Comparing algorithms for automated vessel segmentation in computed tomography scans of the lung: the VESSEL12 study.

The VESSEL12 (VESsel SEgmentation in the Lung) challenge objectively compares the performance of different algorithms to identify vessels in thoracic computed tomography (CT) scans. Vessel segmentation is fundamental in computer aided processing of data generated by 3D imaging modalities. As manual vessel segmentation is prohibitively time consuming, any real world application requires some form of automation. Several approaches exist for automated vessel segmentation, but judging their relative merits is difficult due to a lack of standardized evaluation. We present an annotated reference dataset containing 20 CT scans and propose nine categories to perform a comprehensive evaluation of vessel segmentation algorithms from both academia and industry. Twenty algorithms participated in the VESSEL12 challenge, held at International Symposium on Biomedical Imaging (ISBI) 2012. All results have been published at the VESSEL12 website http://vessel12.grand-challenge.org. The challenge remains ongoing and open to new participants. Our three contributions are: (1) an annotated reference dataset available online for evaluation of new algorithms; (2) a quantitative scoring system for objective comparison of algorithms; and (3) performance analysis of the strengths and weaknesses of the various vessel segmentation methods in the presence of various lung diseases.

Design and numeric evaluation of a novel axial-flow left ventricular assist device.

Virtual design characteristics and performance of the first Turkish axial-flow left ventricular assist device (LVAD) are presented, with emphasis on rotor geometry. The patented rotor design includes a central flow channel carved inside the main block, which carries permanent magnets. A concentric rotor-stator gap minimizes the distance between respective magnets, improving electromagnetic efficiency and creating a second blood pathway. Dual sets of three helical blades, placed on the shaft and external surface of the rotor block, ensure unidirectionality. Hemodynamic performance was tested with computational fluid dynamics (CFD); and rotor-blade geometry was optimized, to maximize overall efficiency d and minimize backflow and wall shear stresses. For a shaft radius of 4.5 mm, rotor blade height of 2.5 mm, and blade inlet and exit metal angles of 67° and 32°, pump operation at the nominal head-flow combination (5 L/min and 100.4 mm Hg) was achieved at a rotor speed of 10,313 rpm. At the nominal point, backflow as percent of total flow was 7.29 and 29.87% at rotor inlet and exit, respectively; overall hydraulic efficiency reached 21.59%; and maximum area-averaged shroud shear was 520 Pa. Overall efficiency peaked at 24.07% for a pump flow of 6.90 L/min, and averaged at 22.57% within the flow range of 4-8 L/min. We concluded that the design satisfies initial rotor design criteria, and that continued studies with diffuser optimization and transient flow analysis are warranted.

Continuous flow total artificial heart: modeling and feedback control in a mock circulatory system.

We developed a mock circulatory loop and used mathematical modeling to test the in vitro performance of a physiologic flow control system for a total artificial heart (TAH). The TAH was constructed from two continuous flow pumps. The objective of the control system was to maintain loop flow constant in response to changes in outflow resistance of either pump. Baseline outflow resistances of the right (pulmonary vascular resistance) and the left (systemic vascular resistance) pumps were set at 2 and 18 Wood units, respectively. The corresponding circuit flow was 4 L/min. The control system consisted of two digital integral controllers, each regulating the voltage, hence, the rotational speed of one of the pumps. The in vitro performance of the flow control system was validated by increasing systemic and pulmonary vascular resistances in the mock loop by 4 and 8 Wood units (simulating systemic and pulmonary hypertension conditions), respectively. For these simulated hypertensive states, the flow controllers regulated circuit flow back to 4 L/min within seconds by automatically adjusting the rotational speed of either or both pumps. We conclude that this multivariable feedback mechanism may constitute an adequate supplement to the inherent pressure sensitivity of rotary blood pumps for the automatic flow control and left-right flow balance of a dual continuous flow pump TAH system.

Total heart replacement using dual intracorporeal continuous-flow pumps in a chronic bovine model: a feasibility study.

