Precision of cortical bone reconstruction based on 3D CT scans

A Comprehensive Review of Vision-Based 3D Reconstruction Methods

With the rapid development of 3D reconstruction, especially the emergence of algorithms such as NeRF and 3DGS, 3D reconstruction has become a popular research topic in recent years. 3D reconstruction technology provides crucial support for training extensive computer vision models and advancing the development of general artificial intelligence. With the development of deep learning and GPU technology, the demand for high-precision and high-efficiency 3D reconstruction information is increasing, especially in the fields of unmanned systems, human-computer interaction, virtual reality, and medicine. The rapid development of 3D reconstruction is becoming inevitable. This survey categorizes the various methods and technologies used in 3D reconstruction. It explores and classifies them based on three aspects: traditional static, dynamic, and machine learning. Furthermore, it compares and discusses these methods. At the end of the survey, which includes a detailed analysis of the trends and challenges in 3D reconstruction development, we aim to provide a comprehensive introduction for individuals who are currently engaged in or planning to conduct research on 3D reconstruction. Our goal is to help them gain a comprehensive understanding of the relevant knowledge related to 3D reconstruction.

A Practical Method for Slice Spacing Measurement Using the American Association of Physicists in Medicine Computed Tomography Performance Phantom

Background

The slice spacing has a crucial role in the accuracy of computed tomography (CT) images in sagittal and coronal planes. However, there is no practical method for measuring the accuracy of the slice spacing.

Purpose

This study proposes a novel method to automatically measure the slice spacing using the American Association of Physicists in Medicine (AAPM) CT performance phantom.

Methods

The AAPM CT performance phantom module 610-04 was used to measure slice spacing. The process of slice spacing measurement involves a pair of axial images of the module containing ramp aluminum objects located at adjacent slice positions. The middle aluminum plate of each image was automatically segmented. Next, the two segmented images were combined to produce one image with two stair objects. The centroid coordinates of two stair objects were automatically determined. Subsequently, the distance between these two centroids was measured to directly indicate the slice spacing. For comparison, the slice spacing was calculated by accessing the slice position attributes from the DICOM header of both images. The proposed method was tested on phantom images with variations in slice spacing and field of view (FOV).

Results

The results showed that the automatic measurement of slice spacing was quite accurate for all variations of slice spacing and FOV, with average differences of 9.0% and 9.3%, respectively.

Conclusion

A new automated method for measuring the slice spacing using the AAPM CT phantom was successfully demonstrated and tested for variations of slice spacing and FOV. Slice spacing measurement may be considered an additional parameter to be checked in addition to other established parameters.

Database of segmentations and surface models of bones of the entire lower body created from cadaver CT scans

The range of applications of digital surface models of the bones in science and industry is wide. Three-dimensional reconstructions of bones are used in biomechanics, biomedical engineering, medical image processing, orthopedics, traumatology, radiology, patient education, anatomy, anthropometry, forensic anthropology, ergonomics, usability and human factors engineering, or accident and injury analysis and prevention. No open access database or repository of skeletal surface models of the full lower extremities exists. Therefore, the objective of this publication was to provide access to consistent complete bone models of the pelvis and lower limbs of multiple subjects, including biometric data. Segmentations and surface models of the bones of the lower extremities of more than twenty subjects were created from open access postmortem whole-body computed tomography scans. The database provides a broad range of applications by giving access to the data of the complete process chain, from the raw medical imaging data through the segmentations to the surface models.

Visualization and quantification of the degenerative pattern of the distal tibia and fibula in unilateral varus ankle osteoarthritis

The present study aimed to quantify and visualize the degenerative patterns of the distal tibia and fibula due to ankle osteoarthritis (OA). We analyzed differences in tibial and fibular surface deviation between sides of patients with unilateral varus ankle OA (medial talar tilt > 4°) by registering each surface model to the mirror image of corresponding bone. Computed tomography images of both feet of 33 patients (OA: 22, control: 11) were examined. Statistically significant surface depression of approximately 2.5 mm on the anterior articular surface of the medial malleolus, and surface elevation of approximately 1 mm on the anterodistal edge of the tibiofibular joint and the lateral malleolus were observed in OA patients. These bone degenerations were found to be correlated with those on the other side of the ankle joint, the medial margin of the talar trochlea and the lateral articular surface of the talus, respectively. In contrast, the amount of bone depression on the plafond was smaller than previously anticipated. Such quantitative information about stereotypical patterns of bone degeneration in ankle OA would contribute to better understanding of the development of ankle OA and possible therapeutic interventions.

