Search for: All records

The burgeoning field of brain health research increasingly leverages artificial intelligence (AI) to analyze and interpret neuroimaging data. Medical foundation models have shown promise of superior performance with better sample efficiency. This work introduces a novel approach towards creating 3-dimensional (3D) medical foundation models for multimodal neuroimage segmentation through self-supervised training. Our approach involves a novel two-stage pretraining approach using vision transformers. The first stage encodes anatomical structures in generally healthy brains from the large-scale unlabeled neuroimage dataset of multimodal brain magnetic resonance imaging (MRI) images from 41,400 participants. This stage of pertaining focuses on identifying key features such as shapes and sizes of different brain structures. The second pretraining stage identifies disease-specific attributes, such as geometric shapes of tumors and lesions and spatial placements within the brain. This dual-phase methodology significantly reduces the extensive data requirements usually necessary for AI model training in neuroimage segmentation with the flexibility to adapt to various imaging modalities. We rigorously evaluate our model, BrainSegFounder, using the Brain Tumor Segmentation (BraTS) challenge and Anatomical Tracings of Lesions After Stroke v2.0 (ATLAS v2.0) datasets. BrainSegFounder demonstrates a significant performance gain, surpassing the achievements of the previous winning solutions using fully supervised learning. Our findings underscore the impact of scaling up both the model complexity and the volume of unlabeled training data derived from generally healthy brains. Both of these factors enhance the accuracy and predictive capabilities of the model in neuroimage segmentation tasks. Our pretrained models and code are at https://github.com/lab-smile/BrainSegFounder. 

The International Affective Picture System (IAPS) contains 1,182 well-characterized photographs depicting natural scenes varying in affective content. These pictures are used extensively in affective neuroscience to investigate the neural correlates of emotional processing. Recently, in an effort to augment this dataset, we have begun to generate synthetic emotional images by combining IAPS pictures and diffusion-based AI models. The goal of this study is to compare the neural responses to IAPS pictures and matching AI-generated images. The stimulus set consisted of 60 IAPS pictures (20 pleasant, 20 neutral, 20 unpleasant) and 60 matching AI-generated images (20 pleasant, 20 neutral, 20 unpleasant). In a recording session, a total of 30 IAPS pictures and 30 matching AI-generated images were presented in random order, where each image was displayed for 3 seconds with neighboring images being separated by an interval of 2.8 to 3.5 seconds. Each experiment consisted of 10 recording sessions. The fMRI data was recorded on a 3T Siemens Prisma scanner. Pupil responses to image presentation were monitored using an MRI-compatible eyetracker. Our preliminary analysis of the fMRI data (N=3) showed that IAPS pictures and matching AI-generated images evoked similar neural responses in the visual cortex. In particular, MVPA (Multivariate Pattern Analysis) classifiers built to decode emotional categories from neural responses to IAPS pictures can be used to decode emotional categories from neural responses to AI-generated images and vice versa. Efforts to confirm these findings are underway by recruiting additional participants. Analysis is also being expanded to include the comparison of such measures as functional connectivity and pupillometry. 

Recent neuroimaging studies have shown that the visual cortex plays an important role in representing the affective significance of visual input. The origin of these affect-specific visual representations is debated: they are intrinsic to the visual system versus they arise through reentry from frontal emotion processing structures such as the amygdala. We examined this problem by combining convolutional neural network (CNN) models of the human ventral visual cortex pre-trained on ImageNet with two datasets of affective images. Our results show that in all layers of the CNN models, there were artificial neurons that responded consistently and selectively to neutral, pleasant, or unpleasant images and lesioning these neurons by setting their output to zero or enhancing these neurons by increasing their gain led to decreased or increased emotion recognition performance respectively. These results support the idea that the visual system may have the intrinsic ability to represent the affective significance of visual input and suggest that CNNs offer a fruitful platform for testing neuroscientific theories.