Continuous-flow pumps are small, simple, and respond physiologically to input variations, making them potentially ideal for total heart replacement. However, the physiological effects of complete pulseless flow during long-term circulatory support without a cardiac interface or with complete cardiac exclusion have not been well studied. We evaluated the feasibility of dual continuous-flow pumps as a total artificial heart (TAH) in a chronic bovine model. Both ventricles of a 6-month-old Corriente crossbred calf were excised and sewing rings attached to the reinforced atrioventricular junctions. The inlet portions of 2 Jarvik 2000 pumps were positioned through their respective sewing rings at the mid-atrial level and the pulseless atrial reservoir connected end-to-end to the pulmonary artery and aorta. Pulseless systemic and pulmonary circulations were thereby achieved. Volume status was controlled, and systemic and pulmonary resistance were managed pharmacologically to keep mean arterial pressures at 100+/-10 mmHg (systemic) and 20+/-5 mmHg (pulmonary) and both left and right atrial pressures at 15+/-5 mmHg. The left pump speed was maintained at 14,000 rpm and its output autoregulated in response to variations in right pump flow, systemic and pulmonary pressures, fluid status, and activity level. Hemodynamics, end-organ function, and neurohormonal status remained normal. These results suggest the feasibility of using dual continuous-flow pumps as a TAH.

Total heart replacement with dual centrifugal ventricular assist devices.

In an ovine feasibility study, we implanted two HeartMate-III centrifugal ventricular assist devices (VADs) for total heart replacement. With cardiopulmonary bypass support, both ventricles were transected at the atrioventricular groove, preserving a rim of ventricular tissue. The atrioventricular valves were excised, and the aorta and pulmonary artery were transected above the ventriculoarterial valves. An interatrial septal window was created by excising the foramen ovale. The VADs' sewing rings were attached to the left and right ventricular remnants, respectively. Outflow grafts were anastomosed to the aorta and pulmonary artery. The left VAD operated continuously at 4,500 rpm. Right VAD speed increased from 2,000 to 4,500 rpm in 500 rpm increments. Outflow graft flow, pressure, oxygen saturation, and shunt direction were recorded. The pulmonary artery to aortic ratio of flow and pressure increased from 0.26 and 0.15 (at 2,000 rpm) to 1.21 and 0.53, respectively (at 4,500 rpm). The interatrial shunt, which was right to left at lower right VAD speeds, progressed to bidirectional, then to left dominant as right VAD speed increased. Outflow-graft oxygen saturation was reflective of the shunt direction. In this acute experiment, total heart replacement with continuous flow VADs satisfactorily balanced left and right ventricular flows and preserved the physiologic circulatory response.

The effect of intermittent low speed mode upon aortic valve opening in calves supported with a Jarvik 2000 axial flow device.

We assessed the effects of an axial flow left ventricular assist device (LVAD) upon aortic valve opening, pump outflow, and biologic and hematologic parameters when operated in intermittent low speed (ILS) mode. An ILS controller equipped Jarvik 2000 LVAD was implanted in six calves. Pump speed was maintained at 10,000 rpm, and pump outflow was measured throughout the study period (71 +/- 6 days [mean +/- SD]). Hematologic and biochemical parameters were analyzed daily for the first 10 days, weekly for the first month, and biweekly thereafter to monitor for kidney or liver dysfunction, hemolysis, bleeding, or infection. Before study termination, esmolol hydrochloride was infused to induce low cardiac output and totally impair aortic valve opening. Radiopaque cineaortography was performed over 30 second intervals (10 seconds before, 10 seconds during, and 10 seconds immediately after ILS controller activation) to assess the effect of ILS mode upon aortic valve opening. After study termination, major end organs and the major vascular tree were removed and examined macroscopically and histologically for thrombus formation and infarction; the aortic valve was examined for thickening and fusion. All pumps were explanted and examined for thrombus formation. All six calves recovered without surgical or mechanical complications. Hematologic and biochemical parameters did not change significantly between baseline and study termination. The aortic valve successfully opened when ILS mode was activated, even under low cardiac output conditions. No thrombus was detected in the major end organs and vascular tree, except for some small renal infarcts in three calves that did not affect renal function. These results indicate that operating an axial flow LVAD in ILS mode allows aortic valve opening and aortic root washout.

Effect of pump flow mode of novel left ventricular assist device upon end organ perfusion in dogs with doxorubicin induced heart failure.

