Manual Landmark Identification Research Articles

Accuracy of web-based automated versus digital manual cephalometric landmark identification.

The purpose of this study was to assess the accuracy of two web-based automated cephalometric landmark identification and analysis programs. Manual landmark identification using Dolphin Imaging software was used as reference. 105 cephalograms were selected and divided into three groups of 35 subjects each, Class I, II and III. Radiographs were traced using Dolphin imaging software. WebCeph™ (South Korea) and Cephio™ (Poland) were used for the automated cephalometric analysis. Bland-Altman limits of agreement and the concordance correlation coefficient (CCC) were calculated. Kruskal Wallis test was used to compare the accuracy of WebCeph™ and Cephio™ measurements between the three groups. Mann-Whitney U test was used to compare the absolute difference between cephalometric measurements obtained using WebCeph™ and Cephio™. The mean difference (MD) between AI and manually-derived measurements was less than 1mm/degree and ranged from 0.01 to 0.8 except for upper lip protrusion (MD 1.35°), nasolabial angle (MD 5.01°), SN-GoGn (MD 1.41°), Ramus height (MD 1.46°), and IMPA (MD 1.94°). The mean CCC was 0.91 (range 0.60 to 0.96). No statistically significant differences were found between the three malocclusion groups for most of the measurements (P > 0.05). For most of the measurements, automated cephalometric measurements were clinically acceptable. Few differences were found between Webceph™ and Cephio™ for most measurements. Measurements including SNA, SN-PP, IMPA as well as soft tissue measurements require extra consideration and manual adjustment of respective landmarks for higher precision and improved efficiency.

Accuracy of and dental students' preferences toward manual and digital cephalometric landmark identification: A randomized cross-over study.

To evaluate dental students' perceptions of manual and digital cephalometric landmark identification methods based on their preferences, difficulty level, and procedure time required to provide insights into the future of dental education, considering incorporating digital technology in dental schools. Fifty-five second-year dental students were randomly divided into two groups: (1) group A, students who performed manual landmark identification first, followed by digital method; and (2) group B, students who performed digital method first, followed by manual method. The duration of the procedure was recorded. Subsequently, all students completed a questionnaire regarding the difficulty they experienced using a visual analog scale and their preferences. Landmark identification accuracy was measured. Digital landmark identification was preferred by 93% of students. The mean procedure time for digital method was significantly lower than that of manual method (13.00±5.60vs. 9.70±4.60; p=0.002). Group B completed manual and digital methods in a shorter time than group A. Group A experienced less difficulty with manual procedure than group B. However, statistically significant differences were not observed in the difficulty level of digital technique. A statistically significant difference in the mean accuracy was shown in favor of the manual method. However, this difference is clinically insignificant (p=0.001). Students considered digital method to be effective for learning and preferred it over manual method. Furthermore, digital landmark identification demonstrated better performance and was faster than manual method, suggesting that this must be incorporated in undergraduate dental education.

An Evaluation of the Reliability of Manual Landmark Identification on 3D Segmented Wrists.

Three-dimensional (3D) preoperative planning is increasingly used in orthopaedic surgery. Two-dimensional (2D) characterization of distal radial deformities remains inaccurate, and 3D planning requires a reliable reference frame at the wrist. We aim to evaluate the reliability of the determination of anatomical points placed manually on 3D models of the radius to determine which of those points allow reliable morphometric measurements. Twenty-three radial scans were reconstructed in 3D. Five operators specialized in the upper limb manually positioned 8 anatomical points on each model. One of the operators repeated the operation 6 times. The anatomical points were based on previously published 3D models used for radial inclination and dorsopalmar tilt measurements. The repeatability and reproducibility of the measurements derived using this manual landmarking were calculated using different measurement methods based on the identified points. An error of ≤2° was considered clinically acceptable. This study of intraobserver and interobserver variability of the anatomic points allowed us to determine the least variable and most accurately defined points. The middle of the ulnar border of the radius, the radial styloid, and the midpoint of the ulnar incisura of the radius were the least variable. The palmar and dorsal ends of the ridge delineating the scaphoid and lunate facets were the most variable. Only 1 of the radial inclination measurement methods was clinically acceptable; the others had a repeatability and reproducibility limit of >2°, making those measurements clinically unacceptable. The use of isolated points seems insufficient for the development of a wrist reference frame, especially for the purpose of measuring dorsopalmar tilt. If one concurs that an error of 2° is unacceptable for all distal radial measurements, then clinicians should avoid using 3D landmarked points, due to their unreliability, except for radial inclination measured using the radial styloid and the midpoint of the ulnar edge of the radius. A characterization of the wrist using 3D shapes that fit the articular surface of the radius should be considered. Diagnostic Level III . See Instructions for Authors for a complete description of levels of evidence.