Decreased mandibular cortical bone quality after botulinum toxin injections in masticatory muscles in female adults

This study aimed to clarify how masticatory muscle atrophy induced by botulinum toxin (BTX) injection affects cortical bone quality of the mandible using 3D modeling technology. A total of 39 young (26.9 ± 6.0 years) and 38 post-menopausal (55.3 ± 6.3 years) females were included. Computed tomography (CT) images were obtained before and after 12 months of treatment. Predictor variables were application of a stabilization splint, and/or two times of BTX injection in the bilateral temporalis and masseter muscles within a six-month interval. Outcome variables were changes in average Hounsfield units (HU) and cortical thickness of region of interest (ROI). 3D mandibular models were reconstructed using CT images, and models were used to calculate average HU and cortical thickness of ROIs, including inferior half of the lateral surface of ascending ramus, coronoid process, and temporomandibular joint condyle. Cortical bone quality at muscle insertion site was influenced by decreased muscle thickness but seemed not to be affected by decreased functional loading. Reduced functional loading seemed to influence cortical bone quality of the condyles. These effects were more remarkable in post-menopausal females. Hence, decreased masticatory muscle thickness may lead to alterations of the mandibular cortical structures, especially in post-menopausal females.

Performance of Cone Beam Computed Tomography Systems in Visualizing the Cortical Plate in 3D Image Reconstruction: An In Vitro Study

INTRODUCTION

Cortical bone is an important anatomical structure and its thickness needs to be determined prior to many dental procedures to ensure treatment success. Imaging modalities are necessarily used in dentistry for treatment planning and dental procedures. Three-dimensional image reconstruction not only provides visual information but also enables accurate measurement of anatomical structures; thus, it is necessarily required for maxillofacial examination and in case of skeletal problems in this region.

AIMS

This study aimed to assess the ability of three Cone Beam Computed Tomography (CBCT) systems including Cranex 3D, NewTom 3G and 3D Promax for Three-Dimensional (3D) image reconstruction of the cortical plate with variable thicknesses.

METHODS

Depending on the cortical bone thickness, samples were evaluated in three groups of ≤ 0. 5 mm, 0.6 -1 mm and 1.1-1.5 mm cortical bone thickness. The CBCT scans were obtained from each sample using three systems, their respective FOVs, and 3D scans were reconstructed using their software programs. Two observers viewed the images twice with a two-week interval. The ability of each system in the 3D reconstruction of different thicknesses of cortical bone was determined based on its visualization on the scans. The data were analyzed using SPSS and Kappa test.

RESULTS

The three systems showed the greatest difference in the 3D reconstruction of cortical bone with < 0.5 mm thickness. Cranex 3D with 4×6 cm2 FOV had the highest and 3D Promax with 8×8 cm2 FOV had the lowest efficacy for 3D reconstruction of cortical bone. Cranex 3D with 4×6 cm2 and 6×8 cm2 FOVs and NewTom 3G with 5×5 cm2 and 8×5 cm2 FOVs showed significantly higher efficacy for 3D reconstruction of cortical bone with 0.6-1mm thickness while 3D Promax followed by NewTom 3G with 8×8 cm2 FOV had the lowest efficacy for this purpose.

CONCLUSION

Most CBCT systems have high efficacy for 3D image reconstruction of cortical bone with thicknesses over 1 mm while they have poor efficacy for image reconstruction of cortical bone with less than 0.5 mm thickness. Thus, for accurate visualization of anatomical structures on CBCT scans, systems with smaller FOVs and consequently smaller voxel size are preferred.

Interlaboratory comparison of femur surface reconstruction from CT data compared to reference optical 3D scan

BACKGROUND

The present study contrasts the accuracy of different reconstructed models with distinctive segmentation methods performed by various experts. Seven research groups reconstructed nine 3D models of one human femur based on an acquired CT image using their own computational methods. As a reference model for accuracy assessment, a 3D surface scan of the human femur was created using an optical measuring system. Prior to comparison, the femur was divided into four areas; "neck and greater trochanter", "proximal metaphysis", "diaphysis", and "distal metaphysis". The deviation analysis was carried out in GEOMAGIC studio v.2013 software.

