Towards clinical applicability and computational efficiency in automatic cranial implant design: An overview of the AutoImplant 2021 cranial implant design challenge.

Cranial implants are commonly used for surgical repair of craniectomy-induced skull defects. These implants are usually generated offline and may require days to weeks to be available. An automated implant design process combined with onsite manufacturing facilities can guarantee immediate implant availability and avoid secondary intervention. To address this need, the AutoImplant II challenge was organized in conjunction with MICCAI 2021, catering for the unmet clinical and computational requirements of automatic cranial implant design. The first edition of AutoImplant (AutoImplant I, 2020) demonstrated the general capabilities and effectiveness of data-driven approaches, including deep learning, for a skull shape completion task on synthetic defects. The second AutoImplant challenge (i.e., AutoImplant II, 2021) built upon the first by adding real clinical craniectomy cases as well as additional synthetic imaging data. The AutoImplant II challenge consisted of three tracks. Tracks 1 and 3 used skull images with synthetic defects to evaluate the ability of submitted approaches to generate implants that recreate the original skull shape. Track 3 consisted of the data from the first challenge (i.e., 100 cases for training, and 110 for evaluation), and Track 1 provided 570 training and 100 validation cases aimed at evaluating skull shape completion algorithms at diverse defect patterns. Track 2 also made progress over the first challenge by providing 11 clinically defective skulls and evaluating the submitted implant designs on these clinical cases. The submitted designs were evaluated quantitatively against imaging data from post-craniectomy as well as by an experienced neurosurgeon. Submissions to these challenge tasks made substantial progress in addressing issues such as generalizability, computational efficiency, data augmentation, and implant refinement. This paper serves as a comprehensive summary and comparison of the submissions to the AutoImplant II challenge. Codes and models are available at https://github.com/Jianningli/Autoimplant_II.

Sarcopenia in Men With Bone-Predominant Metastatic Castration-Resistant Prostate Cancer Undergoing Ra-223 Therapy.

INTRODUCTION: Osteosarcopenia is the progressive loss of musculoskeletal structure and functionality, contributing to disability and mortality. Despite complex interactions between bone and muscle, osteosarcopenia prevention and treatment in men with metastatic castration-resistant prostate cancer (mCRPC) focuses predominantly on bone health. It is unknown whether Radium-223 (Ra-223) therapy affects sarcopenia. METHODS: We identified 52 patients with mCRPC who had received Ra-223 and had a baseline plus ≥1 follow-up abdominopelvic CT scan. The total contour area (TCA) and averaged Hounsfield units (HU) of the left and right psoas muscles were obtained at the inferior L3 endplate, and the psoas muscle index (PMI) was calculated therefrom. Intrapatient musculoskeletal changes were analyzed across various time points. RESULTS: TCA and PMI gradually declined over the study period (P = .002, P = .003, respectively), but Ra-223 therapy did not accelerate sarcopenia, nor the decline of HU compared to the pre-Ra-223 period. The median overall survival of patients with baseline sarcopenia was numerically worse (14.93 vs. 23.23 months, HR 0.612, P = .198). CONCLUSIONS: Ra-223 does not accelerate sarcopenia. Thus, worsening muscle parameters in men with mCRPC undergoing Ra-223 therapy are attributable to other factors. Further research is needed to determine whether baseline sarcopenia predicts poor overall survival in such patients.

AutoImplant 2020-First MICCAI Challenge on Automatic Cranial Implant Design.

The aim of this paper is to provide a comprehensive overview of the MICCAI 2020 AutoImplant Challenge. The approaches and publications submitted and accepted within the challenge will be summarized and reported, highlighting common algorithmic trends and algorithmic diversity. Furthermore, the evaluation results will be presented, compared and discussed in regard to the challenge aim: seeking for low cost, fast and fully automated solutions for cranial implant design. Based on feedback from collaborating neurosurgeons, this paper concludes by stating open issues and post-challenge requirements for intra-operative use. The codes can be found at https://github.com/Jianningli/tmi.

Trabecular bone loss at a distant skeletal site following noninvasive knee injury in mice.