Threshold-Based Segmentation for Landmark Detection Using CBCT Images

The aim of this study is to examine the influence of threshold-based segmentation on the mean error of automatic landmark detection in 3D CBCT images. A GUI was developed for radiologists, allowing manual landmark identification and visualization of CBCT images. After a threshold-based segmentation, a semi-automatic algorithm for landmark detection was designed using the anatomic definition of each landmark. A step of 50 Hounsfield units was used for threshold variation to assess the detection error. 5 CBCT images were used to validate the proposed approach. The measurement of error detection for one patient was influenced by the threshold variation. For this patient, the error changed from 1.49 mm to 10.32 mm at a low threshold value, while for another patient, the error changed from 1.96 mm to 12.28 mm at high a threshold value. In a CBCT scanner, the choice of threshold value for segmentation can be an important factor in causing error in measurements.

Preciseness of artificial intelligence for lateral cephalometric measurements.

The aim of the study was to assess the accuracy and efficiency of anew artificial intelligence (AI) method in performing lateral cephalometric radiographic measurements. A total of 200lateral cephalometric radiographs were assessed for quality and included. Three methods were used to perform the cephalometric measurements: (1) the AI method using WebCeph software (AssembleCircle Corp., Gyeonggi-do, Republic of Korea), (2)the modified AI method using WebCeph software after manual modification of the landmarks' position, and (3)using OnyxCeph software (Image Instruments GmbH, Chemnitz, Germany) by manual landmark identification and digital measurements generation. The results of the measurements produced by the three methods were compared, in addition to comparing the time required for the measurements' generation required for each method. Statistically significant differences were detected between the measurements resulting from the three used methods. Fewer differences were detected between the modified AI method and the OnyxCeph method. The AI method produced the measurements the fastest followed by the modified AI method and then the OnyxCeph method. Considering the used AI software, AI followed by manual tuning of the landmarks' position might be an accurate method in lateral cephalometric analysis. AI alone is still not fully reliable at locating the different landmarks on the lateral cephalometric radiographs.

Reliability of Artificial Intelligence-Assisted Cephalometric Analysis. A Pilot Study

Recently, Artificial Intelligence (AI) has spread in orthodontics, in particular within cephalometric analysis, where computerized digital software is able to provide linear-angular measurements upon manual landmark identification. A step forward is constituted by fully automated AI-assisted cephalometric analysis, where the landmarks are automatically detected by software. The aim of the study was to compare the reliability of a fully automated AI-assisted cephalometric analysis with the one obtained by a computerized digital software upon manual landmark identification. Fully automated AI-assisted cephalometric analysis of 13 lateral cephalograms were retrospectively compared to the cephalometric analysis performed twice by a blinded operator with a computerized software. Intra- and inter-operator (fully automated AI-assisted vs. computerized software with manual landmark identification) reliability in cephalometric parameters (maxillary convexity, facial conicity, facial axis angle, posterior and lower facial height) was tested with the Dahlberg equation and Bland–Altman plot. The results revealed no significant difference in intra- and inter-operator measurements. Although not significant, higher errors were observed within intra-operator measurements of posterior facial height and inter-operator measurements of facial axis angle. In conclusion, despite the small sample, the cephalometric measurements of a fully automated AI-assisted cephalometric software were reliable and accurate. Nevertheless, digital technological advances cannot substitute the critical role of the orthodontist toward a correct diagnosis.

Errors in Measuring Glenohumeral Arthrokinematics With 2-Dimensional Fluoroscopy.

Two-dimensional fluoroscopic imaging allows measurement of small magnitude humeral head translations that are prone to errors due to optical distortion, out-of-plane imaging, repeated manual identification of landmarks, and magnification. This article presents results from in vivo and in vitro fluoroscopy-based experiments that measure the errors and variability in estimating the humeral head translated position in true scapular plane and axillary views. The errors were expressed as bias and accuracy. The variability with repeated digitization was calculated using the intraclass correlation coefficient (ICC) and the standard error of measurement. Optical distortion caused underestimation of linear distances. The accuracy was 0.11 and 0.43mm for in vitro and in vivo experiments, respectively, for optical distortion. The intrarater reliability was excellent for both views (ICC = .94 and .93), and interrater reliability was excellent (ICC = .95) for true scapular view but moderate (ICC = .74) for axillary views. The standard error of measurement ranged from 0.27 to 0.58mm. The accuracy for the humeral head position in 10° out of true scapular plane images ranged from 0.80 to 0.87mm. The current study quantifies the magnitude of error. The results suggest that suitable measures could be incorporated to minimize errors and variability for the measurement of glenohumeral parameters.

