Dataset Information

Feasibility of Deep Learning-Based Analysis of Auscultation for Screening Significant Stenosis of Native Arteriovenous Fistula for Hemodialysis Requiring Angioplasty.

ABSTRACT:

Objective

To investigate the feasibility of using a deep learning-based analysis of auscultation data to predict significant stenosis of arteriovenous fistulas (AVF) in patients undergoing hemodialysis requiring percutaneous transluminal angioplasty (PTA).

Materials and methods

Forty patients (24 male and 16 female; median age, 62.5 years) with dysfunctional native AVF were prospectively recruited. Digital sounds from the AVF shunt were recorded using a wireless electronic stethoscope before (pre-PTA) and after PTA (post-PTA), and the audio files were subsequently converted to mel spectrograms, which were used to construct various deep convolutional neural network (DCNN) models (DenseNet201, EfficientNetB5, and ResNet50). The performance of these models for diagnosing ≥ 50% AVF stenosis was assessed and compared. The ground truth for the presence of ≥ 50% AVF stenosis was obtained using digital subtraction angiography. Gradient-weighted class activation mapping (Grad-CAM) was used to produce visual explanations for DCNN model decisions.

Results

Eighty audio files were obtained from the 40 recruited patients and pooled for the study. Mel spectrograms of "pre-PTA" shunt sounds showed patterns corresponding to abnormal high-pitched bruits with systolic accentuation observed in patients with stenotic AVF. The ResNet50 and EfficientNetB5 models yielded an area under the receiver operating characteristic curve of 0.99 and 0.98, respectively, at optimized epochs for predicting ≥ 50% AVF stenosis. However, Grad-CAM heatmaps revealed that only ResNet50 highlighted areas relevant to AVF stenosis in the mel spectrogram.

Conclusion

Mel spectrogram-based DCNN models, particularly ResNet50, successfully predicted the presence of significant AVF stenosis requiring PTA in this feasibility study and may potentially be used in AVF surveillance.

SUBMITTER: Park JH

PROVIDER: S-EPMC9523235 | biostudies-literature | 2022 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Feasibility of Deep Learning-Based Analysis of Auscultation for Screening Significant Stenosis of Native Arteriovenous Fistula for Hemodialysis Requiring Angioplasty.

Park Jae Hyon JH Park Insun I Han Kichang K Yoon Jongjin J Sim Yongsik Y Kim Soo Jin SJ Won Jong Yun JY Lee Shina S Kwon Joon Ho JH Kwon Joon Ho JH Moon Sungmo S Kim Gyoung Min GM Kim Man-Deuk MD

Korean journal of radiology 20221001 10

<h4>Objective</h4>To investigate the feasibility of using a deep learning-based analysis of auscultation data to predict significant stenosis of arteriovenous fistulas (AVF) in patients undergoing hemodialysis requiring percutaneous transluminal angioplasty (PTA).<h4>Materials and methods</h4>Forty patients (24 male and 16 female; median age, 62.5 years) with dysfunctional native AVF were prospectively recruited. Digital sounds from the AVF shunt were recorded using a wireless electronic stethos ...[more]

PMID: 36174999

Similar Datasets

Project description:BackgroundStenosis resulting in dysfunctional dialysis access may occur simultaneously on the anastomotic and central venous side. The purpose of this study was to retrospectively evaluate the feasibility of a single sheath inverse technique using the vertical puncture approach to perform bidirectional transvenous percutaneous transluminal angioplasty (PTA) from a single sheath for such dialysis access stenoses.Materials and methodsTwenty patients (26 cases; 13 males; median age, 74 [range: 50-89] years) who underwent PTA using the sheath inverse technique for dysfunctional arteriovenous fistula stenoses between April 2019 and June 2023 were included. All procedures were performed in an outpatient setting. A 4-cm sheath (4Fr, four cases; 5Fr, 19 cases; 6Fr, three cases) was inserted by vertical puncture through a cutaneous vein in the forearm (20 cases) or upper arm (six cases). After treating one side of the lesion, the sheath was reversed to treat the lesion on the opposite side. The vessel diameter at the sheath insertion site, the success rate of sheath inversion, the number of PTA balloon catheters used, the PTA success rate, adverse events, and primary and secondary patency rates up to one year after PTA were evaluated.ResultsThe median diameter at the sheath indwelling site was 5.2 (range: 3.6-9.5) mm, and sheath inversion was successful in all cases, eliminating the need to place an additional sheath at another site for contralateral stricture treatment. The number of balloon catheters used was one and two in 17 (65%) and eight cases (31%), respectively, and three in one case wherein a drug-coated balloon was used. PTA was successful in all cases and major complications were not observed. However, in one case wherein a sheath had to be placed at the arterial needle puncture site, the skin was hard, leading to difficulty in inversion, and transient venous spasm occurred post-inversion. The primary patency rates at 3, 6 and 12 months after the PTA were 87.5%, 41.7%, and 20.8%, respectively. The secondary patency rates at 6 and 12 months were 100% and 75%, respectively.ConclusionThe single-sheath inverse technique for arteriovenous fistulas was feasible without sheath withdrawal.

Dataset Information

Feasibility of Deep Learning-Based Analysis of Auscultation for Screening Significant Stenosis of Native Arteriovenous Fistula for Hemodialysis Requiring Angioplasty.

Objective

Materials and methods

Results

Conclusion

Publications

Feasibility of Deep Learning-Based Analysis of Auscultation for Screening Significant Stenosis of Native Arteriovenous Fistula for Hemodialysis Requiring Angioplasty.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets