American Institute of Ultrasound in Medicine <strong>Proceedings</strong> J Ultrasound Med 32(suppl):S1–S134, 2013Results—Spearman’s coefficient between measurements usingdifferent machines was 0.686 for automatic and 0.721 for manual measurement(P < .0001 for both measurements). A hyperechoic SN was detectedin the same 26 (86.7%) PD patients using both automatic andmanual measurements. Cohen’s κ coefficients for automatic and manualmeasurements were 0.787 and 0.762, respectively (P < .0001 for bothmeasurements).Conclusions—The results of the presented study showedcomparable findings for SN feature measurement using the designed applicationwith manual measurement. (Supported by a grant from theMoravian-Silesian Region.)1540954 Investigation of Asymmetries in Cerebral Collateral Flowfor Patients With Carotid StenosisKhalid Al Muhanna, 1 * Limin Zhao, 2 Kirk Beach, 3 BrajeshLal, 4,5 Gregory Kowalewski, 5 Siddhartha Sikdar 1 1 Electricaland Computer Engineering and Bioengineering, GeorgeMason University, Fairfax, Virginia USA; 2 University ofMaryland Medical Center, Baltimore, Maryland USA; 3 Surgeryand Bioengineering, University of Washington, Seattle, WashingtonUSA; 4 Vascular Surgery, University of MarylandMedical Center, Baltimore, Maryland USA; 5 Baltimore VeteransAffairs Medical Center, Baltimore, Maryland USAObjectives— Stroke affects millions of people in the worldeach year. About 25% of ischemic strokes are caused by rupture ofcarotid artery plaque. Currently, stenosis severity is used as a surrogatefor the risk of plaque rupture; however, other factors may play a largerrole, such as the hemodynamics around the plaque, which may be affectedboth by extracranial hemodynamics and intracranial collateralization.In the present study, we investigated how an incomplete(noncollateralized) intracranial circle of Willis (COW) might affect intraluminalvelocity around the carotid plaque and whether there areasymmetries of flow in the middle cerebral artery (MCA) in patientswith extracranial carotid stenosis.Methods—We created a simple linear simulation model of theintracranial and extracranial circulation to investigate the relationship betweenMCA flow waveforms on the contralateral (normal) and ipsilateral(diseased) sides and carotid stenosis for a complete and incomplete COW.Then we compared the predictions of this model with bilateral MCAvelocity measurements performed in patients with asymptomatic carotidstenosis using transcranial Doppler.Results—Simulation results showed no asymmetries in MCAflow waveforms for a complete (collateralized) COW, but for an incompleteCOW, the systolic peak had a lower magnitude and was delayed byabout 100 milliseconds on the ipsilateral side. In our clinical measurementson 32 patients, we found that 25 had waveforms consistent withthose predicted for a collateralized COW, with minimal differences indelay, velocity magnitude, and resistivity index between the ipsilateral andcontralateral sides. In 6 cases, some unexpected findings were noted, suchas large delays for 2 patients who had ≤50% stenosis and a larger velocitydifference with low delays for 5 patients with >50% stenosis.Conclusions—Our results indicate that intracranial flow is animportant variable when interpreting intrastenotic velocities. The presentstudy does not allow us to definitely interpret the reason for MCA flowasymmetries, since the COW was not directly imaged. We intend to enrolladditional patients in our cohort with concomitant imaging of the COW tofurther strengthen our results.1512012 Left-to-Right Image Registration of Longitudinal CarotidImages Improves Intima-Media Thickness and AtheroscerlsosisDisease MonitoringFilippo Molinari, 1 Nobutaka Ikeda, 2 U Rajendra Acharya, 3Luca Saba, 4 Andrew Nicolaides, 5 Jasjit Suri 6,7 * 1Electronicsand Telecommunications, Politecnico Torino, Torino, Italy;2Division of Cardiovascular Medicine, Toho University MedicalCenter, Tokyo, Japan; 3 Electronics and Computer Engineering,Ngee Ann