Machine Learning Applications in Orthopaedic Imaging.

Machine Learning Applications in Orthopaedic Imaging.

Precision of manual landmark identification between as-received and oriented volume-rendered cone-beam computed tomography images.

The objective of this study was to evaluate the effect of the orientation of cone-beam computed tomography (CBCT) images on the precision and reliability of 3-dimensional cephalometric landmark identification. Ten CBCT scans were used for manual landmark identification. Volume-rendered images were oriented by aligning the Frankfort horizontal and transorbital planes horizontally, and the midsagittal plane vertically. A total of 20 CBCT images (10 as-received and 10 oriented) were anonymized, and 3 random sets were generated for manual landmark plotting by 3 expert orthodontists. Twenty-five landmarks were identified for plotting on each anonymized image independently. Hence, a total of 60 images were marked by the orthodontists. After landmark plotting, the randomized samples were decoded and regrouped into as-received and oriented data sets for analysis and comparison. Means and standard deviations of the x-, y-, and z-axis coordinates were calculated for each landmark to measure the central tendency. Intraclass correlation coefficients were calculated to analyze the interobserver reliability of landmark plotting in the 3 axes in both situations. Paired t tests were applied on the mean Euclidean distance computed separately for each landmark to evaluate the effect of 3-dimensional image orientation. Interobserver reliability (intraclass correlation coefficient,>0.9) was excellent for all 25 landmarks for the x-, y-, and z-axes on both before and after orientation of the images. Paired t test results showed insignificant differences for the orientation of volume-rendered images for all landmarks except 3: R1 left (P=0.0138), sella (P=0.0490), and frontozygomatic left (P=0.0493). Also midline structures such as Bolton and nasion were plotted more consistently or precisely than bilateral structures. Orientation of the CBCT image does not enhance the precision of landmark plotting if each landmark is defined properly on multiplanar reconstruction slices and rendered images, and the clinician has sufficient training. The consistency of landmark identification is influenced by their anatomic locations on the midline, bilateral, and curved structures.

A knowledge-based algorithm for automatic detection of cephalometric landmarks on CBCT images.

Cone-beam computed tomography (CBCT) is now an established component for 3D evaluation and treatment planning of patients with severe malocclusion and craniofacial deformities. Precision landmark plotting on 3D images for cephalometric analysis requires considerable effort and time, notwithstanding the experience of landmark plotting, which raises a need to automate the process of 3D landmark plotting. Therefore, knowledge-based algorithm for automatic detection of landmarks on 3D CBCT images has been developed and tested. A knowledge-based algorithm was developed in the MATLAB programming environment to detect 20 cephalometric landmarks. For the automatic detection, landmarks that are physically adjacent to each other were clustered into groups and were extracted through a volume of interest (VOI). Relevant contours were detected in the VOI and landmarks were detected using corresponding mathematical entities. The standard data for validation were generated using manual marking carried out by three orthodontists on a dataset of 30 CBCT images as a reference. Inter-observer ICC for manual landmark identification was found to be excellent (>0.9) amongst three observers. Euclidean distances between the coordinates of manual identification and automatic detection through the proposed algorithm of each landmark were calculated. The overall mean error for the proposed method was 2.01 mm with a standard deviation of 1.23 mm for all the 20 landmarks. The overall landmark detection accuracy was recorded at 64.67, 82.67 and 90.33 % within 2-, 3- and 4-mm error range of manual marking, respectively. The proposed knowledge-based algorithm for automatic detection of landmarks on 3D images was able to achieve relatively accurate results than the currently available algorithm.

Automated landmark identification for human cortical surface-based registration

Volume-based registration (VBR) is the predominant method used in human neuroimaging to compensate for individual variability. However, surface-based registration (SBR) techniques have an inherent advantage over VBR because they respect the topology of the convoluted cortical sheet. There is evidence that existing SBR methods indeed confer a registration advantage over affine VBR. Landmark-SBR constrains registration using explicit landmarks to represent corresponding geographical locations on individual and atlas surfaces. The need for manual landmark identification has been an impediment to the widespread adoption of Landmark-SBR. To circumvent this obstacle, we have implemented and evaluated an automated landmark identification (ALI) algorithm for registration to the human PALS-B12 atlas. We compared ALI performance with that from two trained human raters and one expert anatomical rater (ENR). We employed both quantitative and qualitative quality assurance metrics, including a biologically meaningful analysis of hemispheric asymmetry. ALI performed well across all quality assurance tests, indicating that it yields robust and largely accurate results that require only modest manual correction (<10 min per subject). ALI largely circumvents human error and bias and enables high throughput analysis of large neuroimaging datasets for inter-subject registration to an atlas.

