Translating AI to Clinical Practice: Overcoming Data Shift with Explainability.

To translate artificial intelligence (AI) algorithms into clinical practice requires generalizability of models to real-world data. One of the main obstacles to generalizability is data shift, a data distribution mismatch between model training and real environments. Explainable AI techniques offer tools to detect and mitigate the data shift problem and develop reliable AI for clinical practice. Most medical AI is trained with datasets gathered from limited environments, such as restricted disease populations and center-dependent acquisition conditions. The data shift that commonly exists in the limited training set often causes a significant performance decrease in the deployment environment. To develop a medical application, it is important to detect potential data shift and its impact on clinical translation. During AI training stages, from premodel analysis to in-model and post hoc explanations, explainability can play a key role in detecting model susceptibility to data shift, which is otherwise hidden because the test data have the same biased distribution as the training data. Performance-based model assessments cannot effectively distinguish the model overfitting to training data bias without enriched test sets from external environments. In the absence of such external data, explainability techniques can aid in translating AI to clinical practice as a tool to detect and mitigate potential failures due to data shift. ©RSNA, 2023 Quiz questions for this article are available in the supplemental material.

Retinal tissue oxygen tension and consumption during light flicker stimulation in rat.

Light flicker stimulation has been shown to increase inner retinal oxygen metabolism and supply. The purpose of the study was to test the hypothesis that sustained light flicker stimulation of various durations alters the depth profile metrics of oxygen partial pressure in the retinal tissue (tPO2) but not the outer retinal oxygen consumption rate (QO2). In 17 rats, tPO2 depth profiles were derived by phosphorescence lifetime imaging after intravitreal injection of an oxyphor. tPO2 profile metrics, including mean inner retinal tPO2, maximum outer retinal tPO2 and minimum outer retinal tPO2 were determined. QO2 was calculated using a one-dimensional oxygen diffusion model. Data were acquired at baseline (constant light illumination) and during light flicker stimulation at 10 Hz under the same mean illumination levels, and differences between values obtained during flicker and baseline were calculated. None of the tPO2 profile metrics or QO2 differences depended on the duration of light flicker stimulation (R2 ≤ 0.03). No significant change in any of the tPO2 profile metrics was detected with light flicker compared with constant light (P ≥ 0.08). Light flicker decreased QO2 from 0.53 ±â€¯0.29 to 0.38 ±â€¯0.30 mL O2/(min*100 gm), a reduction of 28% (P = 0.02). The retinal compensatory responses to the physiologic challenge of light flicker stimulation were effective in maintaining the levels of oxygen at or near baseline in the inner retina. Oxygen availability to the inner retina during light flicker may also have been enhanced by the decrease in QO2.

The effect of intravitreal vascular endothelial growth factor on inner retinal oxygen delivery and metabolism in rats.

Vascular endothelial growth factor (VEGF) is stimulated by hypoxia and plays an important role in pathologic vascular leakage and neovascularization. Increased VEGF may affect inner retinal oxygen delivery (DO2) and oxygen metabolism (MO2), however, quantitative information is lacking. We tested the hypotheses that VEGF increases DO2, but does not alter MO2. In 10 rats, VEGF was injected intravitreally into one eye, whereas balanced salt solution (BSS) was injected into the fellow eye, 24 h prior to imaging. Vessel diameters and blood velocities were determined by red-free and fluorescent microsphere imaging, respectively. Vascular PO2 values were derived by phosphorescence lifetime imaging of an intravascular oxyphor. Retinal blood flow, vascular oxygen content, DO2 and MO2 were calculated. Retinal arterial and venous diameters were larger in VEGF-injected eyes compared to control eyes (P < 0.03), however no significant difference was observed in blood velocity (P = 0.21). Thus, retinal blood flow was greater in VEGF-injected eyes (P = 0.007). Retinal vascular PO2 and oxygen content were similar between control and VEGF-injected eyes (P > 0.11), while the arteriovenous oxygen content difference was marginally lower in VEGF-injected eyes (P = 0.05). DO2 was 950 ± 340 and 1380 ± 650 nL O2/min in control and VEGF-injected eyes, respectively (P = 0.005). MO2 was 440 ± 150 and 490 ± 190 nL O2/min in control and VEGF-injected eyes, respectively (P = 0.31). Intravitreally administered VEGF did not alter MO2 but increased DO2, suggesting VEGF may play an offsetting role in conditions characterized by retinal hypoxia.