RESULTS

The results revealed that the highest deviation errors occurred in "neck and greater trochanter" area and "proximal metaphysis" area with RMSE of 0.84 and 0.83 mm respectively.

CONCLUSION

In conclusion, this study shows that the average deviation of reconstructed models prepared by experts with various methods, skills and software from the surface 3D scan is lower than 0.79 mm, which is not a significant discrepancy.

In Vivo Evaluation of 3D-Printed Polycaprolactone Scaffold Implantation Combined with β-TCP Powder for Alveolar Bone Augmentation in a Beagle Defect Model

Insufficient bone volume is one of the major challenges encountered by dentists after dental implant placement. This study aimed to evaluate the efficacy of a customized three-dimensional polycaprolactone (3D PCL) scaffold implant fabricated with a 3D bio-printing system to facilitate rapid alveolar bone regeneration. Saddle-type bone defects were surgically created on the healed site after extracting premolars from the mandibles of four beagle dogs. The defects were radiologically examined using computed tomography for designing a customized 3D PCL scaffold block to fit the defect site. After fabricating 3D PCL scaffolds using rapid prototyping, the scaffolds were implanted into the alveolar bone defects along with β-tricalcium phosphate powder. In vivo analysis showed that the PCL blocks maintained the physical space and bone conductivity around the defects. In addition, no inflammatory infiltrates were observed around the scaffolds. However, new bone formation occurred adjacent to the scaffolds, rather than directly in contact with them. More new bone was observed around PCL blocks with 400/1200 lattices than around blocks with 400/400 lattices, but the difference was not significant. These results indicated the potential of 3D-printed porous PCL scaffolds to promote alveolar bone regeneration for defect healing in dentistry.

ARE MRIs NECESSARY TO DEVELOP SUBJECT-SPECIFIC CARTILAGE AND MENISCI GEOMETRIES FOR SUBJECT-SPECIFIC KNEE MODELS?

Native subject-specific knee geometries are usually based on CT and MRI images reconstruction. Unfortunately, while the definition of bone geometries using CTs is quite consistent, MRIs are often hardly readable, due to the usual lower resolution, and the final shape of cartilage and menisci is not consequently detailed enough. Moreover, further smoothing techniques, necessary to efficiently use these structures for numerical modeling, could result in bad interfaces and/or geometry inaccuracies. In this study a CAD-based approach to generate 3D cartilages and menisci geometries, avoiding the use of MRIs, was proposed and tested versus the traditional methods that use MRIs segmentation. The femoral, tibial and patellar cartilage layers were generated as offset from the bone geometries, the menisci were obtained by an extrusion based on tibia borders. Such geometries were compared to the reconstructions obtained from MRIs of healthy knee specimens. Overlapping the resulting geometries with the ones traditionally reconstructed, volumes differ from 2% to 14%. By using the new methodology, the geometries are obtained in 75% less time. The CAD-based methods shown in this pilot study is able to generate faster and accurate subject-specific knee cartilage layers and menisci geometries and can be suitable to be applied for numerical modeling.

Quantitative analysis of the patella following the harvest of a quadriceps tendon autograft with a bone block

PURPOSE

The objective of this study was to determine parameters associated with patellar fracture after quadriceps tendon autograft harvest.

METHODS

Thirteen non-fractured and five fractured patella surface models were created based on patient data obtained from a prospective randomized clinical trial in order to assess geometric parameters and bending stress. Measurements that describe the bone block harvest site geometry were used to calculate three normalized parameters. The relative depth parameter describes the thickness of the bone block harvest site with respect to the thickness of the patella at the harvest site. The asymmetry parameter defines the medial-lateral location of the bone bock harvest site. The normalized bending stress parameter assesses the bending stress experienced by the remaining bone beneath the bone block harvest site.

