About

David Lalush is an Associate Professor at the College of Engineering at NC State University, located in the Engineering Building III (EB3). His contact information includes the phone number 919-513-7671 and the email address dslalush@ncsu.edu. The college is situated at 915 Partners Way, Raleigh, NC 27695-7901. The webpage indicates his affiliation with the College of Engineering and his role within the faculty, but does not provide specific details about his research focus, background, or key contributions.

Research topics

• Anatomy
• Pathology
• Orthodontics
• Biomedical engineering
• Radiology
• Physical medicine and rehabilitation
• Psychology
• Neuroscience
• Developmental psychology
• Medicine

Selected publications

• Aberrant Gait Biomechanics Linked to Cartilage Changes After ACL Reconstruction in Those With High Body Mass Index

  Journal of Orthopaedic Research® · 2025-05-19 · 2 citations

  articleOpen access

  ABSTRACT A history of anterior cruciate ligament reconstruction (ACLR) and high body mass index (BMI) are strong risk factors for incident knee osteoarthritis. Limited research has evaluated the interaction between ACLR and high BMI on limb‐level loading and early deleterious changes in cartilage health. The purpose of this study was to separately investigate the association between vertical ground reaction force (vGRF) loading profiles during gait and tibiofemoral cartilage composition in ACLR patients with high and normal BMI. Forty‐three participants with primary unilateral ACLR (17 ± 14 months post‐ACLR) were categorized as high (≥ 25 kg/m 2 ; n = 18) or normal (< 25 kg/m 2 ; n = 25) BMI and performed an overground gait at self‐selected speed. For biomechanical outcomes, we calculated the differences between first peak and midstance minimum (∆vGRF1) and between the second peak and midstance minimum (∆vGRF2). T1ρ relaxation time interlimb ratios (ILR), calculated as the T1ρ relaxation time in the ACLR relative to the uninjured limb, were calculated for the medial and lateral tibia and femur. Stepwise linear regressions were used to determine associations between biomechanical outcomes and T1ρ relaxation time ILR for each region of interest. Lesser ∆vGRF1 and ∆vGRF2 in the high‐BMI group significantly associated with greater T1ρ relaxation time ILR for the medial femoral condyle (Δ R 2 = 0.28, p = 0.03; Δ R 2 = 0.25, p = 0.04, respectively) and tibial plateau (Δ R 2 = 0.55, p < 0.001; Δ R 2 = 0.25, p = 0.004, respectively). Aberrant limb‐level loading, characterized by less dynamic limb loading, is linked to deleterious changes in tibiofemoral cartilage in ACLR patients with high BMI, suggesting that gait retraining may be more critical for ACLR with a BMI ≥ 25 kg/m 2 .

  PublisherOA PDFDOI

• Feasibility of deep learning-based cancer detection in ultrasound microvascular images

  Ultrasonics · 2025-11-20

  article
  PublisherDOI

• A Practical Introduction to Wavelet Analysis in Electroretinography

  medRxiv · 2025-07-27 · 1 citations

  preprintOpen access

  Purpose: To provide a conceptual understanding of the continuous and discrete wavelet transforms (CWT, DWT) for clinical electroretinography (ERG) analysis, and how these methods uncover time-frequency features that complement traditional time-domain analysis. Methods: A technical overview without the use of mathematical formula describing the basics of CWT and DWT and implementation considerations. We also review an example of four standard ISCEV ERG recordings from a healthy male (between 30-34 years of age) and a male (between 15-19 years of age) with complete congenital stationary night blindness (CSNB). Results: Wavelet analysis uncovered time-frequency signatures absent in raw traces. In light-adapted flicker, the normal ERG showed a ~30 Hz response with harmonics up to 90 Hz, whereas CSNB was largely attenuated. For LA 3 and dark-adapted flashes, normal CWTs concentrated energy < 100 Hz between 0.04-0.08 s, while CSNB demonstrated lowered or almost absent energy profiles in comparison. DWT indices exhibited a similar pattern, with normal recordings demonstrating high energy responses early in the 7, 15, and 29 Hz frequency bands, while CSNB registered markedly lower values. Conclusions: CWT and DWT provide complementary and objective insight into ERG responses. Open-source MATLAB toolkit and step-by-step tutorial provided herein lower technical barriers and enable use by the broader community.