Traumatic injuries can have systemic consequences, as the early inflammatory response after trauma can lead to tissue destruction at sites not affected by the initial injury. This systemic catabolism may occur in the skeleton following traumatic injuries such as anterior cruciate ligament (ACL) rupture. However, bone loss following injury at distant,unrelated skeletal sites has not yet been established. In the current study, we utilized a mouse knee injury model to determine whether acute knee injury causes a mechanically significant trabecular bone loss at a distant, unrelated skeletal site (L5 vertebral body).Knee injury was noninvasively induced using either high-speed (HS; 500 mm/s) or lowspeed(LS; 1 mm/s) tibial compression overload. HS injury creates an ACL rupture by midsubstance tear, while LS injury creates an ACL rupture with an associated avulsion bone fracture. At 10 days post-injury, vertebral trabecular bone structure was quantified using high-resolution microcomputed tomography (lCT), and differences in mechanical properties were determined using finite element modeling (FEM) and compressive mechanical testing. We hypothesized that knee injury would initiate a loss of trabecular bone structure and strength at the L5 vertebral body. Consistent with our hypothesis, we found significant decreases in trabecular bone volume fraction (BV/TV) and trabecular number at the L5 vertebral body in LS injured mice compared to sham (8.8% and 5.0%, respectively), while HS injured mice exhibited a similar, but lower magnitude response (5.1% and 2.5%, respectively). Contrary to our hypothesis, this decrease intrabecular bone structure did not translate to a significant deficit in compressive stiffness or ultimate load of the full trabecular body assessed by mechanical testing or FEM. However,we were able to detect significant decreases in compressive stiffness in both HS and LS injured specimens when FE models were loaded directly through the trabecular bone region (9.9% and 8.1%, and 3, respectively). This finding may be particularly important for osteoporotic fracture risk, as damage within vertebral bodies has been shown to initiate within the trabecular bone compartment. Altogether, these data point to a systemic trabecular bone loss as a consequence of fracture or traumatic musculoskeletal injury, which may be an underlying mechanism contributing to increased risk of refracture following an initial injury. This finding may have consequences for treatment of acute musculoskeletal injuries and the prevention of future bone fragility.

3D atlas-based registration can calculate malalignment of femoral shaft fractures in six degrees of freedom.

OBJECTIVE: This study presents and evaluates a semi-automated algorithm for quantifying malalignment in complex femoral shaft fractures from a single intraoperative cone-beam CT (CBCT) image of the fractured limb. METHODS: CBCT images were acquired of complex comminuted diaphyseal fractures created in 9 cadaveric femora (27 cases). Scans were segmented using intensity-based thresholding, yielding image stacks of the proximal, distal and comminuted bone. Semi-deformable and rigid affine registrations to an intact femur atlas (synthetic or cadaveric-based) were performed to transform the distal fragment to its neutral alignment. Leg length was calculated from the volume of bone within the comminution fragment. The transformations were compared to the physical input malalignments. RESULTS: Using the synthetic atlas, translations were within 1.71 ± 1.08 mm (medial/lateral) and 2.24 ± 2.11 mm (anterior/posterior). The varus/valgus, flexion/extension and periaxial rotation errors were 3.45 ± 2.6°, 1.86 ± 1.5° and 3.4 ± 2.0°, respectively. The cadaveric-based atlas yielded similar results in medial/lateral and anterior/posterior translation (1.73 ± 1.28 mm and 2.15 ± 2.13 mm, respectively). Varus/valgus, flexion/extension and periaxial rotation errors were 2.3 ± 1.3°, 2.0 ± 1.6° and 3.4 ± 2.0°, respectively. Leg length errors were 1.41 ± 1.01 mm (synthetic) and 1.26 ± 0.94 mm (cadaveric). The cadaveric model demonstrated a small improvement in flexion/extension and the synthetic atlas performed slightly faster (6 min 24 s ± 50 s versus 8 min 42 s ± 2 min 25 s). CONCLUSIONS: This atlas-based algorithm quantified malalignment in complex femoral shaft fractures within clinical tolerances from a single CBCT image of the fractured limb.

Strain-induced optical changes in demineralized bone.

Bone "stress-whitens," becoming visibly white during mechanical loading, immediately prior to failure. Stress-whitening is known to make materials tougher by dissipating mechanical energy. A greater understanding of stress-whitening, both an optical and mechanical phenomenon, may help explain age-related increases in fracture risk that occur without changes in bone mineralization. In this work, we directly measure the optical properties of demineralized bone as a function of deformation and immersing fluid (with different hydrogen-bonding potentials, water, and ethanol). The change in refractive index of demineralized bone was linear: with deformation and not applied force. Changes in refractive index were likely due to pushing low-refractive-index fluid out of specimens and secondarily due to changes in the refractive index of the collagenous phase. Results were consistent with stress-whitening of demineralized bone previously observed. In ethanol, the refractive index values were lower and less sensitive to deformation compared with deionized water, corroborating the sensitivity to fluid hydration. Differences in refractive index were consistent with structural changes in the collagenous phase such as densification that may also occur under mechanical loading. Understanding bone quality, particularly stress-whitening investigated here, may lead to new therapeutic targets and noninvasive methods to assess bone quality.

