Optical coherence tomography (OCT) retinal volumes are prone to motion artifacts due to the movement of the eye during acquisition. Current retrospective motion correction algorithms are either computationally expensive or limited to pair-wise formulations, based on registration of consecutive slices (B-scans). This type of approach can lead to errors when individual B-scans contain artifacts or lack sufficient signal. Instead, we propose a framework, based on unsupervised deep learning, that corrects motion by aligning groups of consecutive B-scans. The network architecture is fully-convolutional and, thus, it can perform inference on the entire OCT volume, even though it was trained on groups of smaller size. Moreover, we improved performance by inferring in a multi-shot recurrent manner, which was further leveraged by a novel data augmentation technique. We used an exhaustive search algorithm (brute-force) to compare the proposed method against, both quantitatively and qualitatively based on visual assessment. In a dataset of 146 (training: 106, validation: 40) macula and optic disc volumes from 19 healthy subjects, our best performing configuration achieved 72% reduction in registration errors compared to the exhaustive search algorithm, with a computation time of 2.35 seconds. These results demonstrated that our framework has the potential to provide a fast and robust solution, based on deep learning registration, for the motion correction of OCT images.
Although computed tomography (CT) perfusion (CTP) imaging enables rapid diagnosis and prognosis of ischemic stroke, current CTP analysis methods have several shortcomings. We propose a fast nonlinear regression method with a box-shaped model (boxNLR) that has important advantages over the current state-of-the-art method, block-circulant singular value decomposition (bSVD). These advantages include improved robustness to attenuation curve truncation, extensibility, and unified estimation of perfusion parameters. The method is compared with bSVD and with a commercial SVD-based method. The three methods were quantitatively evaluated by means of a digital perfusion phantom, described by Kudo et al. and qualitatively with the aid of 50 clinical CTP scans. All three methods yielded high Pearson correlation coefficients (>0.9) with the ground truth in the phantom. The boxNLR perfusion maps of the clinical scans showed higher correlation with bSVD than the perfusion maps from the commercial method. Furthermore, it was shown that boxNLR estimates are robust to noise, truncation, and tracer delay. The proposed method provides a fast and reliable way of estimating perfusion parameters from CTP scans. This suggests it could be a viable alternative to current commercial and academic methods.
Assessment of the extent of cerebral damage on admission in patients with acute ischemic stroke could play an important role in treatment decision making. Computed tomography perfusion (CTP) imaging can be used to determine the extent of damage. However, clinical application is hindered by differences among vendors and used methodology. As a result, threshold based methods and visual assessment of CTP images has not yet shown to be useful in treatment decision making and predicting clinical outcome. Preliminary results in MR studies have shown the benefit of using supervised classifiers for predicting tissue outcome, but this has not been demonstrated for CTP. We present a novel method for the automatic prediction of tissue outcome by combining multi-parametric CTP images into a tissue outcome probability map. A supervised classification scheme was developed to extract absolute and relative perfusion values from processed CTP images that are summarized by a trained classifier into a likelihood of infarction. Training was performed using follow-up CT scans of 20 acute stroke patients with complete recanalization of the vessel that was occluded on admission. Infarcted regions were annotated by expert neuroradiologists. Multiple classifiers were evaluated in a leave-one-patient-out strategy for their discriminating performance using receiver operating characteristic (ROC) statistics. Results showed that a RandomForest classifier performed optimally with an area under the ROC of 0.90 for discriminating infarct tissue. The obtained results are an improvement over existing thresholding methods and are in line with results found in literature where MR perfusion was used.
CT plays an important role in the diagnosis of acute stroke patients. Dynamic contrast enhanced CT (DCE-CT) can estimate local tissue perfusion and extent of ischemia. However, hemodynamic information of the large intracranial vessels may also be obtained from DCE-CT data and may contain valuable diagnostic information. We describe a novel method to estimate intravascular blood velocity (IBV) in large cerebral vessels using DCE-CT data, which may be useful to help predict stroke outcome. DCE-CT scans from 34 patients with isolated M1 occlusions were included from a large prospective multi-center cohort study of patients with acute ischemic stroke. Gaussians fitted to the intravascular data yielded the time-to-peak (TTP) and cerebral-blood-volume (CBV). IBV was computed by taking the inverse of the TTP gradient magnitude. Voxels with a CBV of at least 10% of the CBV found in the arterial input function were considered part of a vessel. Mid-sagittal planes were drawn manually and averages of the IBV over all vessel-voxels (arterial and venous) were computed for each hemisphere. Mean-hemisphere IBV differences, mean-hemisphere TTP differences, and hemisphere vessel volume differences were used to differentiate between patients with good and bad outcome (modified Rankin Scale score <3 versus ≥3 at 90 days) using ROC analysis. AUCs from the ROC for IBV, TTP, and vessel volume were 0.80, 0.67 and 0.62 respectively. In conclusion, IBV was found to be a better predictor of patient outcome than the parameters used to compute it and may be a promising new parameter for stroke outcome prediction.
CT perfusion (CTP) imaging allows for rapid diagnosis of ischemic stroke. Generation of perfusion maps from CTP data usually involves deconvolution algorithms providing estimates for the impulse response function in the tissue. We propose the use of a fast non-linear regression (NLR) method that we postulate has similar performance to the current academic state-of-art method (bSVD), but that has some important advantages, including the estimation of vascular permeability, improved robustness to tracer-delay, and very few tuning parameters, that are all important in stroke assessment. The aim of this study is to evaluate the fast NLR method against bSVD and a commercial clinical state-of-art method. The three methods were tested against a published digital perfusion phantom earlier used to illustrate the superiority of bSVD. In addition, the NLR and clinical methods were also tested against bSVD on 20 clinical scans. Pearson correlation coefficients were calculated for each of the tested methods. All three methods showed high correlation coefficients (>0.9) with the ground truth in the phantom. With respect to the clinical scans, the NLR perfusion maps showed higher correlation with bSVD than the perfusion maps from the clinical method. Furthermore, the perfusion maps showed that the fast NLR estimates are robust to tracer-delay. In conclusion, the proposed fast NLR method provides a simple and flexible way of estimating perfusion parameters from CT perfusion scans, with high correlation coefficients. This suggests that it could be a better alternative to the current clinical and academic state-of-art methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.