.png){kind=icon}
.png){kind=icon}
Return
Decoding dynamic bamboo cell shrinkage with time-lapse microscopy and machine-learning
.png){kind=icon}
PRE
AI
DOI：10.1016/j.indcrop.2024.118965.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The shrinkage of bamboo can be traced to the cellular scale; however, research on the shrinkage of bamboo cells' scale is rare. This gap primarily arises from the intrinsic heterogeneity of biomass materials, leading to significant differences in the shrinkage behaviors of bamboo cells. Consequently, conventional measurement and analysis techniques struggle to obtain comprehensive and accurate results. To fill in this gap, this study investigated the dynamic shrinking behavior of individual bamboo cells, including fiber cells and parenchyma cells, using timelapse microscopy and machine-learning techniques. Time-lapse observations revealed that the shrinkage rate of fiber cells across the cross-section surpassed that of parenchyma cells, accompanied by a greater shrinkage deformation. Specifically, fiber cells exhibited a shrinkage rate of 2.53 %/min, whereas parenchyma cells showed a rate of 0.58 %/min. This resulted in corresponding diameter reductions of approximately -30 % and -7 %, respectively. Additionally, machine-learning results demonstrated the efficacy of the trained long short-term memory (LSTM) model in accurately predicting morphological changes in individual parenchyma cells during the shrinkage process. This study advances understanding of bamboo-moisture interaction mechanisms and offers crucial theoretical data for improving bamboo drying and minimizing crack formation.
Keywords:
Bamboo
Dynamic shrinkage
Machine-learning
Time-lapse microscopy
Parenchyma cells
Journal
IF:
6.2
Papers:
1.8W
Citations:
6.8W