  PublisherOA PDFDOI

• Articular Cartilage Differences Between Those With High And Normal Body Mass Index Following Acl Injury

  Medicine & Science in Sports & Exercise · 2025-09-16

  article

  PURPOSE: Those with an anterior cruciate ligament (ACL) injury are at higher risk for developing knee osteoarthritis. However, little is known about how high body mass index (BMI), a strong risk factor for knee osteoarthritis development, affects cartilage prior to ACL reconstruction (ACLR) in those with acute ACL injuries. The purpose of this study was to determine differences in osteoarthritis-related biomarkers of T1ρ relaxation time and serum cartilage oligomeric matrix protein (sCOMP) concentration between individuals with high and normal BMI following an ACL injury. METHODS: Forty-two participants with primary ACL injuries who were seeking ACLR were dichotomized into either the high-BMI (>25 kg/m2; n = 20; 22.9 ± 5.1 years; days between injury and visit: 25.2 ± 15.4) or normal-BMI (≤25 kg/m2; n = 22; 21.1 ± 4.2 years; days between injury and visit: 23.1 ± 11.9) group based on their BMI at their preoperative visit. We used spin-locked T1ρ MR sequence to quantify T1ρ relaxation times to estimate proteoglycan density in the tibiofemoral articular cartilage composition. T1ρ relaxation times were calculated for the medial and lateral tibia (MTC and LTC) and femur (MFC and LFC) for each participant. Separate mixed-model ANOVAs were used to determine differences in T1ρ relaxation times across the regions of interests between the high- and normal-BMI groups in both the injured and uninjured limbs. An independent t-test was used to determine differences in sCOMP concentration between groups. RESULTS: The high-BMI group demonstrated greater T1ρ relaxation times in the LTC (50.1 ± 2.6 vs 48.0 ± 2.4 ms; p = 0.004; d = 0.77) and MTC (51.8 ± 5.2 vs 49.4 ± 3.4 ms; p = 0.04; d = 0.44) in both limbs compared to the normal-BMI group. The high-BMI group also showed greater concentrations of sCOMP compared to the normal-BMI group (148.8 ± 44.4 vs 122.4 ± 28.6 ng/mL; p = 0.02; d = 0.72). CONCLUSIONS: These findings suggest that BMI may impact bilateral articular cartilage composition and systemic biochemical changes linked to knee osteoarthritis development in individuals with acute ACL injuries. Individuals with BMI >25 kg/m2 may exhibit a predisposition for greater articular cartilage breakdown even prior to ACLR, which is important for guiding therapeutic interventions to prevent knee osteoarthritis in this patient population. Supported by: Arthritis Foundation and the National Institute of Arthritis and Musculoskeletal and Skin Diseases (T32AR082310 and P30AR072580).

  PublisherDOI

• Distribution-Guided Multi-tracer Brain PET Synthesis from Structural MRI with Class-Conditioned Weighted Diffusion

  Lecture notes in computer science · 2025-09-18

  book-chapterOpen access
  PublisherOA PDFDOI

• 272. Associations Between Perimenopausal Anhedonia and Striatal Dopamine Binding

