Food Images Research Articles

Using Crowdsourced Food Image Data for Assessing Restaurant Nutrition Environment: A Validation Study.

Crowdsourced online food images, when combined with food image recognition technologies, have the potential to offer a cost-effective and scalable solution for the assessment of the restaurant nutrition environment. While previous research has explored this approach and validated the accuracy of food image recognition technologies, much remains unknown about the validity of crowdsourced food images as the primary data source for large-scale assessments. In this paper, we collect data from multiple sources and comprehensively examine the validity of using crowdsourced food images for assessing the restaurant nutrition environment in the Greater Hartford region. Our results indicate that while crowdsourced food images are useful in terms of the initial assessment of restaurant nutrition quality and the identification of popular food items, they are subject to selection bias on multiple levels and do not fully represent the restaurant nutrition quality or customers' dietary behaviors. If employed, the food image data must be supplemented with alternative data sources, such as field surveys, store audits, and commercial data, to offer a more representative assessment of the restaurant nutrition environment.

Selecting foods from FNDDS when calculating food intake: Does the reference matter?

To estimate food intake, the nutrient content of food must be known or inferred by matching it to a nutrient database. This study evaluated how different procedures for matching foods to the Food and Nutrient Database for Dietary Studies (FNDDS) affect food intake estimates. Adult participants estimated a meal with the PortionSize app, which contains a subset of FNDDS codes (PortionSize FNDDS). Meals were covertly weighed (weigh back), and intake was estimated by referencing the FNDDS dataset. Combination foods were evaluated by 1) a composite generated by summing the nutrient values (from FNDDS) from single food items and 2) using the single composite food code from FNDDS. Raters analyzed food images from the PortionSize app to estimate intake with the digital photography of foods (DPF) method. Raters estimated intake from: 1) PortionSize FNDDS and 2) full FNDDS. Using PortionSize’s FNDDS database, participants’ energy intake estimates were equivalent to rater estimated energy intake. Raters estimated almost identical energy intake when they used PortionSize vs. full FNDDS. Using the full FNDDS, the DPF’s energy estimates were similar to weigh back. Estimation of energy intake with a single FNDDS code was not equivalent to estimation from summing the nutrient values for combination foods.

Risk of suffering from eating disorders students in the 1st and 2nd semester of the Medicine degree

Background: eating Disorders are complex psychiatric illnesses caused by various factors, mainly during adolescence and youth. The symptomatology involves excessive concern about food, weight, and/or body image and unhealthy habits to control or reduce body weight. Methods: a cross-sectional descriptive observational study was conducted using the SCOFF and EAT-26 instruments to assess eating behavior in first and second- semester university students at a private university in Asunción, Paraguay. The questionnaire, consisting of closed-ended multiple-choice questions, was processed using Microsoft Office's Excel, enabling a detailed analysis of the sample. Results: the sample of 83 students, predominantly composed of women (74,7 %) aged between 18 and 24, showed a 41 % risk of eating disorders according to the SCOFF scale and 20 % according to the EAT-26 scale. Notably, there is a significant prevalence of women at higher risk of eating disorders compared to men. The results underscore the severity of these disorders and highlight specific risk factors in university students, such as academic stress and social pressure to maintain a slim body. Conclusion: findings from this cross-sectional study suggest a moderate prevalence of eating disorder risk in first and second-semester medical students, particularly among women. The EAT-26 survey, encompassing more items and response options, identified a 20 % risk, while the briefer and dichotomous SCOFF questionnaire reported 41 % risk prevalence. However, due to the cross-sectional design and the nature of the instruments, these findings should be interpreted cautiously. Longitudinal studies with comprehensive diagnostic assessments are needed to draw more robust conclusions about the true prevalence and incidence of eating disorders in this population.

FoodMask: Real-time food instance counting, segmentation and recognition

Food computing has long been studied and deployed to several applications. Understanding a food image at the instance level, including recognition, counting and segmentation, is essential to quantifying nutrition and calorie consumption. Nevertheless, existing techniques are limited to either category-specific instance detection, which does not reflect precisely the instance size at the pixel level, or category-agnostic instance segmentation, which is insufficient for dish recognition. This paper presents a compact and fast multi-task network, namely FoodMask, for clustering-based food instance counting, segmentation and recognition. The network learns a semantic space simultaneously encoding food category distribution and instance height at pixel basis. While the former value addresses instance recognition, the latter value provides prior knowledge for instance extraction. Besides, we integrate into the semantic space a pathway for class-specific counting. With these three outputs, we propose a clustering algorithm to segment and recognize food instances at a real-time speed. Empirical studies are made on three large-scale food datasets, including Mixed Dishes, UECFoodPixComp and FoodSeg103, which cover Western, Chinese, Japanese and Indian cuisines. The proposed networks outperform benchmarks in both terms of instance map quality and speed efficiency.

