In this work we propose a machine-learning MO based on Naive-Bayes classification (NB-MO) for the diagnostic tasks of detection, localization and assessment of perfusion defects in clinical SPECT Myocardial Perfusion Imaging (MPI), with the goal of evaluating several image reconstruction methods used in clinical practice. NB-MO uses image features extracted from polar-maps in order to predict lesion detection, localization and severity scores given by human readers in a series of 3D SPECT-MPI. The population used to tune (i.e. train) the NB-MO consisted of simulated SPECT-MPI cases – divided into normals or with lesions in variable sizes and locations – reconstructed using filtered backprojection (FBP) method. An ensemble of five human specialists (physicians) read a subset of simulated reconstructed images, and assigned a perfusion score for each region of the left-ventricle (LV). Polar-maps generated from the simulated volumes along with their corresponding human scores were used to train five NB-MOs (one per human reader), which are subsequently applied (i.e. tested) on three sets of clinical SPECT-MPI polar maps, in order to predict human detection and localization scores. The clinical “testing” population comprises healthy individuals and patients suffering from coronary artery disease (CAD) in three possible regions, namely: LAD, LcX and RCA. Each clinical case was reconstructed using three reconstruction strategies, namely: FBP with no SC (i.e. scatter compensation), OSEM with Triple Energy Window (TEW) SC method, and OSEM with Effective Source Scatter Estimation (ESSE) SC. Alternative Free-Response (AFROC) analysis of perfusion scores shows that NB-MO predicts a higher human performance for scatter-compensated reconstructions, in agreement with what has been reported in published literature. These results suggest that NB-MO has good potential to generalize well to reconstruction methods not used during training, even for reasonably dissimilar datasets (i.e. simulated vs. clinical).
Cardiac-gated MRI is widely used for the task of measuring parameters related to heart motion. More specifically, gated tagged MRI is the preferred modality to estimate local deformation (strain) and rotational motion (twist) of myocardial tissue. Many methods have been proposed to estimate cardiac motion from gated MRI sequences. However, when dealing with clinical data, evaluation of these methods is problematic due to the absence of gold-standards for cardiac motion. To overcome that, a linear regression scheme known as regression-without-truth (RWT) was proposed in the past. RWT uses priors to model the distribution of true values, thus enabling us to assess image-analysis algorithms without knowledge of the ground-truth. Furthermore, it allows one to rank methods by means of an objective figure-of-merit γ (i.e. precision). In this work we apply RWT to compare the performance of several gated MRI motion-tracking methods (e.g. non-rigid registration, feature based, harmonic phase) at the task of estimating myocardial strain and left-ventricle (LV) twist, from a population of 18 clinical human cardiac-gated tagged MRI studies.
Model observers (MO) are widely used in medical imaging to act as surrogates of human observers in task-based image quality evaluation, frequently towards optimization of reconstruction algorithms. In SPECT myocardial perfusion imaging (MPI), a realistic task-based approach involves detection and localization of perfusion defects, as well as a subsequent assessment of defect severity. In this paper we explore a machine-learning MO based on Naive- Bayes classification (NB-MO). NB-MO uses a set of polar-map image features to predict lesion detection, localization and severity scores given by five human readers for a set of simulated 3D SPECT-MPI patients. The simulated dataset included lesions with different sizes, perfusion-reduction ratios, and locations. Simulated projections were reconstructed using two readily used methods namely: FBP and OSEM. For validation, a multireader multi-case (MRMC) analysis of alternative free-response ROC (AFROC) curve was performed for NB-MO and human observers. For comparison, we also report performances of a statistical Hotelling Observer applied on polar-map images. Results show excellent agreement between NB-MO and humans, as well as model’s good generalization between different reconstruction treatments.