Correspondence of retinal thinning and vasculopathy in mice with oxygen-induced retinopathy.

The aberrantly vascularized peripheral retina in retinopathy of prematurity (ROP) may be associated with visual field constriction, retinal dysfunction, and abnormalities in retinal thickness which is commonly assessed by spectral domain optical coherence tomography (SDOCT). However, due to the limitation of SDOCT for peripheral retinal imaging, retinal thickness in avascular peripheral retina in ROP has not been evaluated. Oxygen induced retinopathy (OIR) in mice has features of vasculopathy similar to those in human ROP. These features occur in the posterior retina and thereby are accessible by standard imaging methods. The purpose of the current study was to determine the correspondence between abnormalities in retinal thickness and vasculopathy in neonatal OIR mice by simultaneous SDOCT imaging and fluorescein angiography (FA). Newborn mice (N = 19; C57BL/6J strain) were exposed to 77% oxygen from postnatal day 7 (P7) to P12. Age-matched control mice (N = 12) were raised in room air. FA and SDOCT were performed in mice between P17 and P19 to visualize retinal vasculature and measure retinal thickness, respectively. Retinal thickness measurements in vascular regions of interest (ROIs) of control mice, and in hypovascular and avascular ROIs of OIR mice were compared. In control mice, FA showed uniformly dense retinal capillary networks between major retinal vessels and retinal thickness of vascular ROIs was 260 ± 7 µm (N = 12). In OIR mice, FA displayed hypovascular regions with less dense and fewer capillaries and avascular regions devoid of visible capillaries. Retinal thickness measurements of hypovascular and avascular ROIs were 243 ± 21 µm and 209 ± 11 µm (N = 19), respectively. Retinal thickness in hypovascular and avascular ROIs of OIR mice was significantly lower than in vascular ROIs of control mice (p ≤ 0.01). Likewise, retinal thickness in avascular ROIs was significantly lower than in hypovascular ROIs (p < 0.001). Retinal thinning in hypovascular and avascular regions may be due to arrested retinal development and/or ischemia induced apoptosis.

Quantifying lung fissure integrity using a three-dimensional patch-based convolutional neural network on CT images for emphysema treatment planning.

Purpose: Evaluation of lung fissure integrity is required to determine whether emphysema patients have complete fissures and are candidates for endobronchial valve (EBV) therapy. We propose a deep learning (DL) approach to segment fissures using a three-dimensional patch-based convolutional neural network (CNN) and quantitatively assess fissure integrity on CT to evaluate it in subjects with severe emphysema. Approach: From an anonymized image database of patients with severe emphysema, 129 CT scans were used. Lung lobe segmentations were performed to identify lobar regions, and the boundaries among these regions were used to construct approximate interlobar regions of interest (ROIs). The interlobar ROIs were annotated by expert image analysts to identify voxels where the fissure was present and create a reference ROI that excluded non-fissure voxels (where the fissure is incomplete). A CNN configured by nnU-Net was trained using 86 CT scans and their corresponding reference ROIs to segment the ROIs of left oblique fissure (LOF), right oblique fissure (ROF), and right horizontal fissure (RHF). For an independent test set of 43 cases, fissure integrity was quantified by mapping the segmented fissure ROI along the interlobar ROI. A fissure integrity score (FIS) was then calculated as the percentage of labeled fissure voxels divided by total voxels in the interlobar ROI. Predicted FIS (p-FIS) was quantified from the CNN output, and statistical analyses were performed comparing p-FIS and reference FIS (r-FIS). Results: The absolute percent error mean (±SD) between r-FIS and p-FIS for the test set was 4.0% (±4.1%), 6.0% (±9.3%), and 12.2% (±12.5%) for the LOF, ROF, and RHF, respectively. Conclusions: A DL approach was developed to segment lung fissures on CT images and accurately quantify FIS. It has potential to assist in the identification of emphysema patients who would benefit from EBV treatment.