Food cue reward salience does not explain Hyperphagia in adolescents with Prader-Willi syndrome

ABSTRACT Prader-Willi Syndrome (PWS) is characterized by hyperphagia, an extreme and persistent hunger that emerges in early childhood. We used event-related potentials (ERPs) to objectively investigate brain responses to low- and high-calorie foods, animals, and household objects in 20 satiated adolescents with PWS. Late Positive Potential (LPP) responses to food images did not differ from non-food images. Rather, we observed larger ERPs to high-calorie foods relative to animal images (p=.001) in an earlier time window. These responses correlated with greater severity of hyperphagia (p = .01). Thus, hyperphagia associated with PWS may be due to altered satiety regulation rather than increased motivational salience.

Discrimination exposure impacts unhealthy processing of food cues: crosstalk between the brain and gut

Experiences of discrimination are associated with adverse health outcomes, including obesity. However, the mechanisms by which discrimination leads to obesity remain unclear. Utilizing multi-omics analyses of neuroimaging and fecal metabolites, we investigated the impact of discrimination exposure on brain reactivity to food images and associated dysregulations in the brain–gut–microbiome system. We show that discrimination is associated with increased food-cue reactivity in frontal-striatal regions involved in reward, motivation and executive control; altered glutamate-pathway metabolites involved in oxidative stress and inflammation as well as preference for unhealthy foods. Associations between discrimination-related brain and gut signatures were skewed towards unhealthy sweet foods after adjusting for age, diet, body mass index, race and socioeconomic status. Discrimination, as a stressor, may contribute to enhanced food-cue reactivity and brain–gut–microbiome disruptions that can promote unhealthy eating behaviors, leading to increased risk for obesity. Treatments that normalize these alterations may benefit individuals who experience discrimination-related stress.

PyTorch Deep Learning for Food Image Classification with Food Dataset

Classification of food images is crucial in today's increasingly digitally connected world. In the rapidly evolving mobile applications and social media era, the demand for an automated system that can recognize food types from an image is intensifying. This study employs deep learning and the PyTorch framework to develop a dependable and efficient solution for classifying food images. This research is motivated by the growing complexity of food introduction challenges. The primary challenge is improving the accuracy of food type recognition and overcoming variations in the visual presentation of food, such as lighting, shooting angles, and proportional and textural differences. Convolutional Neural Networks (CNN) are effective for image classification and are incorporated into the methods utilized. In addition, we employ ResNet101 transfer learning techniques to capitalize on the knowledge of trained models for large image datasets. The primary objective of this study is to develop a food image classification model that is accurate, training-efficient, and capable of accurately recognizing various types of food. In testing and evaluation, the developed model could realize multiple types of food with satisfactory accuracy. The accuracy of training reached 99.35%, while the accuracy of testing reached 94.65%. This study also reveals how Resnet101 transfer learning is utilized by deep learning technology.

Indian Food Image Classification Using Convolutional Neural Network

Abstract: Food classification using Convolutional Neural Networks (CNNs) has gained significant attention due to its potential applications in dietary analysis, food recommendation systems, and nutrition monitoring. In this study, we present a CNN-based approach for food classification, leveraging image data augmentation to enhance the model's ability to generalize to new and unseen food images. We utilize a dataset containing diverse food categories and preprocess the images by resizing and normalization. The proposed CNN architecture consists of multiple convolutional and pooling layers, along with dropout and batch normalization for regularization. To prevent overfitting, we employ early stopping as a custom callback during model training. The experimental results demonstrate promising training accuracy of 90.13% and relatively low training loss of 0.3066, indicating effective learning from the training data. the validation accuracy of 69.00% and validation loss of 1.270.

A New CNN-Based Single-Ingredient Classification Model and Its Application in Food Image Segmentation.

