Performance Of Cascade Research Articles

Effect of a Tip Groove on the Performance of Compressor Linear Cascade

Effect of a Tip Groove on the Performance of Compressor Linear Cascade

A New Segmented Suction Slot Design Based on the Performance of Compressor Cascade

To develop an effective suction slot arrangement, computational fluid dynamics simulation software and a high subsonic compressor cascade were used to simulate different suction slots. Based on the effects of various suction slots on the cascade performance under various operating conditions, a novel segmented suction slot structure Seg3 was proposed. The results of the study revealed that in the vicinity of the operating conditions ( incidence ≤ 4 ° ), a full-blade height suction slot should be installed at least 5% of the chord length downstream of the corner separation point to effectively remove the separation. For small and medium incidences, the best performance was observed for suction slot SS5, which was located at 60% of the chord length downstream of the leading edge. Analyzing the effects of SS5 and Seg3 on the cascade performance revealed that when the incidence was less than 4°, the reductions in the total pressure loss coefficients for ζsuc of SS5 and Seg3 both exceeded 8.2%. When the incidence was 4° or greater, the ζsuc increased slightly for SS5, whereas it decreased for Seg3, and the ζsuc was reduced. For a 3° incidence, for example, Seg3 reduced the ζsuc and passage blockage by 12.74% and 8.41%, respectively, and increased the static pressure rise coefficient by 18.55% from the baseline values. Thus, the segmented suction slot proposed in this paper outperformed conventional full-blade height suction slots.

Effect of inlet Mach number on performance and flow structure of an axial supersonic through-flow fan cascade

Ensuring the high efficiency and stable operation of a supersonic through-flow fan (STFF) in a wide range of inlet Mach numbers is of vital importance. The influence of inlet Mach number (M) ranging from 0.3 to 2.36 on the aerodynamic performance and flow structure in the STFF cascade is studied. The results indicate that at design incidence, the transonic inflow condition has a greater loss, a lower static pressure ratio, and a larger flow turning than the non-transonic inflow condition. The evolution of shock structure with the increasing inlet Mach number is as follows: shock-free → passage quasi-normal shock → leading edge (LE) local normal shock + λ shock → LE normal shock + trailing edge (TE) fishtail shock → LE bow shock + TE fishtail shock + multiple reflected oblique shocks → multiple oblique shocks. Under subsonic inflow conditions, the blade loss is dominated by boundary layer friction loss and wake loss. Under transonic inflow conditions, when the inlet Mach number rises from 0.66 to 0.86, the rear branch of passage λ shock impinges on the thickened separation region of the adjacent blade. This increases the loss. When the inlet Mach number rises to 1.06, the λ shock evolves into fishtail shocks, and the LE shock becomes stronger, reducing shock loss and separation loss significantly. As the inlet Mach number goes up to 1.26, the LE normal shock evolves into a bow shock, inducing a sharp reduction in LE shock loss. Under the supersonic inflow conditions, when the inlet Mach number rises from 1.96 to 2.36, the loss variation is mainly stemmed from LE bluntness loss, accounting for 66% of overall loss change.

Nitrogen Experiments on a Supersonic Linear Cascade For ORC Applications

A novel experiment has been conceived at Politecnico di Milano for the study of the flow within and downstream of supersonic cascades of Organic Rankine Cycle (ORC) turbines. This paper documents the first phase of the research, focused on the preliminary tests and studies performed by operating the facility with nitrogen as working fluid, to demonstrate the technical relevance of the experiment and the validity of the measurement system in a simplified thermodynamic condition. The set of measured data includes, beside the inlet total thermodynamic state, eight static pressure values obtained via taps manufactured on the test section rear end-wall, both within the bladed and semi-bladed region of the cascade, as well as a total pressure probe to retrieve the cascade performance. A double-passage Schlieren equipment was also employed to visualize the density gradients. Experiments show an outstanding repeatability, indicate a quasi-steady cascade operation during the blow-down process for all the pressure signal considered, and demonstrate a remarkable periodicity among two consecutive channels also in off-design conditions. Experimental data were also compared with CFD simulations, resulting in an excellent agreement for the pressure data acquired both within and downstream of the cascade.

