Artifact Detection in Lung Ultrasound: An Analytical Approach

Abstract

Lung ultrasound is used to detect various artifacts in the lungs that support the diagnosis of different conditions. There is ongoing research to support the automatic detection of such artifacts using machine learning. We propose a solution that uses analytical computer vision methods to detect two types of lung artifacts, namely A- and B-lines. We evaluate the proposed approach on the POCUS dataset and data acquired from a hospital. We show that by using the Fourier transform, we can analyze lung ultrasound images in real-time and classify videos with an accuracy above 70%. We also evaluate the method’s applicability for segmentation, showcasing its high success rate for B-lines (89% accuracy) and its shortcomings for A-line detection. We then propose a hybrid solution that uses a combination of neural networks and analytical methods to increase accuracy in horizontal line detection, emphasizing the pleura.

Lung Ultrasound Reduces Chest X-rays in Postoperative Care after Thoracic Surgery: Is There a Role for Artificial Intelligence?—Systematic Review

“…In the next step of our research, AI was trained to detect A-lines and B-lines [ 49 ]. Ultrasound videos from patients after various non-cardiac thoracic procedures and LUS videos from a freely available POCUS dataset were used [ 66 ].…”

“…Increased accuracy for A-lines was achieved using a hybrid solution that combined neural networks training in a pleura of detection and analytical methods. These trials [ 48 , 49 ] represent our preliminary results tested on a small sample size, which is an important limitation. In our ongoing research, we will continue with an evaluation of the rest of the LUS signs relevant in BLUE protocol by AI methods.…”

“…The LUS signs relevant in BLUE protocol are: lung sliding [1], B-lines [2,3], A-lines [3], lung point [4], pleural effusion [7] and lung consolidations [6]. In the next step of our research, AI was trained to detect A-lines and B-lines [49]. Ultrasound videos from patients after various non-cardiac thoracic procedures and LUS videos from a freely available POCUS dataset were used [66].…”

“… Evaluation of the lung ultrasound video using artificial intelligence: ( a ) Lung ultrasound video from linear probe with A-lines; ( b ) Visualization of anatomical landmarks (pleural line) and ultrasound sign (A-line) after evaluation of the same video as in ( a ) by artificial intelligence using analytical approach [ 49 ]. …”

“…Our initial experience with the evaluation of LUS videos from patients after non-cardiac thoracic surgery using deep learning and AI [ 48 , 49 ] encouraged us to perform a literature review with two aims. The first aim was to review the state of the art regarding the use of LUS in postoperative care and chest tube management after non-cardiac thoracic surgery to assess whether lung ultrasound could replace CXR.…”