It is important for food recognition to separate each ingredient within a food image at the pixel level. Most existing research has trained a segmentation network on datasets with pixel-level annotations to achieve food ingredient segmentation. However, preparing such datasets is exceedingly hard and time-consuming. In this paper, we propose a new framework for ingredient segmentation utilizing feature maps of the CNN-based Single-Ingredient Classification Model that is trained on the dataset with image-level annotation. To train this model, we first introduce a standardized biological-based hierarchical ingredient structure and construct a single-ingredient image dataset based on this structure. Then, we build a single-ingredient classification model on this dataset as the backbone of the proposed framework. In this framework, we extract feature maps from the single-ingredient classification model and propose two methods for processing these feature maps for segmenting ingredients in the food images. We introduce five evaluation metrics (IoU, Dice, Purity, Entirety, and Loss of GTs) to assess the performance of ingredient segmentation in terms of ingredient classification. Extensive experiments demonstrate the effectiveness of the proposed method, achieving a mIoU of 0.65, mDice of 0.77, mPurity of 0.83, mEntirety of 0.80, and mLoGTs of 0.06 for the optimal model on the FoodSeg103 dataset. We believe that our approach lays the foundation for subsequent ingredient recognition.

No effect of hunger on attentional capture by food cues: Two replication studies

Food cues potently capture human attention, and it has been suggested that hunger increases their propensity to do so. However, the evidence for such hunger-related attentional biases is weak. We focus on one recent study that did show significantly greater attentional capture by food cues when participants were hungry, using an Emotional Blink of Attention (EBA) task [Piech, Pastorino, & Zald, 2010. Appetite, 54, 579–582]. We conducted online (N = 29) and in-person (N = 28) replications of this study with British participants and a Bayesian analytical approach. For the EBA task, participants tried to identify a rotated target image in a Rapid Serial Visual Presentation (RSVP). Targets were preceded by “neutral”, “romantic”, or “food” distractor images. Participants completed the task twice, 6–11 days apart, once hungry (overnight plus 6h fast) and once sated (after a self-selected lunch in the preceding 1h). We predicted that food images would create a greater attentional blink when participants were hungry than when they were sated, but romantic and neutral images would not. We found no evidence that hunger increased attentional capture by food cues, despite our experiments passing manipulation and quality assurance checks. Our sample and stimuli differed from the study we were replicating in several ways, but we were unable to identify any specific factor responsible for the difference in results. The original finding may not be generalisable. The EBA is more sensitive to the physical distinctiveness of distractors from filler and target images than their emotional valence, undermining the sensitivity of the EBA task for picking up subtle changes in motivational state. Moreover, hunger-related attentional bias shifts may not be substantial over the intensities and durations of hunger typically induced in laboratory experiments.

The energy content of meals with a similar macronutrient distribution may have a greater impact on appetite sensations than food preferences in adolescents with obesity: A secondary analysis

PurposeThe energy and macronutrient composition of a meal has been shown to influence postprandial appetitive responses, but it is not clear how energy content independent of macronutrient distribution affects postprandial appetite in adolescents with obesity. Extracting data from a primary study testing the effect of energy turnover on appetite, this secondary analysis assessed how fixed meals varying in energy content with similar macronutrient distributions influences postprandial appetite sensations and food reward. MethodsUsing a randomised, counterbalanced crossover design, N = 14 adolescents with obesity (Mage = 12.71, SDage = 0.99; 10 female) consumed fixed lunch meals with similar macronutrient content starting at 750 kcal in energy and progressively increasing by 250 kcal on three different test days. Liking and wanting for food images varying in fat and taste were assessed at baseline and immediately after meal consumption. Appetite sensations were assessed in half-hour intervals from baseline to 1-h post-meal. ResultsThe area under the curve for subjective hunger (p < .001, ω = 0.36), desire to eat (p < .001, ω = 0.54), and prospective food consumption (p = .004, ω = 0.32) were significantly lower after consumption of the higher calorie meals relative to the lowest. Explicit wanting for sweet foods increased after intake of the intermediate calorie meal yet decreased after the highest calorie meal (p = .014, ω = 0.09). Conversely, implicit wanting sweet bias increased after consumption of the test meal independent of energy content (p < .001, ω = 0.15). ConclusionThe consumption of additional calories without a meaningful change in macronutrient distribution may affect appetite sensations more reliably than hedonic responses to energy-dense foods in adolescents with obesity.