Quantification of the effect of osteolytic metastases on bone strain within whole vertebrae using image registration.

The vertebral column is the most frequent site of metastatic involvement of the skeleton with up to 1/3 of all cancer patients developing spinal metastases. Longer survival times for patients, particularly secondary to breast cancer, have increased the need for better understanding the impact of skeletal metastases on structural stability. This study aims to apply image registration to calculate strain distributions in metastatically involved rodent vertebrae utilizing µCT imaging. Osteolytic vertebral lesions were developed in five rnu/rnu rats 2-3 weeks post intracardiac injection with MT-1 human breast cancer cells. An image registration algorithm was used to calculate and compare strain fields due to axial compressive loading in metastatically involved and control vertebrae. Tumor-bearing vertebrae had greatly increased compressive strains, double the magnitude of strain compared to control vertebrae (p=0.01). Qualitatively strain concentrated within the growth plates in both tumor bearing and control vertebrae. Most interesting was the presence of strain concentrations at the dorsal wall in metastatically involved vertebrae, suggesting structural instability. Strain distributions, quantified by image registration were consistent with known consequences of lytic involvement. Metastatically involved vertebrae had greater strain magnitude than control vertebrae. Strain concentrations at the dorsal wall in only the metastatic vertebrae, were consistent with higher incidence of burst fracture secondary to this pathology. Future use of image registration of whole vertebrae will allow focused examination of the efficacy of targeted and systemic treatments in reducing strains and the related risk of fracture in pathologic bones under simple and complex loading.

Non-destructive evaluation of the effects of combined bisphosphonate and photodynamic therapy on bone strain in metastatic vertebrae using image registration.

Skeletal metastases most frequently affect the vertebral column and may lead to severe consequences including fracture. Clinical management of skeletal metastases often utilizes a multimodal treatment approach, including bisphosphonates (BPs). Previous work has demonstrated the synergistic potential of photodynamic therapy (PDT) in combination with BP in treating osteolytic disease through structural, histologic, and destructive mechanical testing analyses. Recent work has developed and validated image-based methods that may be used to non-destructively determine mechanical stability in whole bones, and enable their use for additional (i.e. histologic) analysis. In this work we use an intensity-based 3D image registration technique to compare the strain patterns throughout untreated control and BP + PDT treated rnu/rnu rat spinal motion segments with osteolytic metastases. It was hypothesized that the combination treatment will reduce average and maximum strain values and restore the pattern of strain to that of healthy vertebrae. Mean, median, and 90th percentile strains in the control group were significantly higher than the treatment group. High strain areas in both groups were observed around the endplates; in the control group, large areas of high strains were also observed around the lesions and adjacent to the dorsal wall. Absence of high strains adjacent to the dorsal wall (similar to healthy vertebrae) may correspond to a reduced risk of burst fracture following BP + PDT therapy. This study demonstrates the application of non-destructive image analysis to quantify the positive mechanical effects of combined BP + PDT treatment in the metastatic spine.

Exploring the role of 3-dimensional simulation in surgical training: feedback from a pilot study.

STUDY DESIGN: Randomized control study assessing the efficacy of a pedicle screw insertion simulator. OBJECTIVES: To evaluate the efficacy of an in-house developed 3-dimensional software simulation tool for teaching pedicle screw insertion, to gather feedback about the utility of the simulator, and to help identify the context and role such simulation has in surgical education. SUMMARY OF BACKGROUND DATA: Traditional instruction for pedicle screw insertion technique consists of didactic teaching and limited hands-on training on artificial or cadaveric models before guided supervision within the operating room. Three-dimensional computer simulation can provide a valuable tool for practicing challenging surgical procedures; however, its potential lies in its effective integration into student learning. METHODS: Surgical residents were recruited from 2 sequential years of a spine surgery course. Patient and control groups both received standard training on pedicle screw insertion. The patient group received an additional 1-hour session of training on the simulator using a CT-based 3-dimensional model of their assigned cadaver's spine. Qualitative feedback about the simulator was gathered from the trainees, fellows, and staff surgeons, and all pedicles screws physically inserted into the cadavers during the courses were evaluated through CT. RESULTS: A total of 185 thoracic and lumbar pedicle screws were inserted by 37 trainees. Eighty-two percent of the 28 trainees who responded to the questionnaire and all fellows and staff surgeons felt the simulator to be a beneficial educational tool. However, the 1-hour training session did not yield improved performance in screw placement. CONCLUSIONS: A 3-dimensional computer-based simulation for pedicle screw insertion was integrated into a cadaveric spine surgery instructional course. Overall, the tool was positively regarded by the trainees, fellows, and staff surgeons. However, the limited training with the simulator did not translate into widespread comfort with its operation or into improvement in physical screw placement.