Multi-scale, domain knowledge-guided attention + random forest: a two-stage deep learning-based multi-scale guided attention models to diagnose idiopathic pulmonary fibrosis from computed tomography images.

BACKGROUND: Idiopathic pulmonary fibrosis (IPF) is a progressive, irreversible, and usually fatal lung disease of unknown reasons, generally affecting the elderly population. Early diagnosis of IPF is crucial for triaging patients' treatment planning into anti-fibrotic treatment or treatments for other causes of pulmonary fibrosis. However, current IPF diagnosis workflow is complicated and time-consuming, which involves collaborative efforts from radiologists, pathologists, and clinicians and it is largely subject to inter-observer variability. PURPOSE: The purpose of this work is to develop a deep learning-based automated system that can diagnose subjects with IPF among subjects with interstitial lung disease (ILD) using an axial chest computed tomography (CT) scan. This work can potentially enable timely diagnosis decisions and reduce inter-observer variability. METHODS: Our dataset contains CT scans from 349 IPF patients and 529 non-IPF ILD patients. We used 80% of the dataset for training and validation purposes and 20% as the holdout test set. We proposed a two-stage model: at stage one, we built a multi-scale, domain knowledge-guided attention model (MSGA) that encouraged the model to focus on specific areas of interest to enhance model explainability, including both high- and medium-resolution attentions; at stage two, we collected the output from MSGA and constructed a random forest (RF) classifier for patient-level diagnosis, to further boost model accuracy. RF classifier is utilized as a final decision stage since it is interpretable, computationally fast, and can handle correlated variables. Model utility was examined by (1) accuracy, represented by the area under the receiver operating characteristic curve (AUC) with standard deviation (SD), and (2) explainability, illustrated by the visual examination of the estimated attention maps which showed the important areas for model diagnostics. RESULTS: During the training and validation stage, we observe that when we provide no guidance from domain knowledge, the IPF diagnosis model reaches acceptable performance (AUC±SD = 0.93±0.07), but lacks explainability; when including only guided high- or medium-resolution attention, the learned attention maps are not satisfactory; when including both high- and medium-resolution attention, under certain hyperparameter settings, the model reaches the highest AUC among all experiments (AUC±SD = 0.99±0.01) and the estimated attention maps concentrate on the regions of interests for this task. Three best-performing hyperparameter selections according to MSGA were applied to the holdout test set and reached comparable model performance to that of the validation set. CONCLUSIONS: Our results suggest that, for a task with only scan-level labels available, MSGA+RF can utilize the population-level domain knowledge to guide the training of the network, which increases both model accuracy and explainability.

Feasibility of conjunctival hemodynamic measurements in rabbits: reproducibility, validity, and response to acute hypotension.

OBJECTIVE: To evaluate the feasibility of conjunctival hemodynamic measurements based on assessment of reproducibility, validity, and response to acute hypotension. METHODS: Image sequences of the conjunctival microvasculature of rabbits were captured using a slit lamp biomicroscope under a steady-state condition, after topical administration of phenylephrine, and after intravenous administration of esmolol. Venous hemodynamic parameters (diameter, blood velocity, blood flow, and wall shear stress) were derived. RESULTS: Conjunctival venous diameters ranged from 9 to 34 µm and blood velocities ranged from 0.08 to 0.95 mm/s. Coefficients of variation of venous diameter and blood velocity measurements were, on average, 6% and 14%, respectively. Automated and manual measurements of venous diameter and velocity were highly correlated (R = 0.97; p < 0.001; n = 16). With phenylephrine administration, diameter and velocity were reduced by 21% and 69%, respectively. Following esmolol administration, blood pressure was reduced with a concomitant decrease in velocity, followed by recovery to baseline. Venous blood velocity, flow, and WSS were correlated with blood pressure (R ≥ 0.52; p ≤ 0.01). CONCLUSIONS: The feasibility of quantifying alterations in microvascular hemodynamics in the bulbar conjunctiva was established. The method is of potential value in evaluating microcirculatory hemodynamics related to cardiovascular function.

