Prediction of weight and aerobic fitness with resting-state fMRI

Jeremy Huckins (Dartmouth College), Kathryn Demos (Brown University), Todd Heatherton (Dartmouth College), William Kelley (Dartmouth College)


Obesity is a worldwide problem, caused, in part, by self-regulatory failure of eating behavior. Understanding factors that predict long-term weight fluctuation will be important to identify possible preventative steps. Emerging evidence suggests that neural reward-responsivity is associated with weight gain and eating behaviors.  Cue-reactivity of the nucleus accumbens in response to food images has been shown to predict subsequent weight gain in female college freshmen. Dieters who have recently experienced a self-regulatory failure also show increased cue-reactivity in response to food images compared to those who have not broken their diets (Demos et al., 2010). Alternatively, high levels of physical activity contribute to long-term successful weight loss maintenance (Wing & Hill, 2001).  Aerobic exercise has been shown to upregulate a variety of trophic factors, improve cognition and reduce depression. Identifying the resting neural activity associated with high levels of physical fitness may play an important role in understanding how body mass and aerobic capacity influence cognition. This study uses resting-state functional connectivity to identify neural predictors of Body Mass Index and aerobic capacity (VO2max).


Subjects completed a survey including questions related to height, weight, frequency of physical activity and perceived physical ability (George et al., 1997) that were used to estimate aerobic capacity. Subjects viewed a white fixation cross on a black background for 2 runs of 5 minutes each (BOLD fMRI). Subjects were instructed to simply stay awake, refrain from moving and look at the crosshair. Scan parameters included voxel size of 3x3x3.5, TR=2.5 S and TE=35ms. Standard preprocessing techniques were used and time-series were extracted from 160 regions attributed to 6 functional networks (Dosenbach et al, 2010). Connectivity between each pair of regions was used to predict BMI and VO2max using Support Vector Regression.

Results and Discussion

Functional connectivity, both within and between regions from each of the 6 resting-state networks used in the current study, was found to be important in predicting aerobic capacity and BMI (p<0.001). These results suggest that resting-state neural activity related to BMI and aerobic capacity are diverse and can in part be predicted by within-network coherence and between network interactions. 

Preferred presentation format: Poster
Topic: Neuroimaging

Document Actions