Icon Legend

This session is not in your schedule.

This session is in your schedule. Click again to remove it.

Device Technologies and Biomedical Robotics

Device Technologies and Biomedical Robotics - Poster Session E

Poster W3 - Preliminary Assessment of Intracranial Aneurysm Auscultation via Acoustic Signal Analysis in In Vitro Models

Saturday, October 26, 2024

10:00 AM - 11:00 AM EST

Location: Exhibit Hall E, F & G

Presenting Author(s)

Sinan Altamimi

PhD Student
University at Buffalo, The State University of New York
Williamsville, New York, United States

Co-Author(s)

AR

Ahmad Rahmatpour

PhD Student
University at Buffalo, The State University of New York, United States
FS

Filip Stefanovic, PhD

Assistant Professor
University at Buffalo, The State University of New York, United States
Ciprian N. Ionita, PhD

Associate Professor
University at Buffalo, The State University of New York, New York, United States
AT

Albert Titus, PhD

Professor
University at Buffalo, The State University of New York
Buffalo, New York, United States

Introduction: Advancements in wearable audio technology offer promising avenues for the noninvasive detection of intracranial aneurysms prior to symptom manifestation. This research utilizes cutting-edge acoustic signal analysis integrated into wearable devices to preemptively identify hemodynamic disturbances that could indicate intracranial aneurysms. We hypothesize that these technologies can detect subtle changes in blood flow patterns early, potentially transforming intracranial aneurysm diagnostics into a proactive and patient-friendly approach. This study aims to demonstrate that wearable acoustic sensors, coupled with sophisticated data analysis, can accurately and efficiently predict the risk of intracranial aneurysms, enabling earlier clinical interventions and improving patient outcomes.

Materials and

Methods: To investigate the capability of acoustic signal analysis for detecting neurovascular anomalies, we developed a patient-specific neurovascular phantom replicating the right half of the Circle of Willis with an aneurysm in the MCA artery. It contains the internal carotid and the main aorta arteries 3D printed in an elastomer. The model is encased in a ballistic gel, which is heated, poured over the model, and solidified. The model includes inlet and outlets to allow for flow simulations using a Shelley CardioFlow pump system programmed to mimic patient-specific hemodynamic internal carotid waveforms. Acoustic data collection was facilitated through a stethoscope integrated with a Soundskrit Ladybug kit, which includes both the SKR0400 dipole/directional and omnidirectional MEMS microphones. This assembly was housed within a noise-canceling enclosure analogous to modern earphone designs. The data acquisition process was executed in three distinct phases using a Raspberry Pi 4 to control the experiment and Audacity software for recording. Figures 1-3 show the experimental setup and devices. A signal baseline was established by recording the ambient noise with the pump off. Subsequently, the system was activated to simulate both healthy and aneurysmal flow conditions using the patient-specific waveform. For each state, acoustic signals were captured. Signal processing was conducted using Praat software, which facilitated the spectrogram and intensity analysis, enabling us to inspect the frequency and amplitude variations over time. The objective was to discern distinctive acoustic patterns that could be indicative of intracranial aneurysms.

Results, Conclusions, and Discussions: Quantitative analysis through spectrogram and time-domain observations revealed significant distinctions between the aneurysm and healthy models. (See Figure 4). The analysis also showed marked amplitude fluctuations in the aneurysm model, aligning with hemodynamic events characteristic of intracranial aneurysms, which were subdued in the healthy model. Specifically, the spectrograms of the aneurysm model showed irregular, high-frequency peaks and significant amplitude changes, reflecting the disturbed hemodynamics. In contrast, the healthy model's spectrograms showed smooth, regular frequency bands with minimal variation in intensity, corresponding to undisturbed and uniform flow.

The preliminary results from our acoustic signal analysis using a patient-specific neurovascular phantom indicate that our prototype system allows us to differentiate between normal and aneurysmal hemodynamic patterns. The identification of specific high-frequency signatures and detailed intensity and pitch variations suggests that our approach holds promise. However, these findings primarily demonstrate the potential of our system and warrant further investigation and optimization. These insights are anticipated to guide the development of an AI-driven diagnostic tool, potentially capable of early and accurate identification of neurovascular diseases through acoustic pattern recognition.

Continued development will focus on enhancing hardware capabilities and integrating advanced sound analysis techniques, including machine learning algorithms, to improve diagnostic accuracy and reliability. The next phase of our research will aim to refine these technologies to ensure robustness and clinical applicability, moving towards a non-invasive, wearable device for early detection of neurovascular anomalies. This study supports the feasibility of our approach and underscores the necessity of continued research and development in this field.

Acknowledgements (Optional): The work on this project is funded by the UB Center and Translational Science Institute.
Thank you to Dr. Titus, Dr. Ionita, and Dr. Stefanovic for their guidance and mentorship.