Oxygen tension and gradient measurements in the retinal microvasculature of rats.

BACKGROUND: Oxygen delivery from the retinal vasculature plays a crucial role in maintaining normal retinal metabolic function. Therefore, measurements of retinal vascular oxygen tension (PO(2)) and PO(2) longitudinal gradients (gPO(2)) along retinal blood vessels may help gain fundamental knowledge of retinal physiology and pathological processes. METHODS: Three-dimensional retinal vascular PO(2) maps were generated in rats by optical section phosphorescence lifetime imaging. A major retinal artery and vein pair, and a smaller blood vessel (microvessel) between them were segmented, and PO(2) along each blood vessel was measured. In each blood vessel, an average PO(2) (mPO(2)) was calculated, and gPO(2) was determined by linear regression analysis. Reproducibility of measurements was assessed by calculating intraclass correlation coefficient (ICC) of repeated measurements. The correlations of mPO(2) and gPO(2) measurements with systemic arterial oxygen tension (P(a)O(2)) and carbon dioxide tension (P(a)CO(2)) was determined. RESULTS: Measurements of mPO(2) and gPO(2) in retinal arteries, microvessels and veins were reproducible (ICC > 0.86; p < 0.01; N = 8), except for retinal arterial gPO(2). Retinal arterial, microvessel and venous mPO(2) were 41 ± 8, 32 ± 8 and 25 ± 7 mmHg, respectively (mean ± SD; N = 27). Retinal arterial mPO(2) was correlated with P(a)O(2) and P(a)CO(2) (R > 0.44; p < 0.03), while retinal microvessel and venous mPO(2) were only correlated with P(a)CO(2) (R > 0.68; p < 0.01). Retinal microvessel gPO(2) (-3.8 ± 1.5 mmHg/100 µm) was significantly steeper (more negative) than venous gPO(2) (0.02 ± 0.43 mmHg/100 µm) (p < 0.01; N = 27), and neither were significantly correlated with P(a)O(2) or P(a)CO(2). CONCLUSIONS: Quantitative measurement of mPO(2) and gPO(2) in the retinal microvasculature was demonstrated. A significant decrease in PO(2) was observed along most retinal microvessels, indicative of substantial oxygen extraction by the retinal tissue. This method has the potential to help elucidate retinal microvascular oxygen transport in health and disease.

High throughput image labeling on chest computed tomography by deep learning.

When mining image data from PACs or clinical trials or processing large volumes of data without curation, the relevant scans must be identified among irrelevant or redundant data. Only images acquired with appropriate technical factors, patient positioning, and physiological conditions may be applicable to a particular image processing or machine learning task. Automatic labeling is important to make big data mining practical by replacing conventional manual review of every single-image series. Digital imaging and communications in medicine headers usually do not provide all the necessary labels and are sometimes incorrect. We propose an image-based high throughput labeling pipeline using deep learning, aimed at identifying scan direction, scan posture, lung coverage, contrast usage, and breath-hold types. They were posed as different classification problems and some of them involved further segmentation and identification of anatomic landmarks. Images of different view planes were used depending on the specific classification problem. All of our models achieved accuracy > 99 % on test set across different tasks using a research database from multicenter clinical trials.

A method for volumetric retinal tissue oxygen tension imaging.