End organ effects of nonpulsatile (NP) and pulsatile (P) left ventricular assist device (LVAD) flow were compared in a canine model of doxorubicin-induced heart failure. After heart failure induction, a prototype bimodal LVAD was implanted. Hemodynamics, cardiac dimensions, and myocardial metabolism were monitored with the LVAD off (baseline) and on (in NP and P modes at 70% or 100% power). End organ perfusion was assessed by colored microsphere analysis. Seven dogs were used: two died before pump implantation and were excluded from analysis, and the remaining five survived to study termination. At 70% NP, ascending aortic flow and myocardial oxygen consumption (MVO2) decreased significantly. At 100% NP, LV dimensions decreased, aortic systolic, pulse, and LV pressures decreased but not significantly, and ascending aorta flow reversed. At 100% NP, coronary blood flow, MVO2, and LV free wall subepicardial and subendocardial blood flows decreased significantly. However, as NP support increased, the subepicardial/subendocardial blood flow ratio remained near baseline. At 100% NP, right ventricular perfusion decreased but not significantly, cerebral perfusion decreased significantly, and renal perfusion stayed constant. P mode results were similar, except that ascending aorta flow decreased significantly at 100% P instead of reversing as at 100% NP. These results suggest that end organ perfusion is not differentially affected by LVAD flow mode during chronic heart failure.

Evaluation of myocardial function in patients with end-stage heart failure during support with the Jarvik 2000 left ventricular assist device.

In 2 patients with the Jarvik 2000 left ventricular assist device (LVAD), we assessed left ventricular systolic function through pressure-volume loops and E(max) at the beginning and end of the support period to potentially predict the possibility of pump removal without transplantation. Immediately before LVAD implantation and explantation, pressure and volume measurements were made with catheters and echocardiography, respectively, the E(max) being calculated from the slope of the pressure-volume loops, and the left ventricular ejection fraction (LVEF) being estimated by echocardiography. Transplantation was performed after 14 and 62 days, respectively, during which the LVEF increased by 75% (from 12% to 21%) in Patient 1 and remained unchanged (from 16% to 18%) in Patient 2, whereas the E(max) increased from 0.63 and 0.42 mm Hg/ml, respectively, to 1.31 and 1.07 mm Hg/ml, reflecting a 107% and 155% improvement. In these 2 cases, the E(max) was a more reliable indicator of intrinsic myocardial contractility than was the LVEF.

Transmyocardial laser revascularization as an adjunct to coronary artery bypass grafting: a randomized, multicenter study with 4-year follow-up.

We evaluated transmyocardial laser revascularization (TMLR) with coronary artery bypass grafting (CABG) versus CABG alone for severe coronary artery disease involving 21 myocardial region unsuited for CABG. At 4 centers, 44 consecutive patients were randomized for CABG+TMLR (n = 23) or CABG alone (n = 21). Operative and in-hospital mortality and morbidity rates were monitored. Clinical status was evaluated at hospital discharge, 1 year, and 4 years. Success was characterized by relief of angina and freedom from repeat revascularization and death. Preoperatively, 20 patients (47%) were at high risk. The CABG technique, number of grafts, and target vessels were similar in both groups. Patients undergoing CABG+TMLR received 25 +/- 11 laser channels. Their < or = 30-day mortality was 13% (3/23) compared with 28% (6/21) after CABG alone (P = 0.21). There were no significant intergroup differences in the number of intraoperative or in-hospital adverse events. The follow-up period was 50.3 +/- 17.8 months for CABG alone and 48.1 +/- 16.8 months for CABG+TMLR. Both groups had substantially improved angina and functional status at 1 and 4 years, with no significant differences in cumulative 4-year mortality. The incidence of repeat revascularization was 24% after CABG alone versus none after CABG+TMLR (P < 0.05). The 4-year event-free survival rate was 14% versus 39%, respectively (P < 0.064). In conclusion, CABG+TMLR appears safe and poses no additional threat for high-risk patients. Improved overall success and repeat revascularization rates may be due to better perfusion of ischemic areas not amenable to bypass. Further studies are warranted to determine whether these trends are indeed significant.

Biventricular support with the Jarvik 2000 ventricular assist device in a calf model of pulmonary hypertension.

The Jarvik 2000 ventricular assist device (VAD) is clinically efficacious for treating end-stage left ventricular failure. Because simultaneous right ventricular support is also occasionally necessary, we developed a biventricular Jarvik 2000 technique and tested it in a calf model. One VAD was implanted in the left ventricle with outflow-graft anastomosis to the descending aorta. The other VAD was implanted in the right ventricle with outflow-graft anastomosis to the pulmonary artery. Throughout the 30 day study, hemodynamic values were continuously monitored. On day 30, both pumps were evaluated at different speeds, under various hemodynamic conditions. By gradually occluding the pulmonary artery proximally or distally, we simulated varying degrees of high pulmonary vascular resistance, right ventricular hypertension, global heart failure, or ventricular fibrillation. The two VADs maintained biventricular support even during pulmonary artery occlusion and ventricular fibrillation, yielding a cardiac output of 3-11 L/min, left ventricular end-diastolic pressure of 11-24 mm Hg, and central venous pressure of 9-25 mm Hg. End-organ function was unimpaired, and no major adverse events occurred. The dual VADs offered safe, effective biventricular assistance in the calf. Additional studies are needed to assess the effects of lowered pulse pressure upon the pulmonary circulation and to develop a single pump speed controller.

Thrombogenicity of mechanical aortic valves in an animal model: site specific testing is crucial.

We evaluated a new trileaflet prosthesis and a control bileaflet prosthesis in the mitral and aortic positions in 27 calves. The prototype trileaflet valve (TV1) functioned satisfactorily in the mitral position (TV1m, n = 7) but later yielded thrombogenic complications in the aortic position (TV1a, n = 4). The valve was redesigned (TV2) and retested in the mitral (TV2m n = 4) and aortic (TV2a, n = 5) positions, along with control valves (Cm, n = 4; Ca, n = 3). At necropsy, the valves were graded on a scale of 0 (no visible thrombi) to 4 (thrombi greater than 5 mm and/or obstructed leaflets). The TV1m, TV2m, and Cm animals, respectively, had implant durations of 215+/-112, 140+/-63, and 159+/-89 days and thrombus grades of 0.71+/-0.76, 0.33+/-0.58, and 1.50+/-0.58. The TV1a, TV2a, and Ca animals had implant durations of 18+/-12, 159+/-61, and 108+/-62 days and thrombus grades of 2.75+/-1.00, 0.50+/-0.58, and 0.67+/-0.58 (p < .005; TV2a vs. TV1a). Three TV1a calves died early of valve related complications. A design irregularity, undetected in the mitral position but revealed in the aortic position, caused a high early mortality in the TV1a animals. Redesigning the prosthesis eliminated valve related mortality and significantly reduced the thrombus grade. Because satisfactory performance in the mitral position does not guarantee safety or efficacy in the aortic position, site specific preclinical testing is crucial for mechanical heart valves.

Baseline hemodynamic and echocardiographic indices in anesthetized calves.

The experimental calf model is used to assess mechanical circulatory support devices and prosthetic heart valves. Baseline indices of cardiac function have been established for the normal awake calf but not for the anesthetized calf. Therefore, we gathered hemodynamic and echocardiographic data from 16 healthy anesthetized calves (mean age, 189.0 +/- 87.0 days; mean body weight, 106.9 +/- 32.3 kg) by cardiac catheterization and noninvasive echocardiography, respectively. Baseline hemodynamic data included heart rate (65 +/- 12 beats per minute), mean aortic pressure (113.5 +/- 17.4 mm Hg), left ventricular end-diastolic pressure (16.3 +/- 38.9 mm Hg), and mean pulmonary artery pressure (21.7 +/- 8.3 mm Hg). Baseline two-dimensional echocardiographic data included left ventricular systolic dimension (3.5 +/- 0.7 cm), left ventricular diastolic dimension (5.6 +/- 0.8 cm), end-systolic intraventricular septal thickness (1.7 +/- 0.2 cm), end-diastolic intraventricular septal thickness (1.2 +/- 0.2 cm), ejection fraction (63 +/- 10%), and fractional shortening (37 +/- 10%). Doppler echocardiography revealed a maximum aortic valve velocity of 0.9 +/- 0.5 m/s and a cardiac index of 3.7 +/- 1.1 L/minute/m2. The collected baseline data will be useful in assessing prosthetic heart valves, cardiac assist pumps, new cannulation techniques, and robotics applications in the anesthetized calf model and in developing calf models of various cardiovascular diseases.