Identification of foods in the breakfast and determination of nutritional facts using deep convolution neural networks

Food recognition plays a crucial role in various fields, including healthcare, nutrition, and the food industry. It involves identifying different types of foods or dishes from images, videos, or other data sources. In healthcare, food recognition aids individuals in monitoring their daily food intake and managing their diet. It also assists dietitians and nutritionists in creating personalized meal plans based on patients' nutritional requirements and preferences. This article focuses on the development of software that can recognize food products and predict their nutritional facts. The software extracts essential nutritional facts such as fat, carbohydrates, protein, and energy from the food products and compiles them into a comprehensive list. For each of the 20 food products, 36 food images were obtained, resulting in a total of 720 food images. To validate the accuracy of the trained models, six different images of each food product were set aside for external validation purposes. The rest of the images were then trained using deep learning algorithms, namely, GoogleNet, ResNet-50, and Inception-v3, in the MATLAB software. The training and validation processes yielded over 98% correct predictions for each of the deep learning algorithms. Although there were no significant differences in accuracy among the algorithms, GoogleNet stood out when considering both prediction accuracy and prediction time. The validated deep learning algorithms were employed in developing the software for food recognition and nutritional value determination. The results indicate that the developed software can reliably identify foods and provide their corresponding nutritional facts. This software holds significant potential for application in the nutrition and dietetic field and can be particularly useful in healthcare settings for monitoring the dietary intake of patients with chronic diseases such as diabetes, heart disease, or obesity. The system can track the types and quantities of foods consumed, offering personalized feedback to patients and healthcare providers.

Explore, eat and revisit: does local food consumption value influence the destination’s food image?

PurposeThis research investigates how foreign tourists' revisit intentions are influenced by their local food consumption (LFC) value by emphasizing their attitude towards the local food and the corresponding destination food image. It will also reveal the foreign tourists' food consumption value and explore its influences on the destination's food image.Design/methodology/approachThe data collection was performed from 433 foreign tourists who visited Delhi, India, using a structured survey instrument and chief constructs were measured as the first-order reflective variables. The 39 items associated with the LFC value underwent both exploratory and confirmatory evaluations. The authors employed partial least square structural equation modeling (PLS-SEM). The model's discriminant and convergent validity, consistency and overall fit were evaluated using confirmatory factor analysis (CFA).FindingsThe findings revealed that “tourists' attitudes and behaviors toward the local cuisine” had a significant and positive influence on their “intention to visit/revisit” and “intention to recommend”. Also, “destination food image” significantly and positively influenced the “intention of tourist to visit/revisit”; however, it negatively and insignificantly influenced the “intention of tourists to recommend”.Research limitations/implicationsThis research used five variables related to LFC—quality, health/nutrition, emotion, prestige and price—that impact tourists' attitudes and behaviors toward local foods in Delhi. However, in addition to these constructs, other factors or constructs may be involved that could affect the tourists' attitudes and behaviors. Future studies might explore and include these constructs to provide a more comprehensive image of Delhi's LFC value.Originality/valueUnderstanding tourists' food-linked behaviors is critical for effective market conduct. However, the interrelations between travelers' destination food image, LFC value, their perceptions of the local cuisine and behavioral intentions are still unknown, and this will be one of the first attempts to discuss these behaviors.

Daily fat intake is associated with basolateral amygdala response to high-calorie food cues and appetite for high-calorie food

ABSTRACT Objectives Animal studies have indicated that fat intake mediates amygdala activation, which in turn promotes fat intake, while amygdala activation increases the preference for fat and leads to increased fat intake. However, the association among fat intake, amygdala activation, and appetite for high-calorie foods in humans remains unclear. Thus, to examine this association, we conducted a functional magnetic resonance imaging (fMRI) experiment. Methods Fifty healthy-weight adults (18 females; mean age: 22.9 ± 3.02 years) were included. Participants were shown images of high-calorie and low-calorie foods and were instructed to rate their desire to eat the food items during fMRI. All participants provided information on their daily fat intake using a self-reported questionnaire. Associations among fat intake, the desire to eat high-calorie or low-calorie food items, and amygdala responses to food items were examined. Results The basolateral amygdala (BLA) response was positively associated with fat intake ([x, y, z] = [24, −6, −16], z = 3.91, pFWE-corrected = 0.007) and the desire to eat high-calorie food items ([26, −4, −16], z = 3.75, pFWE-corrected = 0.010). Structural equation modeling showed that the desire for high-calorie food items was predicted by BLA response to high-calorie food items (p = 0.013, β = 3.176), and BLA response was predicted by fat intake (p < 0.001, β = 0.026). Discussion Fat intake influences BLA response to high-fat food, which in turn increases the desire to eat palatable high-fat food. This may lead to additional fat intake and increase the risk of weight gain.

Evaluation of Acceptability, Functionality, and Validity of a Passive Image-Based Dietary Intake Assessment Method in Adults and Children of Ghanaian and Kenyan Origin Living in London, UK.