PURPOSE: Inadequate retinal oxygenation occurs in many vision-threatening retinal diseases, including diabetic retinopathy, retinal vascular occlusions, and age-related macular degeneration. Therefore, techniques that assess retinal oxygenation are necessary to understand retinal physiology in health and disease. The purpose of the current study is to report a method for the three-dimensional (3D) imaging of retinal tissue oxygen tension (tPO2) in rats. METHODS: Imaging was performed in Long Evans pigmented rats under systemic normoxia (N = 6) or hypoxia (N = 3). A vertical laser line was horizontally scanned on the retina and a series of optical section phase-delayed phosphorescence images were acquired. From these images, phosphorescence volumes at each phase delay were constructed and a 3D retinal tPO2 volume was generated. Retinal tPO2 volumes were quantitatively analyzed by generating retinal depth profiles of mean tPO2 (MtPO2) and the spatial variation of tPO2 (SVtPO2). The effects of systemic condition (normoxia/hypoxia) and retinal depth on MtPO2 and SVtPO2 were determined by mixed linear model. RESULTS: Each 3D retinal tPO2 volume was approximately 500 × 750 × 200 µm (horizontal × vertical × depth) and consisted of 45 en face tPO2 images through the retinal depth. MtPO2 at the chorioretinal interface was significantly correlated with systemic arterial oxygen tension (P = 0.007; N = 9). There were significant effects of both systemic condition and retinal depth on MtPO2 and SVtPO2, such that both were lower under hypoxia than normoxia and higher in the outer retina than inner retina (P < 0.001). CONCLUSION: For the first time, 3D imaging of retinal tPO2 was demonstrated, with potential future application for assessment of physiological alterations in animal models of retinal diseases.

Response of inner retinal oxygen extraction fraction to light flicker under normoxia and hypoxia in rat.

PURPOSE: Oxygen extraction fraction (OEF), defined by the ratio of oxygen metabolism (MO2) to delivery (DO2), determines the level of compensation of MO2 by DO2. In the current study, we tested the hypothesis that inner retinal OEF remains unchanged during light flicker under systemic normoxia and hypoxia in rats due to the matching of MO2 and DO2. METHODS: Retinal vascular oxygen tension (PO2) measurements were obtained in 10 rats by phosphorescence lifetime imaging. Inner retinal OEF was derived from vascular PO2 based on Fick's principle. Measurements were obtained before and during light flicker under systemic normoxia and hypoxia. The effects of light flicker and systemic oxygenation on retinal vascular PO2 and OEF were determined by ANOVA. RESULTS: During light flicker, retinal venous PO2 decreased (P < 0.01, N = 10), while inner retinal OEF increased (P = 0.02). Under hypoxia, retinal arterial and venous PO2 decreased (P < 0.01), while OEF increased (P < 0.01). The interaction effect was not significant on OEF (P = 0.52), indicating the responses of OEF to light flicker were similar under normoxia and hypoxia. During light flicker, OEF increased from 0.46 ± 0.13 to 0.50 ± 0.11 under normoxia, while under hypoxia, OEF increased from 0.67 ± 0.16 to 0.74 ± 0.14. CONCLUSIONS: Inner retinal OEF increased during light flicker, indicating the relative change in DO2 is less than that in MO2 in rats under systemic normoxia and hypoxia. Inner retinal OEF is a potentially useful parameter for assessment of the relative changes of MO2 and DO2 under physiologic and pathologic conditions.

Inner retinal oxygen delivery and metabolism in streptozotocin diabetic rats.

PURPOSE: The purpose of the study is to report global measurements of inner retinal oxygen delivery (DO2_IR) and oxygen metabolism (MO2_IR) in streptozotocin (STZ) diabetic rats. METHODS: Phosphorescence lifetime and blood flow imaging were performed in rats 4 (STZ/4 wk; n = 10) and 6 (STZ/6 wk; n = 10) weeks following injection of STZ to measure retinal arterial (O2A) and venous (O2V) oxygen contents and total retinal blood flow (F). DO2_IR and MO2_IR were calculated from measurements of F and O2A and of F and the arteriovenous oxygen content difference, respectively. Data in STZ rats were compared to those in healthy control rats (n = 10). RESULTS: Measurements of O2A and O2V were not significantly different among STZ/4 wk, STZ/6 wk, and control rats (P ≥ 0.28). Likewise, F was similar among all groups of rats (P = 0.81). DO2_IR measurements were 941 ± 231, 956 ± 232, and 973 ± 243 nL O2/min in control, STZ/4 wk, and STZ/6 wk rats, respectively (P = 0.95). MO2_IR measurements were 516 ± 175, 444 ± 103, and 496 ± 84 nL O2/min in control, STZ/4 wk, and STZ/6 wk rats, respectively (P = 0.37). CONCLUSIONS: Global inner retinal oxygen delivery and metabolism were not significantly impaired in STZ rats in early diabetes.

