To generate evidence regarding the safety and efficacy of artificial intelligence (AI) enabled medical devices, AI models need to be evaluated on a diverse population of patient cases, some of which may not be readily available. We propose an evaluation approach for testing medical imaging AI models that relies on in silico imaging pipelines in which stochastic digital models of human anatomy (in object space) with and without pathology are imaged using a digital replica imaging acquisition system to generate realistic synthetic image datasets. Here, we release M-SYNTH, a dataset of cohorts with four breast fibroglandular density distributions imaged at different exposure levels using Monte Carlo x-ray simulations with the publicly available Virtual Imaging Clinical Trial for Regulatory Evaluation (VICTRE) toolkit. We utilize the synthetic dataset to analyze AI model performance and find that model performance decreases with increasing breast density and increases with higher mass density, as expected. As exposure levels decrease, AI model performance drops with the highest performance achieved at exposure levels lower than the nominal recommended dose for the breast type.

Current medical imaging increasingly relies on 3D volumetric data making it difficult for radiologists to thoroughly search all regions of the volume. In some applications (e.g., Digital Breast Tomosynthesis), the volumetric data is typically paired with a synthesized 2D image (2D-S) generated from the corresponding 3D volume. We investigate how this image pairing affects the search for spatially large and small signals. Observers searched for these signals in 3D volumes, 2D-S images, and while viewing both. We hypothesize that lower spatial acuity in the observers' visual periphery hinders the search for the small signals in the 3D images. However, the inclusion of the 2D-S guides eye movements to suspicious locations, improving the observer's ability to find the signals in 3D. Behavioral results show that the 2D-S, used as an adjunct to the volumetric data, improves the localization and detection of the small (but not large) signal compared to 3D alone. There is a concomitant reduction in search errors as well. To understand this process at a computational level, we implement a Foveated Search Model (FSM) that executes human eye movements and then processes points in the image with varying spatial detail based on their eccentricity from fixations. The FSM predicts human performance for both signals and captures the reduction in search errors when the 2D-S supplements the 3D search. Our experimental and modeling results delineate the utility of 2D-S in 3D search-reduce the detrimental impact of low-resolution peripheral processing by guiding attention to regions of interest, effectively reducing errors.

Augmented reality (AR) blends the digital and physical worlds by overlapping a virtual image onto the see-through physical environment. However, contrast reduction and noise superposition in an AR head-mounted display (HMD) can substantially limit image quality and human perceptual performance in both the digital and physical spaces. To assess image quality in AR, we performed human and model observer studies for various imaging tasks with targets placed in the digital and physical worlds. A target detection model was developed for the complete AR system including the optical see-through. Target detection performance using different observer models developed in the spatial frequency domain was compared with the human observer results. The non-prewhitening model with eye filter and internal noise results closely track human perception performance as measured by the area under the receiver operating characteristic curve (AUC), especially for tasks with high image noise. The AR HMD non-uniformity limits the low-contrast target (less than 0.02) observer performance for low image noise. In augmented reality conditions, the detectability of a target in the physical world is reduced due to the contrast reduction by the overlaid AR display image (AUC less than 0.87 for all the contrast levels evaluated). We propose an image quality optimization scheme to optimize the AR display configurations to match observer detection performance for targets in both the digital and physical worlds. The image quality optimization procedure is validated using both simulation and bench measurements of a chest radiography image with digital and physical targets for various imaging configurations.

Virtual clinical trials (VCTs) of medical imaging require realistic models of human anatomy. For VCTs in breast imaging, a multi-scale Perlin noise method is proposed to simulate anatomical structures of breast tissue in the context of an ongoing breast phantom development effort. Four Perlin noise distributions were used to replace voxels representing the tissue compartments and Cooper’s ligaments in the breast phantoms. Digital mammography and tomosynthesis projections were simulated using a clinical DBT system configuration. Power-spectrum analyses and higher-order statistics properties using Laplacian fractional entropy (LFE) of the parenchymal texture are presented. These objective measures were calculated in phantom and patient images using a sample of 140 clinical mammograms and 500 phantom images. Power-law exponents were calculated using the slope of the curve fitted in the low frequency [0.1, 1.0] mm−1region of the power spectrum. The results show that the images simulated with our prior and proposed Perlin method have similar power-law spectra when compared with clinical mammograms. The power-law exponents calculated are −3.10, −3.55, and −3.46, for the log-power spectra of patient, prior phantom and proposed phantom images, respectively. The results also indicate an improved agreement between the mean LFE estimates of Perlin-noise based phantoms and patients than our prior phantoms and patients. Thus, the proposed method improved the simulation of anatomic noise substantially compared to our prior method, showing close agreement with breast parenchyma measures.

Model observers have a long history of success in predicting human observer performance in clinically-relevant detection tasks. New 3D image modalities provide more signal information but vastly increase the search space to be scrutinized. Here, we compared standard linear model observers (ideal observers, non-pre-whitening matched filter with eye filter, and various versions of Channelized Hotelling models) to human performance searching in 3D 1/f2.8filtered noise images and assessed its relationship to the more traditional location known exactly detection tasks and 2D search. We investigated two different signal types that vary in their detectability away from the point of fixation (visual periphery). We show that the influence of 3D search on human performance interacts with the signal's detectability in the visual periphery. Detection performance for signals difficult to detect in the visual periphery deteriorates greatly in 3D search but not in 3D location known exactly and 2D search. Standard model observers do not predict the interaction between 3D search and signal type. A proposed extension of the Channelized Hotelling model (foveated search model) that processes the image with reduced spatial detail away from the point of fixation, explores the image through eye movements, and scrolls across slices can successfully predict the interaction observed in humans and also the types of errors in 3D search. Together, the findings highlight the need for foveated model observers for image quality evaluation with 3D search.