Polytechnic, Singapore; 4 Radiology, AziendaOspedaliero, Universitaria di Cagliari, Cagliari, Italy;5Imperial College, London, England; 6 Global Biomedical Technologies,Roseville, California USA; 7 Biomedical Engineering,Idaho State University, Pocatello, Idaho USAObjectives—Automated systems for the measurement of thecarotid intima-media thickness (CIMT) are useful in clinical practice ifthey ensure high measurement accuracy and high reproducibility. We developeda registration-based method to improve the carotid distal wall segmentationand CIMT measurement in noisy images.Methods—We tested 50 patients and acquired left and rightcommon carotid arteries in 3 projections: anteroposterior, anterolateral,and lateroposterior. The total number of images was 300 (50 subjects, 2 arteries,3 insonation angles), and we had all images manually segmented by3 independent expert readers. We processed each image by a 3-stage system.Stage 1 is relative to automated carotid localization and far adventitiatracing. Stage 2 is relative to the definition of a guidance zone andregistration of the left to the right distal wall. Registration was performedby relying on the profile of the far adventitia. The segmentation is carriedout in stage 3 by using a edge snapper. We compared the CIMT measurementaccuracy of the registered and unregistered image sets.Results—Stage 1 was successful in all 300 images. Left-torightregistration was successful in 140 of 150 cases (93.3% success),whereas right-to-left registration was successful in 138 cases (92.0% success).The average CIMT measurement bias in the unregistered case was0.012 ± 0.079 mm, which decreased to 0.006 ± 0.081 mm for the registeredimages. The figure of merit (FoM) increased from 98.19% for the unregisteredto 99.09% for the registered image set.Conclusions—Registering the left to the right carotid arteryimages can increase CIMT measurement accuracy. We plan to extend thiswork by also including arteries with plaques in the registration framework.Table 1. Auto Edge Performance for the 3 OperatorsGT1 GT2 GT3 Average GTOriginal 0.020 ± 0.079 0.051 ± 0.083 –0.034 ± 0.090 0.012 ± 0.079CIMT bias,mmFoM, % 96.99 92.00 95.24 98.19Registered 0.014 ± 0.085 0.045 ± 0.086 –0.040 ± 0.090 0.006 ± 0.081CIMT bias,mmFoM, % 97.90 92.96 94.39 99.091540884 More Easily Deployable Long-term Transcranial DopplerMonitoring of the Middle Cerebral ArteryBill Beck PhysioSonics, Inc, Bellevue, Washington USAS35Objectives—Develop a system to facilitate long-term transcranialDoppler monitoring of the M1 segment of the middle cerebral artery(M1 MCA), allowing continuous data collection over a period of dayswithout operator intervention after initial setup.Methods—Deployment of M1 MCA monitoring is facilitatedby a structured procedure to guide the operator in: (1) positioning a 2Dphased array transducer, mounted in a headset, over the temporal window,
American Institute of Ultrasound in Medicine <strong>Proceedings</strong> J Ultrasound Med 32(suppl):S1–S134, 2013by measuring echo strength from the far side of the skull, then (2) articulatingthe transducer to point at the M1 MCA, through the use of flashcolor Doppler insonation of the 3D conical region of interest (ROI). Thephased array is then electronically steered to gather data from the peakDoppler signal in the ROI; the Doppler spectrum is analyzed, and standardflow metrics are logged for future review. The peak Doppler signal is automaticallyrelocated as required to maintain continuous monitoring withoutoperator intervention. An alert is generated if the signal is lost or ifflow metrics exceed user-specified limits.Results—The objective was achieved through development ofappropriate acoustics, supported by refinements in signal processing andan enhanced user interface.Conclusions—A system has been developed to make transcranialDoppler monitoring of the M1 MCA more easily deployable and tosupport long-term monitoring over a period of days, without operator intervention.S36