Study on flow separation and suction control of a compressor tandem cascade

To control the boundary layer separation on the front blade suction surface in a compressor tandem cascade at a high incidence angle, three suction schemes at different axial positions were introduced and the cascade performance improvement and separation flow control effects were investigated. Based on the experimental and numerical simulations, the aerodynamic performance of the prototype tandem cascade reaches the peak at the incidence angle of 5°. Continue to increase the incidence angle, the loss increases sharply, and the static pressure rise coefficient and diffusion factor begin to decrease. With the boundary layer suction on the suction surface of the front blade, fluid energy in the upstream boundary layer of the suction slot is increased, and the adverse pressure gradient downstream is decreased. By suction, the development and separation of the boundary layer are inhibited, and the loss in the tandem cascade is significantly reduced. The flow control effect of the suction scheme at the onset location of boundary layer separation is the best, and the flow loss can be reduced by about 34.9% with a suction flow rate of 1% at the incidence angle of 5°. The suction schemes have good flow control effects in a wide working range and broaden the available range of incidence angles of the tandem cascade. This study provides a reference for the design of tandem blades considering flow control and has a positive significance for improving compressor efficiency and pressure ratio.

Operating performance assessment method for industrial process with slowness principle-based LSTM network

Timely and accurate performance assessment and non-optimal regulation of industrial processes can effectively guarantee product quality. Most industrial processes are highly nonlinear and dynamic, so long short-term memory (LSTM) network is suitable for industrial performance assessment. However, in the network learning, the typical LSTM network focuses on the representation learning of input variables, lacks the representation of comprehensive economic indexes (CEI), and cannot selectively learn essential features, which increases the computational burden and easily mixes redundant information. Thus, a supervised slow feature analysis (SFA)-based LSTM (SSFALSTM) network is proposed for industrial operating performance assessment. By utilizing CEI information and SFA constraints, the network is guided to simultaneously learn features related to CEI and slow-changing features that reflect the inherent dynamics of the process. Further, cascade performance recognition model to construct the complete performance assessment framework. For the non-optimal performance, a reconstruction-based contribution plot method is proposed to identify the main cause variables and guide adjustment operations. Finally, the effectiveness of the proposed method is validated on the dense medium coal preparation process.

Control mechanism of vortex-generator jet on turbulent separation in a highly loaded compressor cascade

The active flow control technique of vortex-generator jet (VGJ) was used to control the turbulent separation of a highly loaded compressor cascade at a high incidence. VGJ schemes on suction surface with different jet locations, skewed angles, pitch angles were numerically performed and the control mechanism of VGJ parameters on the flow field of the cascade was revealed. For avoiding the effect of endwall boundary layer, both endwall surfaces of the cascade were set to translational periodicity boundary (TPB). Results show that VGJ significantly eliminates the turbulent separation and improves the performance of cascade. The overall loss coefficient of cascade is reduced by 52.3% at most. The optimal VGJ location in this study is not separation location but 7% axial chord downstream of it, which is probably due to the ultra-high loading of the cascade. The spanwise location and intensity of jet vortices are mainly affected by skewed angle, and will increase with the skewed angle. The VGJ case with a higher pitch angle features better control effect on suction surface but increases the pitchwise scale of the low-velocity region. In addition, as pitch angle increases, the mixing loss between mainstream and jet flow increases, which brings a negative effect on cascade performance. In order to enhance the momentum transporting between mainstream flow and separation flow, counter vortex-generator jet (CVGJ) schemes were designed and studied. Though CVGJ strengthens the jet vortices near the trailing edge and achieves minimum loss near TPB sections at 150% axial chord section, the interaction of jet vortices with opposite rotating direction brings extra loss, which leads to the control effect of CVGJ is weaker than that of VGJ.

Effects of the airfoil probes on the aerodynamic performance of a transonic compressor cascade

The airfoil probe is typical instrumentation for measuring total pressure/temperature in the research fields of turbomachinery. Besides causing measurement error itself, it also generates extra flow loss inevitably and further affects the measuring accuracy. This paper utilizes the Reynolds-averaged Navier Stokes (RANS) and high-fidelity methods to understand the effects of airfoil probes on the aerodynamic performance of a compressor cascade. With the inclusion of airfoil probes, the load capacity and the shock waves decline, to some extent, mainly because of the streamwise vortex. The shedding vortex induced by the probes triggers an earlier and non-separated flow transition, which contributes to a suppression of the separation bubble on the suction surface and a reduction of the profile loss. However, through the quantitative traceability analysis of the flow loss, the extra loss is generated and redistributed to the overall flow field. The separated flow forms at the junction of the end wall, blade, and probe's pipes, due to the high adverse pressure gradient aggravated by shock waves. It is one of the reasons why the measuring accuracy of airfoil probes is quite sensitive to the Mach numbers. The Delayed-Detached Eddy Simulation (DDES) shows superiority in capturing small-scale coherent vortex structures. The shedding streamwise vortex generated by the probe's pipe is distinguished for the first time. Its transport progression to the wake region is revealed basically, which has been masked previously by the conventional RANS method. In addition, due to the supplement of the streamwise vortex generated by probes, the wake vortices maintain a higher shedding frequency. This work offers a glimpse of the complex fluid mechanisms in a transonic regime, which suggests carefully considering the size of instruments in the design and measurement process.

Hypertension Control Cascade and Regional Performance in India: A Repeated Cross-Sectional Analysis (2015-2021).

Background The weak control cascade of hypertension from the time of screening till the attainment of optimal blood pressure (BP) control is a public health challenge, particularly in resource-limited settings. The study objectives were to(1) estimate the change in the rate of prevalence of hypertension, the yield of newly diagnosed cases, initiation of treatment, and attainment of BP control in the age group15 to 49 years; (2)ascertain the magnitude and predictors of undiagnosed hypertension, lack of initiation of treatment, and poor control of those on antihypertensive therapy; and(3)estimate the regional variation and state-level performance of the hypertension control cascade in India. Methodology We analyzed demographic and health surveillance (DHS) data from India's National Family Health Survey Fifth Series (NFHS-5), 2019-2021, and NFHS-4 (2015-2016). The NFHS-5 sample comprised 695,707 women and 93,267 men in the age group of 15 to 49 years. Multiple logistic regressions were performed to find the associated predictors, and respective adjusted odds ratios (aORs) were reported. Results The prevalence of hypertension (cumulative previously diagnosed and new cases) amongindividuals aged 15 to 49 years was 22.8% (22.6%, 23.1%; n = 172,532), out of which 52.06% were newly diagnosed cases. In contrast, in NFHS-4, the prevalence of hypertension amongindividuals aged 15 to 49 years was 20.4% (20.2%, 20.6%; n = 153,384), of which 41.65% were newly diagnosed cases. In NFHS-5, 40.7% (39.8% and 41.6%) of the previously diagnosed cases were on BP-lowering medications compared to 32.6% (31.8%, 33.6%) in NFHS-4. Furthermore, in NFHS-5, controlled BP was observed in 73.7% (72.7% and 74.7%) of the patients on BP-lowering medication compared to 80.8% (80.0%, 81.6%) in NFHS-4.Females compared to males (aOR =0·72 and 0·007), residents of rural areas (aOR = 0·82 and 0·004), and those belonging to the socially disadvantaged groups were not initiated on treatment despite awareness of their hypertension status indicative of poor treatment-seeking behavior.Furthermore, increasing age (aOR = 0·49, P< 0·001), higher body mass index (aOR = 0·51, P< 0·001), and greater waist-to-hip ratio (aOR = 0·78, P= 0·047) were associated with uncontrolled hypertension in patients on antihypertensive drug therapy. Conclusions Hypertension control cascade in India is largely ineffectual although screening yieldand initiation of antihypertensive treatment have improved in NFHS-5 compared to NFHS-4. Identification of high-risk groups for opportunistic screening, implementing community-based screening, strengthening primary care, and sensitizing associated practitioners are urgently warranted.