Micro-computed tomography-based highly automated 3D segmentation of the rat spine for quantitative analysis of metastatic disease.

Noninvasive evaluation of metastatic disease in the spine has generally been limited to 2D qualitative or semiquantitative analysis techniques. This study aims to develop and evaluate a highly automated micro-CT-based quantitative analysis tool that can measure the architectural impact of metastatic involvement in whole vertebrae. Micro-CT analysis of rat whole vertebrae was conducted using a combination of demons deformable registration, level set curvature evolution, and intensity based thresholding techniques along with upsampling and edge enhancement techniques. The algorithm was applied to 6 lumbar vertebrae (L1-3) from 6 rnu/rnu rats (3 healthy rats and 3 with metastatic involvement). Osteolytic metastatic involvement was modeled via MT1 human breast cancer cells. Excellent volumetric concurrency was achieved in comparing the automated micro-CT-based segmentations of the whole vertebrae, trabecular centrums, and individual trabecular networks to manual segmentations (98.9%, 96.1%, and 98.3%, respectively; 6 specimens), and the automated segmentations were achieved in a fraction of the time. The algorithm successfully accounted for discontinuities in the cortical shell caused by vasculature and osteolytic destruction. As such, this work demonstrates the potential of this highly automated segmentation tool to permit rapid precise quantitative structural analysis of the spine with minimum user interaction in the analysis of both healthy and pathological (metastatically involved) vertebrae. Future optimization and the incorporation of lower-resolution imaging parameters may allow automated analysis of clinical CT-based measures in addition to preclinical micro-CT-based analyses of the structural impact and progression of pathological processes in the spine.

Effects of photodynamic therapy on the structural integrity of vertebral bone.

STUDY DESIGN: This study investigates the effects of photodynamic therapy (PDT) on the structural integrity of vertebral bone in healthy rats. OBJECTIVE: To determine the short-term (1 week) and intermediate term (6 weeks) effects of a single PDT treatment on the mechanical and structural properties of vertebral bone. SUMMARY OF BACKGROUND DATA: Spinal metastasis develops in up to one-third of all cancer patients, compromising the mechanical integrity of the spine and thereby increasing the risk of pathologic fractures and spinal cord damage. PDT has recently been adapted to ablate metastatic tumors in the spine in preclinical animal models. However, little is known about the effects of PDT on the structural integrity of vertebral bone. METHODS: A single PDT treatment was administered to healthy Wistar rats at photosensitizer and light doses known to be effective in athymic rats bearing human breast cancer metastases. At both 1 and 6 weeks posttreatment, changes in trabecular architecture, global stiffness and strength of vertebrae were quantified using micro-CT stereological analysis and axial compression testing. RESULTS: At 6 weeks, there was a significant increase in bone volume fraction (to 55.7 +/- 11.1% vs. 38.5 +/- 6.4%, P < 0.001) and decrease in bone surface area-to-volume ratio (16.9 +/- 5.0/mm vs. 22.8 +/- 4.5/mm, P = 0.001), attributed to trabecular thickening (130 +/- 40 microm vs. 90 +/- 20 microm, P < 0.001). Similar trends were found at 1 week after PDT. There was a significant increase in stiffness from control (306 +/- 123 N/mm) at 1 week (399 +/- 150 N/mm, P = 0.04) and 6 weeks (410 +/- 113 N/mm, P = 0.05) post PDT. There was a positive trend toward increased ultimate stress at 1 week, which became statistically significant at 6 weeks compared with control (39.3 +/- 11.3 MPa vs. 27.5 +/- 9.5 MPa control, P = 0.002). CONCLUSION: Not only may PDT be successful in ablating metastatic tumor tissue in the spine, but the positive effects of PDT on bone found in this study suggest that PDT may also improve vertebral mechanical stability.

Defining the therapeutic window of vertebral photodynamic therapy in a murine pre-clinical model of breast cancer metastasis using the photosensitizer BPD-MA (Verteporfin).