Preclinical hemodynamic assessment of a new trileaflet mechanical valve in the aortic position in a bovine model.

BACKGROUND AND AIM OF THE STUDY: The hemodynamic characteristics of a new trileaflet mechanical heart valve (TTV) (TriFlo Inc., Costa Mesa, CA, USA) in the aortic position were evaluated in a bovine model. The TTV was designed to combine the durability of mechanical heart valves with the central flow characteristics of a bioprosthesis. METHODS: Using nine calves, the native aortic valve was replaced with a 21-mm TTV (n = 6) or a St. Jude Medical (SJM) valve (n = 3). Hemodynamic values were assessed with echocardiography at implantation and with catheterization at explantation. All calves underwent a necropsy, followed by gross pathology and light microscopy studies. RESULTS: The mean implant duration was 159 +/- 55 days for the TTV and 102 +/- 67 days for the SJM valve. Immediately before euthanasia, the peak and mean pressure gradients were respectively 35 +/- 14 and 24 +/- 9 mmHg for the TTV, and 100 +/- 72 and 59 +/- 38 mmHg for the SJM valve (p = 0.03). Two of the SJM-valve animals were electively sacrificed after showing symptoms of aortic stenosis. At necropsy, all nine valves were free from thrombi, pannus, occlusive tissue, or mechanical impairment of leaflet motion. The prematurely sacrificed SJM-valve animals had concentric myocardial hypertrophy consistent with severe functional aortic stenosis. CONCLUSION: The significantly higher gradients and left ventricular hypertrophy in the SJM-valve animals were related to a smaller effective orifice area, which precluded adaptation to increasing transvalvular volumes in the growing animal. This problem was not seen with the TTV, which performed hemodynamically as well as the SJM valve. The high transvalvular gradients seen with the SJM valve at study end may suggest that the hemodynamic characteristics of the TTV may be superior, though additional studies are needed to confirm this.

Myocardial hemodynamics, physiology, and perfusion with an axial flow left ventricular assist device in the calf.

The Jarvik 2000 axial flow left ventricular assist device (LVAD) is used clinically as a bridge to transplantation or as destination therapy in end-stage heart disease. The effect of the pump's continuous flow output on myocardial and end-organ blood flow has not been studied experimentally. To address this, the Jarvik 2000 pump was implanted in eight calves and then operated at speeds ranging from 8,000 to 12,000 rpm. Micromanometry, echocardiography, and blood oxygenation measurements were used to assess changes in hemodynamics, cardiac dimensions, and myocardial metabolism, respectively, at different speeds as compared with baseline (pump off, 0 rpm) in this experimental model. Microsphere studies were performed to assess the effects on heart, kidney, and brain perfusion at different speeds. The Jarvik 2000 pump unloaded the left ventricle and reduced end-diastolic pressures and left ventricular dimensions, particularly at higher pump speeds. The ratio of myocardial oxygen consumption to coronary blood flow and the ratio of subendocardial to subepicardial blood flow remained constant. Optimal adjustment of pump speed and volume status allowed opening of the aortic valve and contribution of the native left ventricle to cardiac output, even at the maximum pump speed. Neither brain nor kidney microcirculation was adversely affected at any pump speed. We conclude that the Jarvik 2000 pump adequately unloads the left ventricle without compromising myocardial metabolism or end-organ perfusion.

Preclinical assessment of a trileaflet mechanical valve in the mitral position in a calf model.