Inner retinal oxygen extraction fraction in rat.

PURPOSE: Oxygen extraction fraction (OEF), defined by the ratio of oxygen consumption to delivery, may be a useful parameter for assessing the retinal tissue status under impaired circulation. We report a method for measurement of inner retinal OEF in rats under normoxia and hypoxia based on vascular oxygen tension (PO(2)) imaging. METHODS: Retinal vascular PO(2) measurements were obtained in 10 rats, using our previously developed optical section phosphorescence lifetime imaging system. Inner retinal OEF was derived from retinal vascular PO(2) measurements based on Fick's principle. Measurements of inner retinal OEF obtained under normoxia were compared between nasal and temporal retinal sectors and repeatability was determined. Inner retinal OEF measurements obtained under normoxia and hypoxia were compared. RESULTS: Retinal vascular PO(2) and inner retinal OEF measurements were repeatable (ICC ≥ 0.83). Inner retinal OEF measurements at nasal and temporal retinal sectors were correlated (R = 0.71; P = 0.02; n = 10). Under hypoxia, both retinal arterial and venous PO(2) decreased significantly as compared with normoxia (P < 0.001; n = 10). Inner retinal OEF was 0.46 ± 0.13 under normoxia and increased significantly to 0.67 ± 0.16 under hypoxia (mean ± SD; P < 0.001; n = 10). CONCLUSIONS: Inner retinal OEF is a promising quantitative biomarker for the adequacy of oxygen supply for metabolism under physiologically and pathologically altered conditions.

Inner retinal oxygen delivery and metabolism under normoxia and hypoxia in rat.

PURPOSE: Retinal hypoxia is a common pathological condition usually caused by ischemia that may result in alterations in oxidative energy metabolism. We report measurements of oxygen delivery by the retinal circulation (DO2_IR) and inner retinal oxygen metabolism (MO2_IR) under systemic normoxia and hypoxia in rat. METHODS: Rats were ventilated with fractions of inspired oxygen (FiO2) to induce either normoxia (n = 10), moderate hypoxia (n = 14), or severe hypoxia (n = 10). Oxygen tension was measured in retinal vessels using phosphorescence lifetime imaging and converted to arterial (O2A) and venous (O2V) oxygen contents. Total retinal blood flow (F) was assessed by red-free and fluorescent microsphere imaging. DO2_IR and MO2_IR were calculated as the products of F and O2A, and F and the arteriovenous oxygen content difference (O2A-V), respectively. RESULTS: Measurements of O2A, O2V, and O2A-V were significantly reduced with decreased FiO2 (P < 0.001). In response to reduced oxygen availability, F increased under moderate hypoxia (P < 0.001) but did not increase further under severe hypoxia (P = 0.5). DO2_IR was similar under normoxia and moderate hypoxia (P = 0.7), but significantly lower under severe hypoxia (P < 0.001). Likewise, MO2_IR under normoxia and moderate hypoxia was similar (P = 0.1), but significantly reduced under severe hypoxia (P ≤ 0.02). CONCLUSIONS: DO2_IR and MO2_IR were maintained during moderate hypoxia, but reduced under severe hypoxia, indicating blood flow compensation became insufficient for the reduced oxygen availability. Future studies may aid our understanding of retinal metabolic function in ischemic conditions.

In vivo retinal vascular oxygen tension imaging and fluorescein angiography in the mouse model of oxygen-induced retinopathy.