  Biological Psychiatry · 2025-04-09

  article
  PublisherDOI

• A practical introduction to wavelet analysis in electroretinography

  Documenta Ophthalmologica · 2025-12-15 · 1 citations

  articleOpen access

  Abstract Purpose To provide a conceptual understanding of the continuous and discrete wavelet transforms (CWT, DWT) for clinical electroretinography (ERG) analysis, and how these methods uncover time–frequency features that complement traditional time-domain analysis. Methods A technical overview without the use of mathematical formula describing the basics of CWT and DWT and implementation considerations. We also review an example of four standard ISCEV full-field ERG (ffERG) recordings from a healthy 32-year-old male. Results Wavelet analysis uncovered time–frequency signatures absent in raw traces. In DA 0.01 cd s/m 2 DWT scalogram, energy localized in the 2–5 Hz range, with CWT scalograms corroborating these findings. In DA 3.0 cd s/m 2 , a broader frequency response is seen across 10, 20 and 40 Hz center frequencies. A similar progression was found in the LA 3.0 cd s/m 2 , with additional low energy indices at 80 and 160 Hz. For the LA 30 Hz flicker, all frequency and time–frequency profiles effectively replicated the 30 Hz response of the cone system. Conclusions CWT and DWT provide complementary and objective insight into ERG responses. Open-source MATLAB toolkit and step-by-step tutorial provided herein lower technical barriers and enable use by the broader community.

  PublisherOA PDFDOI

• Striatal dopamine in anhedonia: A simultaneous [11C]raclopride positron emission tomography and functional magnetic resonance imaging investigation

  UNC Libraries · 2024-08-02

  articleOpen access
  PublisherDOI

• BIOFEEDBACK-DRIVEN GAIT DYNAMICS REDUCES KNEE CARTILAGE EXPOSURE TO HIGH STRESSES: A CASE STUDY AND IN SILICO PROOF-OF-CONCEPT IN AN ACL RECONSTRUCTED PATIENT

  Osteoarthritis and Cartilage · 2024-04-01

  articleOpen access
  PublisherOA PDFDOI

• Abstract 4143465: Aortic Root Pressure for Detecting Aortic Stenosis using Machine Learning

  Circulation · 2024-11-12

  article

  Background: Aortic stenosis (AS) is a progressive, deteriorative valvular condition that is associated with significant morbidity and mortality. The best method to diagnose the severity of AS is controversial as all of the current modalities have multiple potential sources of error. As severity of AS increases, the time from aortic valve opening to peak systolic pressure increases. We hypothesized that machine learning applied to pressure measured in the aortic root via a fluid filled catheter at the time of cardiac catheterization would accurately diagnose aortic stenosis. Aims: Use a long-short term memory (LSTM) neural network to identify AS using aortic root pressure. Methods: We assessed aortic root pressure recordings in 102 consecutive patients undergoing transcatheter aortic valve replacement (TAVR) at our institution between 2014 and 2017. A LSTM was trained, validated, and tested using pre- and post- TAVR aortic root pressure 1 Hz digital recordings of 3.5 seconds duration. Recordings were selected to be artifact-free and in sinus rhythm. A 10-fold cross-validation structure was used (170 training, 17 validation, 17 testing tracings). The model was assessed using area under the receiver operating curve (AUROC) and F1 score. Results: The LSTM was able to distinguish between pre-TAVR (severe AS) and post-TAVR (no AS) tracings with sensitivity of 0.764, specificity of 0.703, AUROC of 0.792, and F1 score of 0.741. In comparison, when pressure tracings of the left ventricle were added to the model, the LSTM performance improved to sensitivity of 0.956, specificity of 0.950, AUROC of 0.977, and F1 score of 0.954. Conclusion: This proof-of-concept study demonstrated that aortic pressure alone can be used to detect severe AS. A larger analysis is needed to validate these findings across gradients of severity. This study has implications for using automatic AS severity assessments during cardiac catheterization to accurately diagnose AS.

  PublisherDOI

Frequent coauthors

• Brian Pietrosimone

  Duke University

  37 shared

• B.M.W. Tsui

  Johns Hopkins University

  36 shared

• Cam Patterson

  University of Arkansas for Medical Sciences

  33 shared

• Weili Lin

  Imaging Center

  32 shared

• Dinggang Shen

  28 shared

• Yueh Z. Lee

  27 shared

• C. Brandon Frederick

  University of North Carolina at Chapel Hill

  27 shared

• Hong Yuan

  Lung Institute

  24 shared