Accurate estimation of dietary intake is challenging. However, whilst some progress has been made in high-income countries, low- and middle-income countries (LMICs) remain behind, contributing to critical nutritional data gaps. This study aimed to validate an objective, passive image-based dietary intake assessment method against weighed food records in London, UK, for onward deployment to LMICs. Wearable camera devices were used to capture food intake on eating occasions in 18 adults and 17 children of Ghanaian and Kenyan origin living in London. Participants were provided pre-weighed meals of Ghanaian and Kenyan cuisine and camera devices to automatically capture images of the eating occasions. Food images were assessed for portion size, energy, nutrient intake, and the relative validity of the method compared to the weighed food records. The Pearson and Intraclass correlation coefficients of estimates of intakes of food, energy, and 19 nutrients ranged from 0.60 to 0.95 and 0.67 to 0.90, respectively. Bland-Altman analysis showed good agreement between the image-based method and the weighed food record. Under-estimation of dietary intake by the image-based method ranged from 4 to 23%. Passive food image capture and analysis provides an objective assessment of dietary intake comparable to weighed food records.

Evaluating machine learning technologies for food computing from a data set perspective

Food plays an important role in our lives that goes beyond mere sustenance. Food affects behavior, mood, and social life. It has recently become an important focus of multimedia and social media applications. The rapid increase of available image data and the fast evolution of artificial intelligence, paired with a raised awareness of people’s nutritional habits, have recently led to an emerging field attracting significant attention, called food computing, aimed at performing automatic food analysis. Food computing benefits from technologies based on modern machine learning techniques, including deep learning, deep convolutional neural networks, and transfer learning. These technologies are broadly used to address emerging problems and challenges in food-related topics, such as food recognition, classification, detection, estimation of calories and food quality, dietary assessment, food recommendation, etc. However, the specific characteristics of food image data, like visual heterogeneity, make the food classification task particularly challenging. To give an overview of the state of the art in the field, we surveyed the most recent machine learning and deep learning technologies used for food classification with a particular focus on data aspects. We collected and reviewed more than 100 papers related to the usage of machine learning and deep learning for food computing tasks. We analyze their performance on publicly available state-of-art food data sets and their potential for usage in multimedia food-related applications for various needs (communication, leisure, tourism, blogging, reverse engineering, etc.). In this paper, we perform an extensive review and categorization of available data sets: to this end, we developed and released an open web resource in which the most recent existing food data sets are collected and mapped to the corresponding geographical regions. Although artificial intelligence methods can be considered mature enough to be used in basic food classification tasks, our analysis of the state-of-the-art reveals that challenges related to the application of this technology need to be addressed. These challenges include, among others: poor representation of regional gastronomy, incorporation of adaptive learning schemes, and reverse engineering for automatic food creation and replication.

Emotional eating: elusive or evident? Integrating laboratory, psychometric and daily life measures

PurposeEmotional eating (EE) refers to eating in response to (negative) emotions. Evidence for the validity of EE is mixed: some meta-analyses find EE only in eating disordered patients, others only in restrained eaters, which suggest that only certain subgroups show EE. Furthermore, EE measures from lab-based assessments, ecological momentary assessment (EMA), and psychometric measures often diverge. This paper tested whether the covariance of these three different EE methods can be modeled through a single latent variable (factorial validity), and if so, how this variable would relate to restrained eating (construct validity), Body-Mass-Index (BMI), and subclinical eating disorder symptomatology (concurrent validity).Methods102 non-eating disordered female participants with a wide BMI range completed EE measures from three methods: psychometric questionnaires, a laboratory experiment (craving ratings of food images in induced neutral vs. negative emotion) and EMA questionnaires (within-participant correlations of momentary negative emotions and momentary food cravings across 9 days). Two measures for each method were extracted and submitted to confirmatory factor analysis.ResultsA one-factor model provided a good fit. The resulting EElat factor correlated positively with subclinical eating disorder symptoms and BMI but not with restrained eating.ConclusionsThe one-factor solution shows that the EE construct can validly be assessed with three different methods. Individual differences in EE are supported by the data and are related to eating and weight problem symptomatology but not to restrained eating. This supports learning accounts of EE and underscores the relevance of the EE construct to physical and mental health.Level of evidenceII (Evidence obtained from well-designed controlled trials without randomization).

Eating Disorders and Diabetes: Facing the Dual Challenge.