Breast cancer is known to cause metastatic lesions in the bone, which can lead to skeletal-related events. Currently, radiation therapy and surgery are the treatment of choice, but the success rate varies and additional adjuncts are desirable. Photodynamic therapy (PDT) has been applied successfully as a non-radiative treatment for numerous cancers. Earlier work has shown that the athymic rat model is suitable to investigate the effect of PDT on bone metastasis and benzoporphyrin-derivative monoacid ring A (BPD-MA; verteporfin) has been shown to be a selective photosensitizer. The aim of this study was to define the therapeutic window of photosensitizer with regard to drug and light dose. Human breast carcinoma cells (MT-1)-stable transfected with the luciferase gene-were injected intra-cardiacally into athymic rats. At 14 days, the largest vertebral lesion by bioluminescence imaging was targeted for single treatment PDT. A drug escalating-de-escalating scheme was used (starting drug dose and light energy of 0.2 mg/kg and 50 J, respectively). Outcomes included 48 h post-treatment bioluminescence of remaining viable tumour, histomorphometric assessment of tumour burden, and neurologic evaluation. The region of effect by bioluminescence and histology increased with increasing drug dose and light energy. A safe and effective drug-light dose combination in this model appears to be 0.5 mg/kg BPD-MA and applied light energy of less than 50 J for the thoracic spine and 1.0 mg/kg and 75 J for the lumbar spine. For translation to clinical use, it is an advantage that BPD-MA (verteporfin), a second-generation photosensitizer, is already approved to treat age-related macular degeneration. Overall, PDT represents an exciting potential new minimally-invasive local, safe and effective therapy in the management of patients with spinal metastases.

Whole bone strain quantification by image registration: a validation study.

Quantification of bone strain can be used to better understand fracture risk, bone healing, and bone turnover. The objective of this work was to develop and validate an intensity matching image registration method to accurately measure and spatially resolve strain in vertebrae using microCT imaging. A strain quantification method was developed that used two sequential microCT scans, taken in loaded and unloaded configurations. The image correlation algorithm implemented was a multiresolution intensity matching deformable registration that found a series of affine mapping between the unloaded and loaded scans. Once the registration was completed, the displacement field and strain field were calculated from the mappings obtained. Validation was done in two distinct ways: the first was to look at how well the method could quantify zero strain; the second was to look at how the method was able to reproduce a known applied strain field. Analytically defined strain fields that linearly varied in space and strain fields resulting from finite element analysis were used to test the strain measurement algorithm. The deformable registration method showed very good agreement with all cases imposed, establishing a detection limit of 0.0004 strain and displaying agreement with the imposed strain cases (average R2=0.96). The deformable registration routine developed was able to accurately measure both strain and displacement fields in whole rat vertebrae. A rigorous validation of any strain measurement method is needed that reports on the ability of the routine to measure strain in a variety of strain fields with differing spatial extents, within the structure of interest.

Mesh-morphing algorithms for specimen-specific finite element modeling.

Despite recent advances in software for meshing specimen-specific geometries, considerable effort is still often required to produce and analyze specimen-specific models suitable for biomechanical analysis through finite element modeling. We hypothesize that it is possible to obtain accurate models by adapting a pre-existing geometry to represent a target specimen using morphing techniques. Here we present two algorithms for morphing, automated wrapping (AW) and manual landmarks (ML), and demonstrate their use to prepare specimen-specific models of caudal rat vertebrae. We evaluate the algorithms by measuring the distance between target and morphed geometries and by comparing response to axial loading simulated with finite element (FE) methods. First a traditional reconstruction process based on microCT was used to obtain two natural specimen-specific FE models. Next, the two morphing algorithms were used to compute mappings from the surface of one model, the source, to the other, the target, and to use this mapping to morph the source mesh to produce a target mesh. The microCT images were then used to assign element-specific material properties. In AW the mappings were obtained by wrapping the source and target surfaces with an auxiliary triangulated surface. In ML, landmarks were manually placed on corresponding locations on the surfaces of both source and target. Both morphing algorithms were successful in reproducing the shape of the target vertebra with a median distance between natural and morphed models of 18.8 and 32.2 microm, respectively, for AW and ML. Whereas AW-morphing produced a surface more closely resembling that of the target, ML guaranteed correspondence of the landmark locations between source and target. Morphing preserved the quality of the mesh producing models suitable for FE simulation. Moreover, there were only minor differences between natural and morphed models in predictions of deformation, strain and stress. We therefore conclude that it is possible to use mesh-morphing techniques to produce accurate specimen-specific FE models of caudal rat vertebrae. Mesh morphing techniques provide advantages over conventional specimen-specific finite element modeling by reducing the effort required to generate multiple target specimen models, facilitating intermodel comparisons through correspondence of nodes and maintenance of connectivity, and lends itself to parametric evaluation of "artificial" geometries with a focus on optimizing reconstruction.