Effect investigation of single-slotted and double-slotted configurations on the corner separation and aerodynamic performance in a high-load compressor cascade

Controlling the corner separation through blade-end slotting has been widely proven advantageous in compressors. This study proposed a new double-slotted configuration and deeply revealed its benefits and flow control mechanisms by compared with the single-slotted configuration. Based on a high-load compressor cascade, the Slotted_A scheme with a single blade-end slot1 and the Slotted_B scheme with a smaller slot2 behind the slot1 were established and detailedly evaluated. The results showed that, since the slot jet with high momentum has efficacious re-energizing effects for the local low-energy fluid, both the two slotted schemes can prevent the original cascade from entering the corner stall condition. But the slot2 jet can further control the end-wall secondary flow developing and separating behind the slot1 in the Slotted_A cascade. Therefore, the Slotted_B scheme behaves better in controlling the corner separation than the Slotted_A scheme, resulting in larger turning angles, smaller vortex structures, and weaker flow blockage. As a whole, the Slotted_A scheme and Slotted_B scheme can respectively increase the turning angle by an average of 0.93° and 1.53°, improve the static pressure coefficient by 18.2% and 21.7% on average, and achieve an average reduction of 38.3% and 43.7% in the total pressure loss. In contrast, the Slotted_B scheme is more advantageous in improving the pressure diffusion capacity and stability margin for the high-load compressor blade.

Experimental and Numerical Investigations for Dual−Cavity Tip Aerodynamic Performance in the Linear Turbine Cascade

Experimental and numerical studies of a linear high-loaded turbine cascade with a dual−cavity tip structure are presented in this paper. The experimental conditions contained an increase in the outlet Mach number from 0.42 to 0.92, a change in the incidence angle from −15° to 15° and an increase in the relative clearance size from 0.36% to 1.4%. The ability of the dual−cavity tip to control leakage losses and vortices is assessed using the total pressure coefficient and the Q-criterion. This research indicates that the leakage vortex interacts strongly with the passage vortex, and the change in working conditions affects the balance between the two vortices and thus the flow field structure. The experimental and numerical results prove that the dual−cavity tip can reduce losses in all operating conditions, with the best control effect reduced by 0.025 in a large clearance size condition. In addition, the leakage control effect of the blade tip structure is more influenced by the incoming flow angle and clearance size than the Mach number.

Aerodynamic Optimization and Mechanism Investigation on Performance Improvements in a Transonic Compressor Cascade

In this paper, a transonic compressor cascade was optimized to improve its aerodynamic performance. A new blade parameterization method with 16 control variables was first proposed to fit the shapes of the suction and the pressure side, as well as the leading edge. Then, the Kriging surrogate-model-based genetic algorithm (GA) was used to optimize the performance of the transonic cascade. The optimization algorithm is effective in reducing the total pressure loss while extending the working range of the cascade. The results show that the total pressure loss coefficient could be reduced by 11% at the best airflow angle and the working range could be extended by 6.9% for the optimized cascade in two-dimensional simulations. Similar improvement results could also be obtained in the simulations of their linear cascade cases. Detailed analyses show that the relative maximum thickness positions of the optimized blades move forward by about 10% to the leading edge, and the radii of curvature of the front half of the suction and pressure surfaces increase, compared with the initial blade. This makes the front half of the optimized blades look more closely like a wedge. Consequently, the passage shock strength is reduced and the shock changes from the passage normal shock to oblique shock. The weakened shock strength leads to the disappearance of the flow separation caused by the shock boundary layer interaction on the suction surfaces of the optimized blades, and results in a narrowed wake width at the outlet section.

Physics-informed deep learning cascade loss model

The design procedure of aero-engine compressor requires empirical models such as cascade loss model and deviation model. With accurate predictions of basic cascade performance by empirical model, better designs would be obtained and the cost would be reduced. A number of empirical models have been developed and most of them build mathematical equations based on experimental data. It is difficult to modify and broaden the scope of application for empirical models because of its inadequate ability to fit strong nonlinear function relationship and lack of data. Neural networks, which are widely used nowadays, are applied to solve this problem but a simple end-to-end neural network only slightly improves the model performance due to the neglect of flow field details and the similarity of different samples. A promising way to solve the problems is to embed the physics mechanism or flow field details into neural networks as intermediate variables to guide the learning of models. A physics-informed deep learning cascade loss model is proposed. By adjusting the architecture of neural network and the data structure, model can build the relationship of input and output in a more reasonable way while providing a prediction of flow fields. This paper verified the superiority of deep learning model, which has no physics informed, with a 22.3% error reduction when compared to empirical models. A physics-informed model is built and trained with the velocity distribution in boundary layer is chosen as intermediate variable and the influencing factors of physics-informed models are explored. The error of physics-informed model decreases 57.4% overall, and over 60% at high Mach numbers and high incidences compared to empirical model. The physics informed deep learning model is promising to replace the empirical models in compressor design since it shows better performance and gives a local description of the cascade flow field, which can be utilized in subsequent cascade design.

Impact of Slots on the Aerodynamic Performance of the Variable Inlet Guide Vane Cascade of a Centrifugal Compressor

This study reconstructed the flow field of a symmetrical variable inlet guide vane in a centrifugal compressor through the passive control method of vane slots. Based on the high-fidelity numerical simulation model verified by experiments, the influence of different slot forms on the flow field was investigated, and the passive control mechanism was revealed. The results demonstrated that the vane slot method can effectively suppress the suction surface separation and broaden the range of low-loss incidence angles. Overall, the 50_30 slotted vane achieves the best flow field control, with a 65.6% reduction in the total pressure loss coefficient and a 2.3° reduction in the deviation angle, respectively, at a 25° incidence angle. The linear characteristics of the pre-swirl grade variation curve with variable inlet guide vane incidence angles are also improved. Furthermore, changing the slot outlet angle has the most significant influence on the aerodynamic performance as it changes the throat width of the location, thereby affecting the flow rate and momentum of the jet. Finally, the impact of the velocity varies in the first self-similarity region on the slotted vane. The results indicate that, in contrast to the baseline vane, the suppression effect of the slot jet on the flow separation improves with the inlet velocity, whereas the deviation angle of the slotted vane declines with the inlet velocity. Meanwhile, the higher the incoming flow velocity, the better the slotted jet can inhibition of flow separation.

Mechanisms of oscillating suction in controlling the tip leakage flow in a high-load compressor cascade

To control tip leakage flow and reduce related loss in axial-flow compressors, the oscillating suction scheme was introduced in a high-load compressor cascade with a tip clearance, and its flow control mechanisms on the tip leakage flow were studied numerically. It is found that the flow control effect of oscillating suction is the best when the unsteady excitation frequency is equal or harmonic to the feature frequency of the flow field related to the flow separation. The loss of the cascade controlled by the oscillating suction with a reduced frequency of 0.39 and excitation amplitude of 10 kPa was reduced by about 3.4% compared with the steady suction scheme, and about 11.14% less than that of the non-suction scheme. Oscillating suction discretized the concentrated tip leakage vortex into a series of scale-reduced vortex structures by the periodic unsteady excitation, which can be identified by the analysis using proper orthogonal decomposition (POD). The feature frequencies became consistent with the oscillating excitation frequency or its multiple, and the fluctuation generated by the tip leakage flow was enhanced. The energy proportion of high-order POD modes was increased by oscillating excitation, which indicated the vortex structures with weak fluctuation intensity were more active. The mixing and energy exchange among the mainstream, separated flow, endwall secondary flow, and tip leakage flow were enhanced, so that a better flow control effect could be achieved. Furthermore, flow capacity of tip passage was modified, the overall aerodynamic performance of the compressor cascade was improved thereby.