BACKGROUND: The bileaflet valve is currently the mechanical replacement valve of choice. Though durable, it does not closely mimic native valve hemodynamics and remains potentially thrombogenic. METHODS: Prototype trileaflet valves (T1 and T2) were implanted in the mitral position in calves. Group I calves received either a T1 valve (n = 12) or a control bileaflet valve (n = 5); Group II, either a T2 valve (n = 7) or a control bileaflet valve (n = 5). Valve function, perivalvular leakage, and transvalvular pressure gradients were evaluated. Also, long-term prototype leaflet wear was evaluated in vivo in one Group I calf (502 days) and two Group II calves (385 and 366 days). Calves were euthanized and necropsied at study termination, and major organs weighed and examined. RESULTS: Valve function was excellent and hematologic parameters remained normal in all calves that survived to study termination. Mean peak transvalvular pressure gradients were 10 +/- 7 mm Hg for T1 valves, 6 +/- 3 mm Hg for T2 valves, and 12 +/- 4 mm Hg for bileaflet control valves. Clinically insignificant valvular regurgitation was observed in both prototypes. Explanted valves showed no thrombus-impaired leaflet motion, except in two T1-fitted calves and one T2-fitted calf. Major organs showed no evidence of clinically significant thromboembolic events. There were no other significant differences between the results of experimental and control groups. CONCLUSIONS: Prototype trileaflet valves performed safely and effectively in the mitral position in calves, even without long-term anticoagulation. This warrants their evaluation as an equivalent alternative to bileaflet valves.

Ex vivo resuscitation of adult pig hearts.

One possible way to expand the human heart donor pool is to include non-heart-beating human donors. To begin validating this approach, we developed an ex vivo cardiac perfusion circuit to support large mammalian hearts in Langendorff mode and beating-ejecting mode and to assess and improve their ischemic tolerance. In vivo hemodynamic data and heparinized blood (4.0 +/- 0.5 L) were collected from 6 anesthetized pigs. Hearts were isolated and connected to a recirculating perfusion circuit primed with autologous buffered blood (pH, 7.40). After retrograde aortic perfusion in Langendorff mode, the left atrium was gravity-filled at 10-20 mmHg, and the left ventricle began to eject against a compliance chamber in series with a systemic reservoir set to a hydraulic afterload of 100-120 mmHg. Left ventricular function was restored and maintained in all 6 hearts for 30 min. Cardiac output, myocardial oxygen consumption, stroke work, aortic pressure, left atrial pressure, and heart rate were measured. The mean myocardial oxygen consumption was 4.8 +/- 2.7 mL/min/100 g (95.8% of in vivo value); and mean stroke work, 5.3 +/- 1.1 g x m/100 g (58.95% of in vivo value). One resuscitated heart was exposed to 30 min of normothermic ischemic arrest, then flushed with Celsior and re-resuscitated. The ex vivo perfusion method described herein restored left ventricular ejection function and allowed assessment of ischemic tolerance in large mammalian hearts, potentially a 1st step toward including non-heart-beating human donors in the human donor pool.

Off-pump coronary artery bypass grafting and transmyocardial laser revascularization via a left thoracotomy.

Off-pump coronary artery bypass grafting may be combined with adjunctive transmyocardial laser revascularization to optimize revascularization. This approach may be advantageous for high-risk patients, particularly those having undergone previous sternotomies. From October 2000 through May 2001, 17 patients (9 women and 8 men) underwent off-pump coronary artery bypass grafting and transmyocardial laser revascularization via a left thoracotomy. The patients had a mean age of 63 years and a mean ejection fraction of 0.33. All but 1 patient had undergone previous coronary surgery. In each patient, the heart was approached via a left thoracotomy through the 5th intercostal space, and 37 transmural channels, 1 mm in diameter, were each created with a single pulse of the carbon dioxide laser. Coronary artery bypass grafting was then performed with left internal thoracic artery or saphenous vein grafts. The follow-up period ranged from 2.1 to 9.3 months (mean, 6.2 months). The patients received 28 bypass grafts (mean, 1.6 grafts). Postoperatively, 2 patients required inotropic support. On day 8, 1 patient died of ventricular fibrillation. After a mean hospitalization of 7.7 days, the remaining patients were discharged, free of angina. At follow-up examination after a mean of 6 months (range, 2-9 months), 15 patients remained free of angina and one had mild angina. None had required further hospitalization. Performed via a left thoracotomy, off-pump coronary artery bypass grafting plus transmyocardial laser revascularization yielded an acceptable mortality rate, no major morbidity, and substantial angina relief in this carefully selected group of challenging, high-risk patients.