A comparison of point-tracking algorithms in ultrasound videos from the upper limb

BioMedical Engineering OnLine

Research

Micha Feigin-Almon¹  Brian Anthony²  Uriel Magana-Salgado³  Roger Pallares-Lopez³  Praneeth Namburi⁴ 

[1] Department of Mechanical Engineering, MIT, 02139, Cambridge, MA, USA;Department of Mechanical Engineering, MIT, 02139, Cambridge, MA, USA;Institute for Medical Engineering and Science, Massachusetts Institute of Technology, 77 Massachusetts Ave, 12-3211, 02139, Cambridge, MA, USA;MIT.Nano Immersion Lab, MIT, 02139, Cambridge, MA, USA;Department of Mechanical Engineering, MIT, 02139, Cambridge, MA, USA;Mechanical Engineering Graduate Program, MIT, 02139, Cambridge, MA, USA;Institute for Medical Engineering and Science, Massachusetts Institute of Technology, 77 Massachusetts Ave, 12-3211, 02139, Cambridge, MA, USA;MIT.Nano Immersion Lab, MIT, 02139, Cambridge, MA, USA;

关键词: Ultrasound;    Point tracking;    OpenCV;    Neural network tracking;    Tracking correction;   

DOI  :  10.1186/s12938-023-01105-y

 received in 2023-01-16, accepted in 2023-04-25,  发布年份 2023

来源: Springer

【 摘 要 】

Tracking points in ultrasound (US) videos can be especially useful to characterize tissues in motion. Tracking algorithms that analyze successive video frames, such as variations of Optical Flow and Lucas–Kanade (LK), exploit frame-to-frame temporal information to track regions of interest. In contrast, convolutional neural-network (CNN) models process each video frame independently of neighboring frames. In this paper, we show that frame-to-frame trackers accumulate error over time. We propose three interpolation-like methods to combat error accumulation and show that all three methods reduce tracking errors in frame-to-frame trackers. On the neural-network end, we show that a CNN-based tracker, DeepLabCut (DLC), outperforms all four frame-to-frame trackers when tracking tissues in motion. DLC is more accurate than the frame-to-frame trackers and less sensitive to variations in types of tissue movement. The only caveat found with DLC comes from its non-temporal tracking strategy, leading to jitter between consecutive frames. Overall, when tracking points in videos of moving tissue, we recommend using DLC when prioritizing accuracy and robustness across movements in videos, and using LK with the proposed error-correction methods for small movements when tracking jitter is unacceptable.

【 授权许可】

CC BY   
© The Author(s) 2023

【 图 表 】

Fig. 1

Fig. 5

41116_2023_36_Article_IEq130.gif

41116_2023_36_Article_IEq129.gif

41116_2023_36_Article_IEq128.gif

41116_2023_36_Article_IEq127.gif

41116_2023_36_Article_IEq126.gif

41116_2023_36_Article_IEq125.gif

41116_2023_36_Article_IEq124.gif

41116_2023_36_Article_IEq123.gif

41116_2023_36_Article_IEq122.gif

41116_2023_36_Article_IEq121.gif

41116_2023_36_Article_IEq120.gif

41116_2023_36_Article_IEq119.gif

41116_2023_36_Article_IEq118.gif

41116_2023_36_Article_IEq114.gif

41116_2023_36_Article_IEq112.gif

41116_2023_36_Article_IEq110.gif

41116_2023_36_Article_IEq109.gif

41116_2023_36_Article_IEq108.gif

40517_2023_256_Article_IEq37.gif

41116_2023_36_Article_IEq106.gif

41116_2023_36_Article_IEq105.gif

41116_2023_36_Article_IEq103.gif

41116_2023_36_Article_IEq99.gif

41116_2023_36_Article_IEq98.gif

41116_2023_36_Article_IEq96.gif

41116_2023_36_Article_IEq95.gif

40517_2023_256_Article_IEq28.gif

41116_2023_36_Article_IEq94.gif

41116_2023_36_Article_IEq93.gif

41116_2023_36_Article_IEq92.gif

41116_2023_36_Article_IEq91.gif

Fig. 4

Fig. 1

41116_2023_36_Article_IEq88.gif

41116_2023_36_Article_IEq87.gif

41116_2023_36_Article_IEq86.gif

41116_2023_36_Article_IEq85.gif

41116_2023_36_Article_IEq84.gif

41116_2023_36_Article_IEq83.gif

Fig. 1

12888_2023_4839_Article_IEq2.gif

41116_2023_36_Article_IEq80.gif

12888_2023_4839_Article_IEq1.gif

Fig. 3

41116_2023_36_Article_IEq78.gif

12888_2023_4816_Article_IEq4.gif

12888_2023_4816_Article_IEq3.gif

12888_2023_4816_Article_IEq2.gif

12888_2023_4816_Article_IEq1.gif

41116_2023_36_Article_IEq72.gif

Fig. 1

41116_2023_36_Article_IEq70.gif

41116_2023_36_Article_IEq69.gif

41116_2023_36_Article_IEq68.gif

Fig. 2

Fig. 3

Fig. 1

Fig. 1

40517_2023_256_Article_IEq16.gif

Fig. 4

40517_2023_256_Article_IEq13.gif

40517_2023_256_Article_IEq12.gif

40517_2023_256_Article_IEq11.gif

40517_2023_256_Article_IEq208.gif

Fig. 2

40517_2023_256_Article_IEq5.gif

Fig. 2

Fig. 3

40517_2023_256_Article_IEq2.gif

Fig. 2

40517_2023_256_Article_IEq204.gif

Fig. 7

40517_2023_256_Article_IEq203.gif

Fig. 1

41116_2023_36_Article_IEq61.gif

41116_2023_36_Article_IEq57.gif

41116_2023_36_Article_IEq54.gif

Fig. 1

41116_2023_36_Article_IEq20.gif

Fig. 5

40517_2023_256_Article_IEq146.gif

Fig. 24

12864_2023_9380_Article_IEq7.gif

Fig. 2

Fig. 4

40517_2023_256_Article_IEq133.gif

Fig. 3