Eating disorders and diabetes mellitus are distinct yet closely linked health conditions, presenting distinct challenges in terms of care and management. Eating disorders encompass a spectrum of mental health disorders characterized by abnormal eating behaviors and disruptions in weight regulation. Research indicates that individuals with diabetes might be at an elevated risk of developing eating disorders. The necessity to adhere to specific dietary guidelines, monitor blood sugar levels vigilantly, and manage drug administration can collectively contribute to the emergence of detrimental attitudes toward food and body image. On the other hand, incorrect eating behaviors such as binge eating and purging can disrupt blood sugar control, significantly impacting the development and management of diabetes. This intricate relationship emphasizes the crucial necessity for a comprehensive understanding and specialized care to effectively address the dual challenges faced by individuals dealing with both diabetes and eating disorders. This paper represents the inaugural comprehensive review delving into the intricate connection between eating disorders and diabetes, thereby illuminating previously under-researched areas. The insights gleaned from this review may contribute to developing integrated interventions that aim to improve the overall well-being and quality of life for individuals grappling with the complexities of eating disorders and diabetes.

The effects of virtual reality environmental enrichments on craving to food in healthy volunteers

RationaleEnvironmental enrichment (EE) is a non-pharmacological approach that has been shown to be effective in reducing food-taking in rats. Studies in human volunteers are still in their infancy, given the difficulty to translate the complexity of EE in clinical practice. Virtual reality (VR) is a promising methodological approach, but no study has yet applied it to model and test EE in humans.ObjectivesThe present study is the first to assess the effects of virtual EE on craving for palatable food.MethodsEighty-one healthy volunteers (43 women) were divided into three groups: (i) exposure to a virtual EE (VR-EE), (ii) exposure to a virtual neutral environment (VR-NoEE), and (iii) without exposure to VR (No VR). Craving for palatable food at basal level and evoked by neutral and palatable food images was assessed before and after the VR simulation. Behavior during VR exposure and subjective measures related to the experience were also collected.ResultsVR-EE group showed a significantly greater decrease in pre-post craving difference compared to No VR for all assessments and at basal level compared to VR-NoEE. Interestingly, an inverse correlation between craving and deambulation in the VR simulation emerged in VR-EE group only.ConclusionsThe study highlighted the feasibility of exposing human subjects to an EE as a virtual simulation. Virtual EE induced effects on basal craving for food that suggest the potential for further improvements of the protocol to extend its efficacy to palatable food cues.

No evidence of compensatory changes in energy balance, despite reductions in body weight and liver fat, during dapagliflozin treatment in type 2 diabetes mellitus: A randomized, double-blind, placebo-controlled, cross-over trial (ENERGIZE).

This study assessed the impact of dapagliflozin on food intake, eating behaviour, energy expenditure, magnetic resonance imaging (MRI)-determined brain response to food cues and body composition in patients with type 2 diabetes mellitus (T2D). Patients were given dapagliflozin 10 mg once daily in a randomized, double-blind, placebo-controlled trial with short-term (1 week) and long-term (12 weeks) cross-over periods. The primary outcome was the difference in test meal food intake between long-term dapagliflozin and placebo treatment. Secondary outcomes included short-term differences in test meal food intake, short- and long-term differences in appetite and eating rate, energy expenditure and functional MRI brain activity in relation to food images. We determined differences in glycated haemoglobin, weight, liver fat (by 1 H magnetic resonance spectroscopy) and subcutaneous/visceral adipose tissue volumes (by MRI). In total, 52 patients (43% were women) were randomized; with the analysis of 49 patients: median age 58 years, weight 99.1 kg, body mass index 35 kg/m2 , glycated haemoglobin 49 mmol/mol. Dapagliflozin reduced glycated haemoglobin by 9.7 mmol/mol [95% confidence interval (CI) 3.91-16.27, p = .004], and body weight (-2.84 vs. -0.87 kg) versus placebo. There was no short- or long-term difference in test meal food intake between dapagliflozin and placebo [mean difference 5.7 g (95% CI -127.9 to 139.3, p = .933); 15.8 g (95% CI -147.7 to 116.1, p = .813), respectively] nor in the rate of eating, energy expenditure, appetite, or brain responses to food cues. Liver fat (median reduction -4.7 vs. 1.95%), but not subcutaneous/visceral adipose tissue, decreased significantly with 12 weeks of dapagliflozin. The reduction in body weight and liver fat with dapagliflozin was not associated with compensatory adaptations in food intake or energy expenditure.