Pei Xu

All Publications

• Explainable deep learning and biomechanical modeling for TMJ disorder morphological risk factors JCI INSIGHT Sun, S., Xu, P., Buchweitz, N., Hill, C. N., Ahmadi, F., Wilson, M. B., Mei, A., She, X., Sagl, B., Slate, E. H., Lee, J. S., Wu, Y., Yao, H. 2024; 9 (16)

  Abstract

  Clarifying multifactorial musculoskeletal disorder etiologies supports risk analysis, development of targeted prevention, and treatment modalities. Deep learning enables comprehensive risk factor identification through systematic analyses of disease data sets but does not provide sufficient context for mechanistic understanding, limiting clinical applicability for etiological investigations. Conversely, multiscale biomechanical modeling can evaluate mechanistic etiology within the relevant biomechanical and physiological context. We propose a hybrid approach combining 3D explainable deep learning and multiscale biomechanical modeling; we applied this approach to investigate temporomandibular joint (TMJ) disorder etiology by systematically identifying risk factors and elucidating mechanistic relationships between risk factors and TMJ biomechanics and mechanobiology. Our 3D convolutional neural network recognized TMJ disorder patients through participant-specific morphological features in condylar, ramus, and chin. Driven by deep learning model outputs, biomechanical modeling revealed that small mandibular size and flat condylar shape were associated with increased TMJ disorder risk through increased joint force, decreased tissue nutrient availability and cell ATP production, and increased TMJ disc strain energy density. Combining explainable deep learning and multiscale biomechanical modeling addresses the "mechanism unknown" limitation undermining translational confidence in clinical applications of deep learning and increases methodological accessibility for smaller clinical data sets by providing the crucial biomechanical context.

  View details for DOI 10.1172/jci.insight.178578

  View details for Web of Science ID 001296350300001

  View details for PubMedID 38990647

  View details for PubMedCentralID PMC11343598

• AdaptNet: Policy Adaptation for Physics-Based Character Control ACM TRANSACTIONS ON GRAPHICS Xu, P., Xie, K., Andrews, S., Kry, P. G., Neff, M., McGuire, M., Karamouzas, I., Zordan, V. 2023; 42 (6)

  View details for DOI 10.1145/3618375

  View details for Web of Science ID 001139790400005

• Mask R-CNN provides efficient and accurate measurement of chondrocyte viability in the label-free assessment of articular cartilage. Osteoarthritis and cartilage open Fan, H., Xu, P., Chen, X., Li, Y., Zhang, Z., Hsu, J., Le, M., Ye, E., Gao, B., Demos, H., Yao, H., Ye, T. 2023; 5 (4): 100415

  Abstract

  Chondrocyte viability (CV) can be measured with the label-free method using second harmonic generation (SHG) and two-photon excitation autofluorescence (TPAF) imaging. To automate the image processing for the label-free CV measurement, we previously demonstrated a two-step deep-learning method: Step 1 used a U-Net to segment the lacuna area on SHG images; Step 2 used dual CNN networks to count live cells and the total number of cells in extracted cell clusters from TPAF images. This study aims to develop one-step deep learning methods to improve the efficiency of CV measurement.TPAF/SHG images were acquired simultaneously on cartilage samples from rats and pigs using two-photon microscopes and were merged to form RGB color images with red, green, and blue channels assigned to emission bands of oxidized flavoproteins, reduced forms of nicotinamide adenine dinucleotide, and SHG signals, respectively. Based on the Mask R-CNN, we designed a deep learning network and its denoising version using Wiener deconvolution for CV measurement.Using training and test datasets from rat and porcine cartilage, we have demonstrated that Mask R-CNN-based networks can segment and classify individual cells with a single-step processing flow. The absolute error (difference between the measured and the ground-truth CV) of the CV measurement using the Mask R-CNN with or without Wiener deconvolution denoising reaches 0.01 or 0.08, respectively; the error of the previous CV networks is 0.18, significantly larger than that of the Mask R-CNN methods.Mask R-CNN-based deep-learning networks improve efficiency and accuracy of the label-free CV measurement.

  View details for DOI 10.1016/j.ocarto.2023.100415

  View details for PubMedID 38025155

  View details for PubMedCentralID PMC10679817

• Context-Aware Timewise VAEs for Real-Time Vehicle Trajectory Prediction IEEE ROBOTICS AND AUTOMATION LETTERS Xu, P., Hayet, J., Karamouzas, I. 2023; 8 (9): 5440-5447

  View details for DOI 10.1109/LRA.2023.3295990

  View details for Web of Science ID 001036073300011

• Composite Motion Learning with Task Control Xu, P., Shang, X., Zordan, V., Karamouzas, I. ASSOC COMPUTING MACHINERY. 2023

  View details for DOI 10.1145/3592447

  View details for Web of Science ID 001044671300059

• Too Stiff, Too Strong, Too Smart: Evaluating Fundamental Problems with Motion Control Policies PROCEEDINGS OF THE ACM ON COMPUTER GRAPHICS AND INTERACTIVE TECHNIQUES Xie, K., Xu, P., Andrews, S., Zordan, V. B., Kry, P. G. 2023; 6 (3)

  View details for DOI 10.1145/3606935

  View details for Web of Science ID 001059100600017

• SocialVAE: Human Trajectory Prediction Using Timewise Latents Xu, P., Hayet, J., Karamouzas, I., Avidan, S., Brostow, G., Cisse, M., Farinella, G. M., Hassner, T. SPRINGER INTERNATIONAL PUBLISHING AG. 2022: 511-528

  View details for DOI 10.1007/978-3-031-19772-7_30

  View details for Web of Science ID 000898297000030

• Automated chondrocyte viability analysis of articular cartilage based on deep learning segmentation and classification of two-photon microscopic images Fan, H., Xu, P., Le, M., Hsu, J., Chen, X., Li, Y., Zhang, Z., Gao, B., Woolf, S., Ye, T., Brown, T. G., Wilson, T., Waller, L. SPIE-INT SOC OPTICAL ENGINEERING. 2022

  View details for DOI 10.1117/12.2609880

  View details for Web of Science ID 000831998600011

• A GAN-Like Approach for Physics-Based Imitation Learning and Interactive Control PROCEEDINGS OF THE ACM ON COMPUTER GRAPHICS AND INTERACTIVE TECHNIQUES Xu, P., Karamouzas, I. 2021; 4 (3)

  View details for DOI 10.1145/3480148

  View details for Web of Science ID 000703447500014

• Human-Inspired Multi-Agent Navigation using Knowledge Distillation Xu, P., Karamouzas, I., IEEE IEEE. 2021: 8105-8112

  View details for DOI 10.1109/IROS51168.2021.9636463

  View details for Web of Science ID 000755125506058