PURPOSE: Oxygenation abnormalities are implicated in the development of retinopathy of prematurity (ROP). The purpose of this study is to report in vivo retinal vascular oxygen tension (PO2) measurements and fluorescein angiography (FA) findings in the mouse model of oxygen-induced retinopathy (OIR). METHODS: We exposed 19 neonatal mice to 77% oxygen from postnatal day 7 (P7) to P12 (OIR), while 11 neonatal mice were kept under room air (control). Using phosphorescence lifetime imaging, retinal vascular PO2 was measured followed by FA. Repeated measures ANOVA was performed to determine the effects of blood vessel type (artery and vein) and group (OIR and control) on PO2. Avascular retinal areas were measured from FA images in OIR mice. RESULTS: There was a significant effect of vessel type on PO2 (P < 0.001). The effect of group on PO2 was not significant (P = 0.3), indicating similar PO2 between OIR and control mice. The interaction between group and vessel type was significant (P = 0.03), indicating a larger arteriovenous PO2 difference in OIR mice than control mice. In control mice, FA displayed normal vascularization, while FA of OIR mice showed abnormalities, including dilation and tortuosity of major retinal blood vessels, and avascular regions. In OIR mice, the mean percent avascular retinal area was 33% ± 18%. CONCLUSIONS: In vivo assessment of retinal vascular oxygen tension and vascularization patterns demonstrated abnormalities in the mouse model of OIR. This approach has the potential to improve understanding of retinal vascular development and oxygenation alterations due to ROP and other ischemic retinal diseases.

Automated prescription of an optimal imaging plane for measurement of cerebral blood flow by phase contrast magnetic resonance imaging.

This study describes and evaluates a semiautomated method for prescribing an optimal imaging plane that is located as close as possible to the skull base, and is simultaneously nearly perpendicular to the four arteries leading blood to the brain [internal carotid arteries (ICAs) and vertebral arteries (VAs)]. Such a method will streamline and improve reliability of the measurement of total cerebral blood flow and intracranial pressure by velocity encoding phase-contrast magnetic resonance imaging. The method first extracts the vessels' centerline from a 2-D time-of-flight magnetic resonance angiogram of the neck by performing distance transformations. An anatomical marker, the V2 segment of the VAs, is then identified to guide the imaging plane to be as close and below the skull base. An imaging plane that is nearly perpendicular to the ICAs and V2 segment of VAs is then identified by minimizing a misalignment value, estimated by a weighted mean of the angles between the plane's normal and the vessel axes at the vessel-plane intersections. The performance of the semiautomated method was evaluated by comparing manually selected planes to those found semiautomatically in nine magnetic resonance angiogram datasets. The semiautomated method consistently outperformed manual prescription with a significantly smaller misalignment value, 8.6° versus 20.7° (P < 0.001), respectively, and significantly improved reproducibility.

Inner retinal metabolic rate of oxygen by oxygen tension and blood flow imaging in rat.

The metabolic function of inner retinal cells relies on the availability of nutrients and oxygen that are supplied by the retinal circulation. Assessment of retinal tissue vitality and function requires knowledge of both the rate of oxygen delivery and consumption. The purpose of the current study is to report a novel technique for assessment of the inner retinal metabolic rate of oxygen (MO(2)) by combined measurements of retinal blood flow and vascular oxygen tension (PO(2)) in rat. The application of this technology has the potential to broaden knowledge of retinal oxygen dynamics and advance understanding of disease pathophysiology.

Analysis of the correlation between the facial temperature and the stenosis of carotid arteries.

Death caused by stroke above the age of 60 years placed second in the world, and is the fifth leading cause in people aged 15 to 59 years old. Several methods for early detection of stroke are magnetic resonance angiography, and carotid duplex, both diagnoses are cost and time consuming. This research is aimed to provide a noninvasive, cost effective, and rapid technique for diagnosing carotid artery stenosis by using thermography. In this study, 64 images from 32 people were used to analysis the correlation between the temperature of the face and the stenosis of carotid arteries by automatically selecting and calculating the mean and standard deviation of the facial temperature. We find that external carotid artery affects the facial temperature significantly.