanced realism and interactivity and conduct experiments that mimic real-world conditions. To validate this method we conducted experiments with both physical and virtual robots in near real-time in a mixed reality environment. In addition, an open-source, distributed, and modular mixed reality simulation framework was implemented with support libraries and applications. Our work not only demonstrates the potential of MR in swarm robotics but also emphasizes its value in advancing the understanding and application of swarm intelligence principles in practical contexts.

The authentication of digital images poses a significant challenge due to the wide range of image forgery techniques employed, with one notable example being a copy-move forgery. This form of forgery involves duplicating and relocating segments of an image within the same image, often accompanied by geometric transformations to deceive viewers into perceiving the forged image as authentic. Furthermore, additional processing techniques like scaling, rotation, JPEG compression, and the application of Additive White Gaussian Noise (AWGN) are frequently employed to further obscure any traces of forgery, making the detection and verification process even more complex. This paper presents a novel approach for detecting copy-move forgery in digital images using the self-supervised image keypoint detector, SuperPoint. Our approach leverages the advanced capabilities of SuperPoint, which combines keypoint detection and descriptor extraction, to identify and localize copy-move forgery accurately. One important aspect of our approach is its ability to handle images with different textures, including smooth and self-similar structural images. The proposed approach is able to produce stable results in images with various attacks, making it a functional and reliable tool for detecting copy-move forgery in a diverse range of forged images. Comparative analysis with existing forgery detection methods shows the superior performance of our proposed approach. Furthermore, the computational efficiency of our algorithm enables real-time forgery detection. Our approach using SuperPoint offers an effective solution for detecting copy-move forgery in digital images, making it valuable for image forensics and authenticity

The paper manufacturing process is a large-scale process in which the control performance is required to be optimal in the machine direction (MD) and cross direction (CD). This work focuses on CD control, which is more complicated than MD control. Uncertainties in the process arise from different sources, so there is a requirement for better modeling techniques. The control of paper profile properties, namely basis weight, moisture, and caliper, is done with the help of actuators that are distributed spatially. Basis weight is taken as the profile parameter in this work. The work aims to improve robust performance and simplify the control design problem. Here, $\mu $ synthesis methods are used for controller design, as this method provides guaranteed robust performance. To simplify the design problem two-dimensional modeling is explained explicitly. The conventional methods of $\mu $ analysis are extended to spatial coordinates. The performance of the $\mu $ synthesis designed method is compared with the open loop shaping method. Control performance is judged in terms of sensitivity, stability margin, $\sigma $ performance, and robust performance. The effectiveness of the developed $\mu $ synthesis designed method is demonstrated using the industrial example from the pulp and paper industry.