Abstract P039: Framework for genomic cancer risk assessment dissemination and implementation in Latin America: The GRACIAS (genetic risk assessment for cancer implementation and sustainment) study

Abstract Introduction: Lack of public insurance coverage for genetic testing and a limited trained genomic cancer risk assessment (GCRA) workforce are barriers to cancer screening and prevention and to understanding the genetic epidemiology of cancer in low/middle income countries (LMIC) in Latin America. Methods: We implemented an intervention that included professional education for clinicians in Mexico, Peru and Colombia, support from a community of practice and a federated consortium of collaborators, a versatile research registry with a dual purpose (research and clinical care) centralized relational database and pedigree drawing program (web-accessible), and economical NGS multigene panel tests (MGPT). We report here the outcomes of GCRA for 4181 probands (3170 breast, 301 ovarian, 965 other cancer, and 262 unaffected). Results: With a mean age of 45.2 y.o. (39.9 y.o. at breast cancer dx for carriers), a pathogenic variant (PV) was detected in 605 (14.5%) of 2,710 probands (BRCA testing only for 1387, MGPT for 1323): BRCA1 (n=250); BRCA2 (129); CHEK2 (24); PALB2 (20); ATM (14); TP53 (14); PTEN (6); CDH1 (5); MMR genes (62); RAD51C (6); and BRIP1 (3). We have documented growing reach (proportion of those eligible that access GCRA), increasing cascade testing (family member(s) tested (n=541) for 192/470 [40%] probands with a PV), and with median 2.2 years follow up, 209 risk reduction surgeries were performed. Summary: Remarkably, despite systemic limitations, integration of a skilled multidisciplinary team with access to genetic assays and practice support tools, community, and hospital outreach increased GCRA reach, and the expansion of cascade testing and performance of risk-appropriate surgeries reflect maturing prevention programs. Sustainment of progress requires increased multilevel efforts and establishment of GCRA as a supported standard of care, so the opportunity for prevention is not missed for vulnerable and underserved high-risk patients and families. Citation Format: Jeffrey N. Weitzel, Julio Abugattas, Cynthia Villarreal-Garza, Adrian Daneri Navarro, Alicia Cock-Rada, Yenni Rodriguez, Pamela Mora-Alferez, Dione Aguilar, Azucena del Toro-Valero, Francisco Gutierrez-Delgado, Ana Ferrigno, Josef Herzog, Danielle Castillo, Gubidxa Gutierrez Seymour, Bita Nehoray, Kathleen R. Blazer, Maria E. Fernandez, Yanin Chavarri Guerra. Framework for genomic cancer risk assessment dissemination and implementation in Latin America: The GRACIAS (genetic risk assessment for cancer implementation and sustainment) study. [abstract]. In: Proceedings of the AACR Special Conference: Precision Prevention, Early Detection, and Interception of Cancer; 2022 Nov 17-19; Austin, TX. Philadelphia (PA): AACR; Can Prev Res 2023;16(1 Suppl): Abstract nr P039.

Diatomic catalysts for Fenton and Fenton-like reactions: a promising platform for designing/regulating reaction pathways.

The optimization of the single-atom catalyst (SAC) performance has been the hot spot for years. It is widely acknowledged that the incorporation of adjacent single-atom sites (diatomic catalysts (DACs)) can enable synergistic effects, which can be used in cascade catalysis, dual-function catalysis, and performance regulation of intrinsic active sites. DACs have been widely applied in the CO2 reduction reaction (CO2RR), oxygen reduction reaction (ORR), hydrogen evolution reaction (HER), etc.; however, their application is limited in Fenton or Fenton-like reactions. This perspective summarizes the most advanced achievements in this field, followed by the proposed opportunities in further research, including regulation of the magnetic moment, inter-atomic distance effect, strain engineering, atomic cluster (AC)/nanoparticle (NP) modification, etc. It is demonstrated that this perspective can contribute to the DAC application in Fenton or Fenton-like reactions with innovative design and mechanisms being put forward.

