Youness Arjoune

Background: Primary care providers (PCPs) are not successful in accurately identifying Still's murmur with no available clinical tools to aid in the process. Existing deep learning (DL) methods primaryly focused on adult pathological murmurs or murmur detection, lacking dedicated approaches for Still's murmur. Furthermore, the absence of a specialized database hampers the development and validation of AI models for pediatric populations. Objective: To collect a pediatric dataset and evaluate DL methods in identifying Still's murmur in pediatric patients from phonocardiograms (PCGs) acquired using the StethAid digital auscultation platform. Methods: We collected a large pediatric dataset (>2 hours of heart sound data) from four pediatric medical centers using StethAid, in addition to our previous Littmann 4100 dataset, which consisted of 1450 heart sound recordings. The StethAid dataset comprises 527 PCGs with diagnoses from four expert pediatric cardiologists. This dataset was used to train convolution- and pure attention-based models, which are our designed architecture SQ-NET, ResNet18, audio spectrogram transformer (AST), data-efficient transformer (DeiT), shifted windows (Swin) transformer, self-distillation with no labels (DINO). These methods were trained on segmented heart sounds except AST. Results: DL methods achieved a sensitivity of 90.7%-100%, specificity of 75%-98.2%, and accuracy of 91.3%-98.5%. These results reflect post-hoc comparisons and highlight the potential of DL models to distinguish Still's murmur from other innocent and pathological murmurs. Pretraining on domain specific Littmann 4100 dataset improved models' performance by at least 2% over training from scratch. Conclusion: Both ConvNets and transformers enjoyed high performance in Still's murmur identification. Significance: The proposed Still's murmur solution can assist PCPs in recognizing Still's murmur from other heart murmurs, potentially reducing unnecessary specialty referrals and the use of echocardiography. Additionally, StethAid could enhance PCPs' auscultation skills for Still's murmur by providing instant feedback, a feature often lacking in current clinical workflows.

Background Trackable ultrasound probes facilitate ultrasound-guided procedures, allowing real-time fusion of augmented ultrasound images and live video streams. The integration aids surgeons in accurately locating lesions within organs, and this could only be achieved through a precise registration between the ultrasound probe and the ultrasound image. Currently, calibration and registration processes are often manual, labor-intensive, time-consuming, and suboptimal. Technologists manually manipulate a stylus, moving it through various poses within the ultrasound probe’s imaging plane to detect its tip in the ultrasound image. This paper addresses this challenge by proposing a novel automated calibration approach for trackable ultrasound probes. Methods We utilized a robotic manipulator (KUKA LBR iiwa 7) to execute stylus movements, eliminating the cumbersome manual positioning of the probe. We incorporated a 6-degree-of-freedom electromagnetic tracker into the ultrasound probe to enable real-time pose and orientation tracking. Also, we developed a feature detection algorithm to effectively identify in plane stylus tip coordinates from recorded ultrasound feeds, facilitating automatic selection of calibration correspondences. Results The proposed system performed comparably to manual ultrasound feature segmentation, yielding a mean re-projection error of 0.38 mm compared to a manual landmark selection error of 0.34 mm. We also achieved an image plane reconstruction of 0.80 deg with manual segmentation and 0.20 deg with automatic segmentation. Conclusion The proposed system allowed for fully automated calibration while maintaining the same level of accuracy as the state-of-the-art methods. It streamlines the process of using a trackable US probe by simplifying recalibration after sterilization when the electromagnetic tracker is externally attached and is required to be disassembled for cleaning and sterilization, or as a part of out-of-factory calibration of US probe with embedded trackers where probes are in mass production.

Heart sound segmentation has been shown to improve the performance of artificial intelligence (AI)-based auscultation decision support systems increasingly viewed as a solution to compensate for eroding auscultatory skills and the associated subjectivity. Various segmentation approaches with demonstrated performance can be utilized for this task, but their robustness can suffer in the presence of noise. A noise-robust heart sound segmentation algorithm was developed and its accuracy was tested using two datasets: the CirCor DigiScope Phonocardiogram dataset and an in-house dataset – a heart murmur library collected at the Children’s National Hospital (CNH). On the CirCor dataset, our segmentation algorithm marked the boundaries of the primary heart sounds S1 and S2 with an accuracy of 0.28 ms and 0.29 ms, respectively, and correctly identified the actual positive segments with a sensitivity of 97.44%. The algorithm also executed four times faster than a logistic regression hidden semi-Markov model. On the CNH dataset, the algorithm succeeded in 87.4% cases, achieving a 6% increase in segmentation success rate demonstrated by our original Shannon energy-based algorithm. Accurate heart sound segmentation is critical to supporting and accelerating AI research in cardiovascular diseases. The proposed algorithm increases the robustness of heart sound segmentation to noise and viability for clinical use.

