China
Africa
Explore chapters and articles related to this topic
Noninvasive Diagnosis Using Sounds Originating from within the Body
Published in Robert B. Northrop, Non-Invasive Instrumentation and Measurement in Medical Diagnosis, 2017
As air passes in and out of the normal respiratory system during normal breathing, certain sounds can be heard by auscultation of the back and chest over the lungs, trachea, and bronchial tubes with an acoustic or electronic stethoscope. As in the case of heart sounds, it requires a good ear and much experience to use breath sounds effectively for NI diagnosis. The normal sounds perceived are due in part to air turbulence, air turbulence exciting damped mechanical resonances in connective tissues, and alveoli stretching open on inspiration, and shrinking on expiration. Normal breath sounds are classified as tracheal, bronchial, bronchovesicular, and vesicular. Tracheal sounds are heard over the trachea; they have a harsh quality and sound like air moving through a pipe. Heard over the anterior chest over the sternum and at the second and third intercostal spaces, the bronchial sounds originate in the bronchial tubes (Figure 3.9) and have a more hollow quality, not as harsh as tracheal sound. They are generally louder and higher in pitch; expiratory bronchial sounds last longer than inspiratory sounds, and there is a pause at peak inspiration. Bronchovesicular sounds are heard in the posterior chest, between the scapulae, and also in the center of the anterior chest. They are softer than bronchial sounds and also have a tubular quality. Vesicular sounds are soft, breezy or rustling in nature, and can be heard throughout most of the lung fields. Heard throughout inspiration, they continue with no pause through the beginning of expiration, and fade away about one-third of the way through expiration.
Deep Learning and Multimodal Artificial Neural Network Architectures for Disease Diagnosis and Clinical Applications
Published in Om Prakash Jena, Bharat Bhushan, Nitin Rakesh, Parma Nand Astya, Yousef Farhaoui, Machine Learning and Deep Learning in Efficacy Improvement of Healthcare Systems, 2022
India has 18% of the world's population, but according to the Global Burden of Disease Study (GBD), India bears 32% of the global burden of respiratory diseases. Pollution is the biggest contributor to this disease burden. If the doctor thinks that there is an issue with the lungs, the location and type of certain breathing sounds can possibly help figure out the cause. Auscultation of the lungs is an important respiratory examination and helps diagnose various lung diseases. The high-pitched whistling noise that can occur while breathing in or out is known as wheezing. Doctors typically use a stethoscope to detect the breath sounds of the patient. Faint heart murmurs and rales in the lungs might lead to a false diagnosis. The recurrent neural network (RNN) is trained with analyzing the breathing sounds to ease the examination by the cardiologists. RNN has the ability to remember the events that happened in the past, and that can be used to predict future outcomes. This principle can be used in medical diagnosis to ascertain the status of the patient based on previous clinical data sets. The output of the RNN can be calculated as y=σ(W0ht) where W0 is a matrix contains parameters of the model, ht a vector contains hidden states of the neural network, σ is the logistic sigmoid activation function. The hidden state of the RNN, based on current input, xt can be represented as ht=f(ht−1,xt)
Assessing the accuracy of artificial intelligence enabled acoustic analytic technology on breath sounds in children
Published in Journal of Medical Engineering & Technology, 2022
Zai Ru Cheng, Huiyu Zhang, Biju Thomas, Yi Hua Tan, Oon Hoe Teoh, Arun Pugalenthi
Two trained paediatric respiratory physicians blinded to each subject’s diagnosis, separately auscultated the subject’s chest with a stethoscope for breath and adventitious sounds. The physicians provided their auscultation findings independently. Similar to conventional auscultation, the stethoscope head of the novel sensing device was then placed on different zones on the chest wall for recording of breath sound samples. Only 1 breath sound recording was collected per subject, to ensure that the samples contributed unique data for effective machine learning. Labels of normal, wheeze or crackles were used according to the European Respiratory Society (ERS) nomenclature [14]. For the standardisation of breath sounds labelling as per the ERS, wheeze was defined as a continuous musical adventitious sound that included both low- and high-pitched sounds. Crackles were defined as discontinuous, explosive, non-musical sounds and for the purposes of this study, taken to include both coarse and fine crackles. Children aged between 0–16 years in both respiratory outpatient clinics and inpatient units were studied. The samples contained inadvertent ambient noise resulting from crying, coughing, talking and movements from the stethoscopic sensors, reflective of a real life paediatric clinical setting.