Organizations: Division AI4Health, Department for Health Services Research, Faculty of Medicine and Health Sciences, Carl von Ossietzky Universität Oldenburg, 26129 Oldenburg, Germany · Fraunhofer IDMT, Institute Part HSA, 26129 Oldenburg, Germany · University Clinic for Visceral Surgery, Faculty of Medicine and Health Sciences, Carl von Ossietzky Universität Oldenburg, 26121 Oldenburg, Germany
Bowel sounds (BS) are typically momentary and have low amplitude, making them difficult to detect accurately through manual auscultation. This leads to significant variability in clinical assessment. Digital acoustic sensors allow the acquisition of high-quality BS and enable automated signal analysis, offering the potential to provide clinicians with both objective and quantitative feedback on bowel activity. This study presents an automated pipeline for bowel sound segmentation and classification using a wearable acoustic SonicGuard sensor. BS signals from 83 subjects were recorded using a SonicGuard sensor. Data from 40 subjects were manually annotated by clinical experts and used to train an automatic annotation algorithm, while the remaining subjects were used for further model evaluation. An energy-based event detection algorithm was developed to detect BS events. Detected sound segments were then classified into BS patterns using a pretrained Audio Spectrogram Transformer (AST) model. Model performance was evaluated separately for healthy individuals and patients. The best configuration used two specialized models, one trained on healthy subjects and one on patients, achieving (accuracy: 0.97, AUROC: 0.98) for healthy group and (accuracy: 0.96, AUROC: 0.98) for patient group. The auto-annotation method reduced manual labeling time by approximately 70%, and expert review showed that less than 12% of automatically detected segments required correction. The proposed automated segmentation and classification system enables quantitative assessment of bowel activity, providing clinicians with an objective diagnostic tool that may improve the diagnostic of gastrointestinal function and support the annotation of large-scale datasets.
Figures & tables
Figure 1: SonicGuard wearable acoustic sensing platform. Four sensors are placed on the abdominal quadrants (Right Upper Quadrant (RUQ), Left Upper Quadrant (LUQ), Right Lower Quadrant (RLQ), and Left Lower Quadrant (LLQ)) to continuously capture bowel sounds. The collected signals are transmitted to the processing platform, which forwards them to the companion mobile application for automated analysis and real-time user feedback.
Figure 2: The methodology used in the auto-annotation algorithm starting from BS event detection by measuring the changes in the energy, followed by BS pattern classification using AST model, then adjusting the segment and finish by comparing the manual created label to the auto labels.
Figure 3: The four BS patterns investigated in this study in time and frequency domain, starting from Harmonic Sound (HS), followed by Continuous Random Sound (CRS), Single Burst (SB) and Multiple Burst (MB)
Figure 4: Illustration of the proposed bowel sound (BS) event detection features. The first row shows the bowel sound waveform in the time domain. The second row presents the temporal variations of the RMS amplitude along the waveform with the corresponding adaptive threshold. The third row depicts the changes in signal energy within each short frame together with its adaptive threshold. The fourth row shows the changes in frame energy relative to the baseline energy of the signal and the associated adaptive threshold. The final row displays the detected BS events using different features: RMS-based detection (red), energy variation within frames (blue), and energy variation relative to the baseline energy (green).
Figure 5: The differences in distribution (first row) and duration (second row) between the bowel sound patterns in the data that has been manually labeled titled as Original, and the same data if it is labeled using the auto annotation technique titled predicted and the group of patient that we do not have the manual label for and we have created the labels using the auto annotation algorithm labeled auto annotation, by using the model that has been trained on healthy data only and tested on healthy data too.
Train
Test
Wav2Vec 2.0
AST
ACC
AUROC
ACC
AUROC
Healthy
Healthy
0.82
0.74
0.97
0.98
Patients
0.77
0.74
0.76
0.82
All
0.79
0.73
0.87
0.94
Patients
Healthy
0.67
0.61
0.91
0.90
Patients
0.75
0.69
0.96
0.98
Table 1: Performance comparison between Wav2Vec 2.0 and AST models using Accuracy (ACC) and Area Under the Curve (AUROC).
Figure 6: The differences in normalized counts (first row) and duration (second row) between the bowel sound patterns between ground truth labels (Original; left panel), model predictions for the same set of samples (Predicted; center panel) and auto-annotated samples (AutoLabelled, right panel).
Class
Auto
Expert
Mean Duration
Mean Duration
Count
Count
Auto (s)
Expert (s)
None
929.7
825.3
0.278
0.200
SB
400.4
314.9
0.084
0.101
MB
42.4
61.4
0.415
0.525
CRS
21.3
27.2
0.531
0.625
HS
7.3
3.0
0.740
0.307
Table 2: Class-wise comparison between automatically generated and expert-adjusted bowel sound labels (overall mean across 10 subjects and all quadrants).
