Bone-conduction digital auscultation system for automatic separation of heart and lung sounds

Abstract

The invention discloses an auscultation system for heart and lung sound separation based on deep learning; bone conduction is used for auscultation. The auscultation device includes an auscultation head, a signal acquisition module, a heart and lung sound signal separation module, a memory, a user interface, an output driving unit and a bone conduction unit. The signal acquisition module acquires heart and lung sound signals and digitizes the heart and lung sound signals, and the heart and lung sound signal separation module adopts a time-frequency masking deep circulatory neural network for heart and lung sound signal separation; the user interface can select heart-lung sound modes and store expert tags. The output driving unit receives heart or lung sound time-domain signals separated from previous stage for power amplification, and drives the bone conduction unit to generate vibration. The auscultator can automatically separate clean heart and lung sounds from the overlapped heart and lung sounds, facilitate data acquisition and post-analysis, and enhance the auscultation experience.

Description

technical field [0001] The invention belongs to the technical field of digital medical equipment, and proposes a bone conduction digital auscultation system for automatically separating heart and lung sounds. Background technique [0002] The stethoscope is an inseparable tool for doctors in their daily work, and once became synonymous with doctors. So far, the stethoscope has been developed for more than 200 years. Its shape and sound transmission method have been continuously improved, but the basic structure has not changed much. It mainly consists of a sound pickup part, a sound transmission part, and a listening part. Traditional stethoscopes generally use Y-shaped fixtures, which are uncomfortable to wear and susceptible to environmental sound interference. Moreover, the sound signals collected by the stethoscope include not only heart sounds, but also lung sounds and external environmental noise. This mixture of heart sounds Acoustic signals of , lung sound and envir...

AI Technical Summary

Problems solved by technology

[0003] To solve this problem, the current solutions mainly include the following: For example, the Chinese utility model patent 201620968593.5 also mentions the function of heart-lung sound separation, but it uses a linear band-pass filter, which cannot separate the time-frequency aliased heart sound and lung sound. Fine separation may even lead to the loss of part of the frequency band information of heart sounds or lung sounds, thus making auscultation inaccurate; for example, Chinese invention patent 201710470514.7 uses the idea of ​​difference enhancement to perform blind channel separation on heart and lung sound signals, but this method requires simultaneous acquisition of multiple signals , it will inevitably make the stethoscope large in size and complex in structure; another example is the Chinese invention patent 201710651856.9, which uses non-negative matrix decomposition to separate heart and lung sounds, because this method assumes the independence between heart sounds and lung sounds, and cannot effectively represent heart sounds and lung sounds The complex nonlinear coupling relationship between them invisibly limits the performance of the method

[0004] In addition, the general electronic stethoscope finally restores the electrical signal to sound through the earphone for auscultation. The common air conduction earphone is mostly used. The sound signal is easily disturbed by noise in the air, and the auscultation effect is not ideal.

Images