Liquid-liquid extraction of calcium in a scaled-out microfluidic device: Process intensification using a crown ether-ionic liquid system

Solvent extraction of calcium was investigated in four microchannels parallelized on a microfluidic chip. Uniform flow distribution of observed flow patterns in parallel microchannels shows successful numbering-up of the microfluidic device. First, flow regime maps were plotted for six liquid-liquid systems. A parallel flow regime with complete phase separation was chosen as the desired flow regime for extracting the calcium ion. Organic phase and crown ether types were tested to obtain the maximum extraction of calcium from the aqueous phase. In these conditions, the effects of crown ether concentration, the concentration of calcium ion and pH were studied in order to optimize the extraction efficiency using the Box-Behnken method. The mass transfer coefficient was also determined as a function of residence time at different concentrations of calcium metal ions. In the second step, ionic liquid-based solvent extraction was studied. In this regard, the performance of the diluent (conventional organic solvent) type on the calcium separation efficiency was investigated. An excellent improvement in separation efficiency (7.1–64.2%) was achieved for the EMIM NTf2/n-butyl acetate mixture compared to butyl acetate. In the final step, continuous multistage liquid-liquid extraction of the calcium ion was performed using three series microchips to evaluate the microfluidic cascade's performance.

Flexible cascade and parallel operations of discrete memristor

As a special nonlinear component, discrete memristor (DM) can lead to excellent chaos performance in its constructed discrete systems. To enhance the diversity and performance of DM-based systems, cascade, parallel and composite operations for DMs are proposed in this paper. Examples are given to demonstrate the operation process, through which new discrete memristive systems can be constructed with larger parameter space and chaotic region. Complex dynamical behaviors are investigated, and these designed maps are realized on a digital signal processor (DSP) development board. The generated chaotic sequences are further tested by NIST software to show random performance.

Segmentation and differentiation of periventricular and deep white matter hyperintensities in 2D T2-FLAIR MRI based on a cascade U-net.

White matter hyperintensities (WMHs) are a subtype of cerebral small vessel disease and can be divided into periventricular WMHs (pvWMHs) and deep WMHs (dWMHs). pvWMHs and dWMHs were proved to be determined by different etiologies. This study aimed to develop a 2D Cascade U-net (Cascade U) for the segmentation and differentiation of pvWMHs and dWMHs on 2D T2-FLAIR images. A total of 253 subjects were recruited in the present study. All subjects underwent 2D T2-FLAIR scan on a 3.0 Tesla MR scanner. Both contours of pvWMHs and dWMHs were manually delineated by the observers and considered as the gold standard. Fazekas scale was used to evaluate the burdens of pvWMHs and dWMHs, respectively. Cascade U consisted of a segmentation U-net and a differentiation U-net and was trained with a combined loss function. The performance of Cascade U was compared with two other U-net models (Pipeline U and Separate U). Dice similarity coefficient (DSC), Matthews correlation coefficient (MCC), precision, and recall were used to evaluate the performances of all models. The linear correlations between WMHs volume (WMHV) measured by all models and the gold standard were also conducted. Compared with other models, Cascade U exhibited a better performance on WMHs segmentation and pvWMHs identification. Cascade U achieved DSC values of 0.605 ± 0.135, 0.517 ± 0.263, and 0.510 ± 0.241 and MCC values of 0.617 ± 0.122, 0.526 ± 0.263, and 0.522 ± 0.243 on the segmentation of total WMHs, pvWMHs, and dWMHs, respectively. Cascade U exhibited strong correlations with the gold standard on measuring WMHV (R2 = 0.954, p < 0.001), pvWMHV (R2 = 0.933, p < 0.001), and dWMHV (R2 = 0.918, p < 0.001). A significant correlation was found on lesion volume between Cascade U and gold standard (r > 0.510, p < 0.001). Cascade U showed competitive results in segmentation and differentiation of pvWMHs and dWMHs on 2D T2-FLAIR images, indicating potential feasibility in precisely evaluating the burdens of WMHs.