RESULTS

The relative depth of the bone block harvest site in the non-fractured patellae was 27 ± 12 % and for the fractured patellae was 42 ± 14 % (p < 0.05). With a value <1 indicating a more lateral location of the harvest site, asymmetry for the non-fractured group was 1.0 ± 0.5 and 0.7 ± 0.4 for the fractured group (n.s.). The maximum bending stress experienced by the non-fractured patellae was (1.8 × 10(-3) ± 1.3 × 10(-3)) mm(-3) × M and for the fractured patellae was over three times greater (6.3 × 10(-3) ± 3.7 × 10(-3)) mm(-3) × M (p < 0.05).

CONCLUSION

Based on the non-uniform geometry of the patella, an emphasis should be made on harvesting a standard percentage of patella thickness rather than a fixed depth. In order to minimize the incidence of a patellar fracture, bone blocks should not be taken laterally and should not exceed 30 % of the total patella thickness at the harvest site.

Accuracy assessment of 3D bone reconstructions using CT: an intro comparison

Computed tomography provides high contrast imaging of the joint anatomy and is used routinely to reconstruct 3D models of the osseous and cartilage geometry (CT arthrography) for use in the design of orthopedic implants, for computer assisted surgeries and computational dynamic and structural analysis. The objective of this study was to assess the accuracy of bone and cartilage surface model reconstructions by comparing reconstructed geometries with bone digitizations obtained using an optical tracking system. Bone surface digitizations obtained in this study determined the ground truth measure for the underlying geometry. We evaluated the use of a commercially available reconstruction technique using clinical CT scanning protocols using the elbow joint as an example of a surface with complex geometry. To assess the accuracies of the reconstructed models (8 fresh frozen cadaveric specimens) against the ground truth bony digitization-as defined by this study-proximity mapping was used to calculate residual error. The overall mean error was less than 0.4 mm in the cortical region and 0.3 mm in the subchondral region of the bone. Similarly creating 3D cartilage surface models from CT scans using air contrast had a mean error of less than 0.3 mm. Results from this study indicate that clinical CT scanning protocols and commonly used and commercially available reconstruction algorithms can create models which accurately represent the true geometry.

Three-dimensional printing to facilitate anatomic study, device development, simulation, and planning in thoracic surgery

BACKGROUND

The development and deployment of new technologies in additive 3-dimensional (3D) printing (ie, rapid prototyping and additive manufacturing) in conjunction with medical imaging techniques allow the creation of anatomic models based on patient data.

OBJECTIVE

To explore this rapidly evolving technology for possible use in care and research for patients undergoing thoracic surgery.

METHODS

Because of an active research project at our institution on regional lung chemotherapy, human pulmonary arteries (PAs) were chosen for this rapid prototyping project. Computed tomography (CT) and CT angiography in combination with segmentation techniques from 2 software packages were used for rapid generation and adjustment of the 3D polygon mesh and models reconstruction of the PAs. The reconstructed models were exported as stereolithographic data sets and further processed by trimming, smoothing, and wall extrusion.

RESULTS

Flexible 3D printed replicas of 10 patient PAs were created successfully with no print failures; however, 1 initial test print with a 1 mm mural thickness was too fragile so the whole group was printed with a 1.5 mm wall. The design process took 8 hours for each model (CT image to stereolithographic) and printing required 97 hours in aggregate. Useful differences in anatomy were defined by this method, such as the expected greater number of proximal branches on the left versus right (2.5 ± 1.1 vs 1.0 ± 0.0; P = .001).

CONCLUSIONS

Reconstructed models of pulmonary arteries using 3D rapid prototyping allow replication of sophisticated anatomical structures that can be used to facilitate anatomic study, surgical planning, and device development.

Modelling and analysis on biomechanical dynamic characteristics of knee flexion movement under squatting

The model of three-dimensional (3D) geometric knee was built, which included femoral-tibial, patellofemoral articulations and the bone and soft tissues. Dynamic finite element (FE) model of knee was developed to simulate both the kinematics and the internal stresses during knee flexion. The biomechanical experimental system of knee was built to simulate knee squatting using cadaver knees. The flexion motion and dynamic contact characteristics of knee were analyzed, and verified by comparing with the data from in vitro experiment. The results showed that the established dynamic FE models of knee are capable of predicting kinematics and the contact stresses during flexion, and could be an efficient tool for the analysis of total knee replacement (TKR) and knee prosthesis design.

