Jichu Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Jichu Zhang has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Andrew J. Shoffstall
Department of Biomedical Engineering, Case Western Reserve University, Cleveland, USA
3 shared publications - Andrew R. Crofton
Department of Anatomy, Case Western Reserve University, Cleveland, OH
2 shared publications Havisha Kalpatthi
Department of Biomedical Engineering, Case Western Reserve University, Cleveland, OH
2 shared publications- Nicole A. Pelot
Duke University
2 shared publications - Aniruddha R. Upadhye
Department of Biomedical Engineering, Case Western Reserve University 2 , Cleveland, Ohio 44106
1 shared publication - Ashlesha Deshmukh
Wisconsin Institute for Translational Neuroengineering, Department of Neurological Surgery and Department of Biomedical Engineering, University of Wisconsin-Madison, 500 Lincoln Dr, Madison, 53706-1314, United States
1 shared publication
Latest Publications
- Open Access
Mapping the Fascicular Morphology and Organization of the Human Sciatic Nerve via High-Resolution MicroCT Imaging
Objective: Implanted neuroprostheses can restore standing and walking after spinal cord injury and somatosensation after limb loss. Yet current approaches often fail to reliably activate hamstring muscles crucial for upright stability and mobility or to target afferent fibers for sensory restoration. We developed a novel methodology using high-resolution micro-computed tomography (microCT) to visualize and track fascicle groups innervating distinct hamstring muscles along the human sciatic nerve. This approach provides a framework for mapping fascicular topography in complex neural pathways to optimize standing neuroprostheses. Methods: Bilateral sciatic nerves were dissected and excised from an embalmed human cadaver, annotated with branch names, and stained with phosphotungstic acid before undergoing microCT scanning at 11.4 μm isotropic resolution. Images were segmented with a 3D U-Net convolutional neural network. Segmentation results were used to quantify morphological metrics and track fascicular organization along ∼25 cm of the nerve. MicroCT reconstructions were validated against histological cross sections. Results: Gross dissection revealed matched proximal-to-distal branching between left and right sciatic nerves: branch to long head of the biceps femoris (lhBF), branch to hamstring part of the adductor magnus and semimembranosus (HAM/SM), and branch to semitendinosus (ST). All branches originated medially and followed an inferomedial trajectory. Branch-free lengths of the sciatic exhibited asymmetry, especially between the lumbosacral roots to the first branch (5.5 cm left vs. 1.5 cm right) and lhBF to the HAM/SM branch (9.0 cm left vs. 16.5 cm right). MicroCT analysis revealed bilateral symmetry in fascicle diameters (∼0.4 mm) and total fascicle counts (∼84) but asymmetry in hamstring-innervating fascicle counts (left ∼7, right: ∼9). The 3D fascicular maps revealed that hamstring fascicles were located in the anteromedial portion of the sciatic nerve cross section and remained separate for distances up to 15.9 cm proximal to their branching points. Significance: Our microCT-based approach enables efficient, high-resolution 3D mapping of fascicular organization within large, complex peripheral nerves like the sciatic nerve, overcoming previous technical limitations. This methodology informs development of neuroprostheses with improved hamstring muscle activation for enhanced standing function.
Open AccessMapping the Fascicular Morphology and Organization of the Human Sciatic Nerve via High-Resolution MicroCT Imaging
Objective: Implanted neuroprostheses can restore standing and walking after spinal cord injury and somatosensation after limb loss. Yet current approaches often fail to reliably activate hamstring muscles crucial for upright stability and mobility or to target afferent fibers for sensory restoration. We developed a novel methodology using high-resolution micro-computed tomography (microCT) to visualize and track fascicle groups innervating distinct hamstring muscles along the human sciatic nerve. This approach provides a framework for mapping fascicular topography in complex neural pathways to optimize standing neuroprostheses. Methods: Bilateral sciatic nerves were dissected and excised from an embalmed human cadaver, annotated with branch names, and stained with phosphotungstic acid before undergoing microCT scanning at 11.4 μm isotropic resolution. Images were segmented with a 3D U-Net convolutional neural network. Segmentation results were used to quantify morphological metrics and track fascicular organization along ∼25 cm of the nerve. MicroCT reconstructions were validated against histological cross sections. Results: Gross dissection revealed matched proximal-to-distal branching between left and right sciatic nerves: branch to long head of the biceps femoris (lhBF), branch to hamstring part of the adductor magnus and semimembranosus (HAM/SM), and branch to semitendinosus (ST). All branches originated medially and followed an inferomedial trajectory. Branch-free lengths of the sciatic exhibited asymmetry, especially between the lumbosacral roots to the first branch (5.5 cm left vs. 1.5 cm right) and lhBF to the HAM/SM branch (9.0 cm left vs. 16.5 cm right). MicroCT analysis revealed bilateral symmetry in fascicle diameters (∼0.4 mm) and total fascicle counts (∼84) but asymmetry in hamstring-innervating fascicle counts (left ∼7, right: ∼9). The 3D fascicular maps revealed that hamstring fascicles were located in the anteromedial portion of the sciatic nerve cross section and remained separate for distances up to 15.9 cm proximal to their branching points. Significance: Our microCT-based approach enables efficient, high-resolution 3D mapping of fascicular organization within large, complex peripheral nerves like the sciatic nerve, overcoming previous technical limitations. This methodology informs development of neuroprostheses with improved hamstring muscle activation for enhanced standing function.