(1) Background: Mastery of auscultation can be challenging for many healthcare providers. Artificial intelligence (AI)-powered digital support is emerging as an aid to assist with the interpretation of auscultated sounds. A few AI-augmented digital stethoscopes exist but none are dedicated to pediatrics. Our goal was to develop a digital auscultation platform for pediatric medicine. (2) Methods: We developed StethAid—a digital platform for artificial intelligence-assisted auscultation and telehealth in pediatrics—that consists of a wireless digital stethoscope, mobile applications, customized patient-provider portals, and deep learning algorithms. To validate the StethAid platform, we characterized our stethoscope and used the platform in two clinical applications: (1) Still’s murmur identification and (2) wheeze detection. The platform has been deployed in four children’s medical centers to build the first and largest pediatric cardiopulmonary datasets, to our knowledge. We have trained and tested deep-learning models using these datasets. (3) Results: The frequency response of the StethAid stethoscope was comparable to those of the commercially available Eko Core, Thinklabs One, and Littman 3200 stethoscopes. The labels provided by our expert physician offline were in concordance with the labels of providers at the bedside using their acoustic stethoscopes for 79.3% of lungs cases and 98.3% of heart cases. Our deep learning algorithms achieved high sensitivity and specificity for both Still’s murmur identification (sensitivity of 91.9% and specificity of 92.6%) and wheeze detection (sensitivity of 83.7% and specificity of 84.4%). (4) Conclusions: Our team has created a technically and clinically validated pediatric digital AI-enabled auscultation platform. Use of our platform could improve efficacy and efficiency of clinical care for pediatric patients, reduce parental anxiety, and result in cost savings.

Heat loss quantification (HLQ) is an essential step in improving a building’s thermal performance and optimizing its energy usage. While this problem is well-studied in the literature, most of the existing studies are either qualitative or minimally driven quantitative studies that rely on localized building envelope points and are, thus, not suitable for automated solutions in energy audit applications. This research work is an attempt to fill this gap of knowledge by utilizing intensive thermal data (on the order of 100,000 plus images) and constitutes a relatively new area of analysis in energy audit applications. Specifically, we demonstrate a novel process using deep-learning methods to segment more than 100,000 thermal images collected from an unmanned aerial system (UAS). To quantify the heat loss for a building envelope, multiple stages of computations need to be performed: object detection (using Mask-RCNN/Faster R-CNN), estimating the surface temperature (using two clustering methods), and finally calculating the overall heat transfer coefficient (e.g., the U-value). The proposed model was applied to eleven academic campuses across the state of North Dakota. The preliminary findings indicate that Mask R-CNN outperformed other instance segmentation models with an mIOU of 73% for facades, 55% for windows, 67% for roofs, 24% for doors, and 11% for HVACs. Two clustering methods, namely K-means and threshold-based clustering (TBC), were deployed to estimate surface temperatures with TBC providing consistent estimates across all times of the day over K-means. Our analysis demonstrated that thermal efficiency not only depended on the accurate acquisition of thermal images but also relied on other factors, such as the building geometry and seasonal weather parameters, such as the outside/inside building temperatures, wind, time of day, and indoor heating/cooling conditions. Finally, the resultant U-values of various building envelopes were compared with recommendations from the American Society of Heating, Refrigerating, and Air-conditioning Engineers (ASHRAE) building standards.

Cognitive radio technology has the potential to address the shortage of available radio spectrum by enabling dynamic spectrum access. Since its introduction, researchers have been working on enabling this innovative technology in managing the radio spectrum. As a result, this research field has been progressing at a rapid pace and significant advances have been made. To help researchers stay abreast of these advances, surveys and tutorial papers are strongly needed. Therefore, in this paper, we aimed to provide an in-depth survey on the most recent advances in spectrum sensing, covering its development from its inception to its current state and beyond. In addition, we highlight the efficiency and limitations of both narrowband and wideband spectrum sensing techniques as well as the challenges involved in their implementation. TV white spaces are also discussed in this paper as the first real application of cognitive radio. Last but by no means least, we discuss future research directions. This survey paper was designed in a way to help new researchers in the field to become familiar with the concepts of spectrum sensing, compressive sensing, and machine learning, all of which are the enabling technologies of the future networks, yet to help researchers further improve the efficiently of spectrum sensing.

Cognitive radio technology addresses the problem of spectrum scarcity by allowing secondary users to use the vacant spectrum bands without causing interference to the primary users. However, several attacks could disturb the normal functioning of the cognitive radio network. Primary user emulation attacks are one of the most severe attacks in which a malicious user emulates the primary user signal characteristics to either prevent other legitimate secondary users from accessing the idle channels or causing harmful interference to the primary users. There are several proposed approaches to detect the primary user emulation attackers. However, most of these techniques assume that the primary user location is fixed, which does not make them valid when the primary user is mobile. In this paper, we propose a new approach based on the Kalman filter framework for detecting the primary user emulation attacks with a non-stationary primary user. Several experiments have been conducted and the advantages of the proposed approach are demonstrated through the simulation results.