Background: Lung-sound classifiers are usually evaluated on recording- or event-level splits, although each child contributes many recordings. We evaluated a foundation-model pipeline (PulmoVec) with the patient as the unit of partitioning and asked whether disease group can be predicted beyond age and sex. Methods: We analyzed 19693 physician-annotated respiratory events from 736 children in the public SPRSound database. A frozen Health Acoustic Representations (HeAR) encoder with low-rank adapters was trained for screening (normal versus adventitious), sound pattern (normal, crackles, wheeze/rhonchi) and disease group (pneumonia, bronchial disease, normal/other); event probabilities were stacked with age, sex and auscultation site. The primary analysis was nested patient-grouped cross-validation; comparators were the majority class, annotated event duration and demographics. Results: The area under the receiver operating characteristic curve (AUC) was 0.95 (95% CI 0.94 to 0.96) for both screening and sound pattern, against 0.78 and 0.75 for event duration alone; the positive predictive value for adventitious events was 0.72. Disease-group prediction reached an AUC of 0.58 (95% CI 0.54 to 0.62), and its accuracy was below the majority-class rate at event level (0.44 versus 0.62) and at patient level (0.46 versus 0.60). Conclusions: Recognition of physician-annotated adventitious events holds under patient-level validation and is not explained by event duration or demographics, although its positive predictive value would limit unaided use. Prediction of disease group, a clinical diagnosis, from lung-sound events alone was not demonstrated. Lung-sound studies should partition data by patient and report non-acoustic comparators.
Izzet Turkalp Akbasli, Oguzhan Serin
Department of Pediatric Intensive Care Medicine, Life Support Center, Hacettepe University, Ankara, Türkiye · Department of Pediatric Emergency Medicine, Life Support Center, Hacettepe University Faculty of Medicine, Ankara, Türkiye
The automatic identification of cough segments in audio through the determination of start and end points is pivotal to building scalable screening tools in health technologies for pulmonary related diseases. We propose the application of two current pre-trained architectures to the task of cough activity detection. A dataset of recordings containing cough from patients symptomatic for tuberculosis (TB) who self-present at community-level care centres in South Africa and Uganda is employed. When automatic start and end points are determined using XLS-R, an average precision of 0.96 and an area under the receiver-operating characteristic of 0.99 are achieved for the test set. We show that best average precision is achieved by utilising only the first three layers of the network, which has the dual benefits of reduced computational and memory requirements, pivotal for smartphone-based applications. This XLS-R configuration is shown to outperform an audio spectrogram transformer (AST) as well as a logistic regression baseline by 9% and 27% absolute in test set average precision respectively. Furthermore, a downstream TB classification model trained using the coughs automatically isolated by XLS-R comfortably outperforms a model trained on the coughs isolated by AST, and is only narrowly outperformed by a classifier trained on the ground truth coughs. We conclude that the application of large pre-trained transformer models is an effective approach to identifying cough end-points and that the integration of such a model into a screening tool is feasible.
Joshua Jansen van Vüren, Devendra Singh Parihar, Daphne Naidoo +4
Department of Electrical and Electronic Engineering, Stellenbosch University, Stellenbosch, South Africa · DSI-NRF Centre of Excellence for Biomedical Tuberculosis Research, South African Medical Research Council Centre for Tuberculosis Research, Division of Molecular Biology and Human Genetics, Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, South Africa · Faculty of Business and Economics, University of Göttingen, Göttingen, Germany +1
Bowel obstruction is a common and potentially life-threatening gastrointestinal condition. In the face of rising diagnostic workloads, the automated diagnosis of bowel obstruction on CT scans supports radiologists by accelerating detection and improving patient outcomes. In this work, we propose a deep learning framework with a multi-task objective that jointly detects bowel obstruction and localizes its transition zone. Additionally, we extend the method with an inherently interpretable classification method that locates the suspected transition point within a slice. It does so by learning a probabilistic selection mask that faithfully bases the classifier's prediction solely on a small image region. The proposed method is evaluated on an internal dataset comprising 1,427 abdominal CTs. Here, the model achieves an obstruction detection test accuracy of 93% and a Hit@10 transition zone localization of 95%. As the first method to reliably localize the transition zone, this marks a significant step towards the automated identification of this critical clinical landmark.
Moritz Vandenhirtz, Andrea Agostini, Dana Belde +5
Department of Computer Science, ETH Zurich, Zurich, Switzerland · Department of Radiology, Kantonsspital Baden, affiliated Hospital for Research and Teaching of the Faculty of Medicine of the University of Zurich, Baden, Switzerland · Department of Forensic Medicine Zurich, University of Zurich, Zurich, Switzerland