- Open Access
Automated 3D Segmentation of Human Vagus Nerve Fascicles and Epineurium from Micro-Computed Tomography Images Using Anatomy-Aware Neural Networks
Objective. Precise segmentation and quantification of nerve morphology from imaging data are critical for designing effective and selective peripheral nerve stimulation (PNS) therapies. However, prior studies on nerve morphology segmentation suffer from important limitations in both accuracy and efficiency. This study introduces a deep learning approach for robust and automated 3D segmentation of human vagus nerve fascicles and epineurium from high-resolution micro-computed tomography (microCT) images. Methods. We developed a multi-class 3D U-Net to segment fascicles and epineurium that incorporates a novel anatomy-aware loss function to ensure that predictions respect nerve topology. We trained and tested the network using subject-level five-fold cross-validation with 100 microCT sub-volumes (11.4 [&mu]m isotropic resolution) from cervical and thoracic vagus nerves stained with phosphotungstic acid from five subjects. We benchmarked the 3D U-Net's performance against a 2D U-Net using both standard and anatomy-specific segmentation metrics. Results. Our 3D U-Net generated high-quality segmentations (average Dice similarity coefficient: 0.93). Compared to the 2D U-Net, our 3D U-Net yielded significantly better volumetric overlap, boundary delineation, and fascicle-level accuracy. The 3D approach reduced anatomical errors by 2.5-fold, provided more consistent inter-slice boundaries, and improved detection of fascicle splits/merges by nearly 6-fold. Significance. Our automated 3D segmentation pipeline provides anatomically accurate 3D maps of peripheral neural morphology from microCT data. The automation allows for high throughput, and the substantial improvement in segmentation quality and anatomical fidelity enhances the reliability of morphological analysis, vagal pathway mapping, and the implementation of realistic computational models. These advancements provide a foundation for understanding the functional organization of the vagus and other peripheral nerves and optimizing PNS therapies.
Open AccessAutomated 3D Segmentation of Human Vagus Nerve Fascicles and Epineurium from Micro-Computed Tomography Images Using Anatomy-Aware Neural Networks
Objective. Precise segmentation and quantification of nerve morphology from imaging data are critical for designing effective and selective peripheral nerve stimulation (PNS) therapies. However, prior studies on nerve morphology segmentation suffer from important limitations in both accuracy and efficiency. This study introduces a deep learning approach for robust and automated 3D segmentation of human vagus nerve fascicles and epineurium from high-resolution micro-computed tomography (microCT) images. Methods. We developed a multi-class 3D U-Net to segment fascicles and epineurium that incorporates a novel anatomy-aware loss function to ensure that predictions respect nerve topology. We trained and tested the network using subject-level five-fold cross-validation with 100 microCT sub-volumes (11.4 [&mu]m isotropic resolution) from cervical and thoracic vagus nerves stained with phosphotungstic acid from five subjects. We benchmarked the 3D U-Net's performance against a 2D U-Net using both standard and anatomy-specific segmentation metrics. Results. Our 3D U-Net generated high-quality segmentations (average Dice similarity coefficient: 0.93). Compared to the 2D U-Net, our 3D U-Net yielded significantly better volumetric overlap, boundary delineation, and fascicle-level accuracy. The 3D approach reduced anatomical errors by 2.5-fold, provided more consistent inter-slice boundaries, and improved detection of fascicle splits/merges by nearly 6-fold. Significance. Our automated 3D segmentation pipeline provides anatomically accurate 3D maps of peripheral neural morphology from microCT data. The automation allows for high throughput, and the substantial improvement in segmentation quality and anatomical fidelity enhances the reliability of morphological analysis, vagal pathway mapping, and the implementation of realistic computational models. These advancements provide a foundation for understanding the functional organization of the vagus and other peripheral nerves and optimizing PNS therapies.
- Open Access
- 10 October 2024
- Published by Wiley in Journal of Neuroimaging
- Vol. 34 (6) , 646-663
Quantification of porcine lower thoracic spinal cord morphology with intact dura mater using high‐resolution μCT
Background and Purpose: Spinal cord stimulation (SCS) is approved by the Food and Drug Administration for treating chronic intractable pain in the back, trunk, or limbs through stimulation of the dorsal column. Numerous studies have used swine as an analog of the human spinal cord to better understand SCS and further improve its efficacy. We performed high‐resolution imaging of the porcine spinal cord with intact dura mater using micro‐computed tomography (μCT) to construct detailed 3‐dimensional (3D) visualizations of the spinal cord and characterize the morphology of the dorsal and ventral rootlets.Methods: We obtained spinal cords from Yorkshire/Landrace crossbred swine (N = 7), stained samples with osmium tetroxide, and performed μCT imaging of the T12‐T15 levels at isotropic voxel resolutions ranging from 3.3 to 50 μm. We measured the anatomical morphology using the 3D volumes and compared our results to measurements previously collected from swine and human spinal cords via microdissection techniques in prior literature.Results: While the porcine thoracic‐lumbar spinal cord is a popular model for SCS, we highlight multiple notable differences compared to previously published T8‐T12 human measurements including rootlet counts (porcine dorsal/ventral: 12.2 ± 2.6, 26.6 ± 3.4; human dorsal/ventral: 5.3 ± 1.3, 4.4 ± 2.4), rootlet angles (porcine ventral‐rostral: 161 ± 1°, ventral‐caudal: 155 ± 6°, dorsal‐rostral: 148 ± 9°, dorsal‐caudal: 142 ± 6°; human ventral‐rostral: 170 ± 3°, ventral‐caudal: 22 ± 10°, dorsal‐rostral: 171 ± 3°, dorsal‐caudal: 15 ± 7°), and the presence and count of dorsal rootlet bundles.Conclusions: Detailed measurements and highlighted differences between human and porcine spinal cords can inform variations in modeling and electrophysiological experiments between the two species. In contrast to other approaches for measuring the spinal cord and rootlet morphology, our method keeps the dura intact, reducing potential artifacts from dissection.
- 10 October 2024
- Published by Wiley in Journal of Neuroimaging
- Vol. 34 (6) , 646-663
Open AccessQuantification of porcine lower thoracic spinal cord morphology with intact dura mater using high‐resolution μCT
Background and Purpose: Spinal cord stimulation (SCS) is approved by the Food and Drug Administration for treating chronic intractable pain in the back, trunk, or limbs through stimulation of the dorsal column. Numerous studies have used swine as an analog of the human spinal cord to better understand SCS and further improve its efficacy. We performed high‐resolution imaging of the porcine spinal cord with intact dura mater using micro‐computed tomography (μCT) to construct detailed 3‐dimensional (3D) visualizations of the spinal cord and characterize the morphology of the dorsal and ventral rootlets.Methods: We obtained spinal cords from Yorkshire/Landrace crossbred swine (N = 7), stained samples with osmium tetroxide, and performed μCT imaging of the T12‐T15 levels at isotropic voxel resolutions ranging from 3.3 to 50 μm. We measured the anatomical morphology using the 3D volumes and compared our results to measurements previously collected from swine and human spinal cords via microdissection techniques in prior literature.Results: While the porcine thoracic‐lumbar spinal cord is a popular model for SCS, we highlight multiple notable differences compared to previously published T8‐T12 human measurements including rootlet counts (porcine dorsal/ventral: 12.2 ± 2.6, 26.6 ± 3.4; human dorsal/ventral: 5.3 ± 1.3, 4.4 ± 2.4), rootlet angles (porcine ventral‐rostral: 161 ± 1°, ventral‐caudal: 155 ± 6°, dorsal‐rostral: 148 ± 9°, dorsal‐caudal: 142 ± 6°; human ventral‐rostral: 170 ± 3°, ventral‐caudal: 22 ± 10°, dorsal‐rostral: 171 ± 3°, dorsal‐caudal: 15 ± 7°), and the presence and count of dorsal rootlet bundles.Conclusions: Detailed measurements and highlighted differences between human and porcine spinal cords can inform variations in modeling and electrophysiological experiments between the two species. In contrast to other approaches for measuring the spinal cord and rootlet morphology, our method keeps the dura intact, reducing potential artifacts from dissection.