Print  Close

Identifying food cue-induced activity pattern for obesity and normal−weight individuals via LSTM

Poster No:

1460 

Submission Type:

Abstract Submission 

Authors:

Zaichen La1,2, Guanya Li1,2, Wenchao Zhang1,2, Yang Hu1,2, Jingyuan Li1,2, Weibin Ji1,2, Mengshan Li1,2, Huiling Zhou1,2, Yonghuan Feng1,2, Zhao Yan1,2, Yi Zhang1,2

Institutions:

1Center for Brain Imaging, School of Life Science and Technology, Xidian University, Xi'an, Shaanxi 710126, China, 2International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University, Xi'an, shaanxi 710126, China

First Author:

Zaichen La  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China

Co-Author(s):

Guanya Li  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Wenchao Zhang  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Yang Hu  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Jingyuan Li  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Weibin Ji  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Mengshan Li  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Huiling Zhou  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Yonghuan Feng  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Zhao Yan  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China
Yi Zhang  
Center for Brain Imaging, School of Life Science and Technology, Xidian University|International Joint Research Center for Advanced Medical Imaging and Intelligent Diagnosis and Treatment & Xi'an Key Laboratory of Intelligent Sensing and Regulation of trans-Scale Life Information, School of Life Science and Technology, Xidian University
Xi'an, Shaanxi 710126, China|Xi'an, shaanxi 710126, China

Introduction:

Growing evidence indicates individuals with obesity exhibit hyperactive brain responses to high−caloric food cues in regions involved in food reward processing [1]. The enhanced sensitivity of the reward circuitry is associated with higher food craving [2−4], and it is also accompanied by impaired function of the executive control circuitry, which is an important cause of excessive eating [5]. Most neuroimaging studies focused on altered functional activity in certain brain regions. However, it is still poorly unclear whether the whole-brain activity pattern of visual perception in food energy density in obese patients is different from normal weight individuals. With the application of artificial neural network methods in the field of neuroimaging, it provides a new way to understand the brain functional activity pattern from a higher dimension.

Methods:

fMRI task with high− (HiCal) and low−caloric (LoCal) food cues was employed to investigate brain responses to visual perception in food energy density cues in 155 individuals with obesity (OB) and 105 normal−weight controls (NW). Based on Brainnetome Atlas, the time series of 246 brain functional regions were extracted under the stimulation of HiCal and LoCal food pictures. Thus, the 1D Convolution operations (conv1D) and the long short-term memory block (LSTM) were utilized to measure the time sequence information. The attention mechanism was further employed to recognize the significative features from the aforementioned information, which were used to construct HiCal/LoCal food cue-induced activity pattern for OB and NW group respectively via 5-fold cross-validation (Fig. 1A). The importance of features in different brain regions were accessed by the shap interpreter. In order to identify the meaningful brain regions related to activity patterns of OB and NW, test data from OB (NW) were further cross-tested by the NW (OB) classifier (Fig. 1B).

Results:

Based on the brain functional activity signals, the proposed model can effectively distinguish the HiCal/LoCal food cue–induced functional activity patterns from OB to NW group, and the accuracy of the OB model (90.96%) is significantly higher than NW model (83.92%) (t = 5.42, P < 0.001). Cross−testing showed lower accuracy of NW data tested in the OB model (OB model→NW data: 78.84%) than in the NW model(t = -3.92, P < 0.001) and lower accuracy of OB data tested in the NW model (NW model→OB data: 83.12%) than in the NW model(t = -7.86, P < 0.001), however there was no significant difference between the accuracy of OB and NW data tested in the NW model (t = 0.45, P = 0.66, Fig. 1C). Feature importance analysis showed that the top 20% of important features of the NW model (including dorsolateral prefrontal cortex, hippocampal gyrus, parahippocampal gyrus, Lingual_R, Fusiform_R, Frontal_Mid_Orb_L, Frontal_Sup_Medial_L and so on) were significantly different with the OB model (including insula, nucleus accumbens, orbitofrontal cortex, Lingual, Cuneus_R, Fusiform, Calcarine_R, Precuneus_L, Cingulum_Ant_R, Putamen, Frontal_Sup_Orb, Frontal_Sup_Medial, Occipital_Inf, Occipital_Mid_L cortex and so on, Fig. 2A). There was a high degree of overlap between the top 50% of important brain regions of the NW and OB model.
Supporting Image: zhongban.png
   ·Fig 1. Experimental design and results. A. Flowchart of the proposed framework. B. Data splitting, model training and testing. C. Comparison results of experiments.
Supporting Image: zhongbanquan.png
   ·Fig 2. Brain regions contributing most to the classifiers. A. The top 20% regions. B. The top 50% regions.
 

Conclusions:

These findings indicate that the important role of the memory and executive control circuits in the food energy density perception in NW, while the insula, nucleus accumbens and the orbitofrontal gyrus, Calcarine_R, Lingual, Cuneus_R and Occipital_Inf_R in OB group enhance the brain perception of food energy density.

Modeling and Analysis Methods:

Classification and Predictive Modeling 1

Novel Imaging Acquisition Methods:

BOLD fMRI 2

Keywords:

FUNCTIONAL MRI
Other - LSTM; food cue; obesity

1|2Indicates the priority used for review

Provide references using author date format

[1] Li, Guanya et al. “Resting activity of the hippocampus and amygdala in obese individuals predicts their response to food cues.” Addiction biology vol. 26,3 (2021): e12974. doi:10.1111/adb.12974
[2] Dunigan, Anna I, and Aaron G Roseberry. “Actions of feeding-related peptides on the mesolimbic dopamine system in regulation of natural and drug rewards.” Addiction neuroscience vol. 2 (2022): 100011. doi:10.1016/j.addicn.2022.100011
[3] Alsiö, Johan et al. “Feed-forward mechanisms: addiction-like behavioral and molecular adaptations in overeating.” Frontiers in neuroendocrinology vol. 33,2 (2012): 127-39. doi:10.1016/j.yfrne.2012.01.002
[4] McCutcheon, James Edgar. “The role of dopamine in the pursuit of nutritional value.” Physiology & behavior vol. 152,Pt B (2015): 408-15. doi:10.1016/j.physbeh.2015.05.003
[5] Gunstad, John et al. “Cognitive dysfunction is a risk factor for overeating and obesity.” The American psychologist vol. 75,2 (2020): 219-234. doi:10.1037/amp0000585