Three-dimensional reconstruction of computed tomography scan images for estimating skull damage in electrical burned patients

Three cases of skull osteomyelitis due to electrical burn and delayed wound closure are presented. For better estimating skull damage before operation, 3-dimensional reconstruction of computed tomography scan images were used. Three-dimensional computed tomography could provide superior and visible stereoscopic images and help clinicians "see" the damage before operation and make more detailed therapeutic planning.

The Hounsfield value for cortical bone geometry in the proximal humerus--an in vitro study

INTRODUCTION

Fractures of the proximal humerus represent a major osteoporotic burden. Recent developments in CT imaging have emphasized the importance of cortical bone thickness distribution in the prevention and management of fragility fractures. We aimed to experimentally define the CT density of cortical bone in the proximal humerus for building cortical geometry maps.

METHODS

With ethical approval, we used ten fresh-frozen human proximal humeri. These were stripped of all soft tissue and high-resolution CT images were then taken. The humeral heads were then subsequently resected to allow access to the metaphyseal area. Using curettes, cancellous bone was removed down to hard cortical bone. Another set of CT images of the reamed specimen was then taken. Using CT imaging software and a CAD interface, we then compared cortical contours at different CT density thresholds to the reference inner cortical contour of our reamed specimens. Working with 3D model representations of these cortical maps, we were able to accurately make distance comparison analyses based on different CT thresholds.

RESULTS

We could compute a single closest value at 700 HU. No difference was found in the HU-based contours generated along the 500-900 HU pixels (p = 1.000). The contours were significantly different from those generated at 300, 400, 1,000, and 1,100 HU.

CONCLUSIONS

A Hounsfield range of 500-900 HU can accurately depict cortical bone geometry in the proximal humerus. Thresholding outside this range leads to statistically significant inaccuracies. Our results concur with a similar range reported in the literature for the proximal femur. Knowledge of regional variations in cortical bone thickness has direct implications for basic science studies on osteoporosis and its treatment, but is also important for the orthopedic surgeon since our decision for treatment options is often guided by local bone quality.

A computerized analysis of femoral condyle radii in ACL intact and contralateral ACL reconstructed knees using 3D CT

The bony geometry of the distal femoral condyles may have a significant influence on knee joint kinematics. The aim of this study was to analyze the relationship between the size of the medial and lateral femoral condyles in different planes. Seventy-four three dimensional (3D) CT reconstructions of 37 patients with ACL intact and contralateral ACL reconstructed knees were used and the data were imported into a graphical software program. The radii of the medial and lateral femoral condyles were analyzed in the sagittal, coronal, and axial planes by digitally reconstructed circular arcs along the bony condylar profiles marked with multiple digital surface points. Intra- and interobserver testing was performed. In the intact knees the average sagittal radius of the distal medial and lateral femoral condyles was similar. There was a significant difference between the radii of the distal medial femoral condyles compared to lateral femoral condyles in the coronal plane (22.4 vs. 27.8 mm, P < 0.001) as well as between the radii of the medial femoral condyles in the axial plane in 90 knee flexion compared to the lateral femoral condyles (21.3 vs. 18.3 mm, P < 0.001). The average radius of the medial femoral condyles was significantly smaller in extension compared to 90 of flexion (21.2 vs. 22.4 mm, P = 0.05) and the average radius of the lateral femoral condyles was significantly larger in extension compared to 90 of flexion (27.8 vs. 18.3 mm, P < 0.001). The 37 ACL reconstructed knees demonstrated similar radii in all three planes when compared to the intact knees without any significant difference. The described method of assessing the architecture of the distal femoral condyles is non-invasive, reproducible, and provides reliable geometric parameters necessary for the 3D reconstruction of the femoral geometry in vivo. The radii of the FC were similar in the sagittal planes but demonstrate a significant asymmetry in the axial and coronal planes. The average radius of the lateral femoral condyles was significantly larger in extension whereas the radius of the medial femoral condyles was significantly larger in flexion. We did not find any significant difference in the shape of the femoral condyles in ACL intact and contralateral ACL reconstructed knees indicating that the geometry of the femoral condyles might not influence the injury mechanism of ACL rupture. The asymmetry between the femoral condyles may be considered when designing new anatomical femoral components in knee arthroplasty.