Automatic identification of landmarks in digital images

The authors present an automated system for feature recognition in digital images. Morphometric landmarks are points that can be defined in all specimens and located precisely. They are widely used in shape analysis and a typical shape analysis study involves several hundred digital images. Presently, the extraction of landmarks is usually done manually and the process of identifying the landmarks is an important and labour-intensive part of any such analysis. This process is time-consuming, and quite often the research questions are dependent on the duration of obtaining these data. The authors show that a single training image with its landmark co-ordinates is enough to independently estimate the landmarks of any individual within a particular data set. The reliability and accuracy of the method can be further enhanced by using multiple training images. The precision, repeatability and robustness of the algorithm have been evaluated. It is shown in this study that the method is sufficiently accurate to replace the manual identification of landmarks. The generic nature and intrinsic capability of the feature recognition process enables this method to be easily incorporated into other recognition tasks.

Image Registration and Localization Accuracy of a Micro IGRT System

The μIGRT systems bridge the technology gap between pre-clinical radiation research and clinical radiation therapy. Image registration enables accurate target localization based on reference data used for treatment plan design. The purpose of this research was to quantify the registration and localization accuracy of a μCBCT guided radiation therapy system. Images of a cylindrical phantom with five embedded fiducial markers and two mouse cadavers were acquired with a μCT scanner (eXplore Ultra, GE Healthcare, London, ON) and/or the CBCT scanner of a μIGRT system (XRad 225Cx, Precision Inc X-rays, North Branford, USA). Voxel sizes were 0.15x0.15x0.15 mm3 and 0.2x0.2x0.2 mm3 for the μCT scan and μCBCT scans respectively. Dicom images were imported into an in-house developed image registration software package. Fiducial markers for the phantom and naturally occurring calcifications in the abdomen for the mice were manually identified and their locations were stored in the coordinate system of the irradiator. Corresponding scans were aligned using local rigid registration with a shaped region of interest and correlation-ratio as a similarity measure. For the phantom, the fiducial markers were excluded from the region of interest. After registration the distance between the manually placed landmarks was calculated. To quantify the overall localization accuracy, a cylindrical phantom containing a Gafchromic film with a small pinched hole was scanned on the μIGRT system, imported into the image registration tool as a reference scan and the hole was marked as the treatment iso-center. The phantom was manually displaced, rescanned and imported as the localization scan. Following image registration the phantom was repositioned using a robotic couch adjustment. Finally the film was exposed to 8 equi-angular spaced beams and the distance between the pinched hole and the center of the beam intersection was manually determined. This was repeated 5 times. For the phantom the residual landmark distance was 19 μm ± 94 μm (1 SD) in the axial plane and 121 μm ± 86 μm in the longitudinal direction. For the mice cadavers this was about 40 μm ± 60 μm in all directions. The average localization accuracy was 167 μm ± 72 μm (1 SD). We quantified the registration and localization accuracy of a μIGRT system and observed an overall delivery accuracy better than 200 μm including manual landmark identification. Multi-modality image registration and quantification of the localization accuracy for living specimen are subject of further study.

Ex vivo radiographic tooth length measurements with the reference sphere method (RSM)

A reference-based radiographic "reference sphere method" (RSM) for accurate length measurements in (dental) projection radiographs for the assessment of tooth length in dry human mandible sections is evaluated. RSM determines the depth coordinates of reference spheres placed in the object plane from the elliptical distortion of their shadows. Two segments (one canine and one molar) of dry human mandibles were exposed 95 times at different angulations (0-40°) on a dental charge-coupled device receptor. Three steel spheres (diameters d (1) = 2.00 mm, d (2) = 3.00 mm) were attached roughly coplanar with the tooth's main axis. Radiographs were assessed once by visual inspection plus manual landmark identification with a mouse-driven cursor. The results were compared to the true tooth length assessed after extraction and to a conventional method (C), i.e., the rule of proportion based on magnification of the sphere shadows. Mean relative length error was 2.28% (d (1)) and 0.46% (d (2)) for RSM and -13.58% (d (1)) and -9.90% (d (2)) for C. For both methods, length errors were significantly (p < 0.0001) correlated with the inclination relative to the receptor. RSM allows for complete a posteriori determination of the imaging geometry under the assumption of a known source-to-receptor distance. One specific application is foreshortening correction of objects coplanar with the reference spheres. Remaining errors are mainly due to incorrect landmark definition. In our setup, these were exaggerated by the visual/manual image-evaluation process. Automated image analysis has been shown for similar tasks to minimize these errors considerably.

Landmark‐based software for anatomical measurements: A precision study

The aim of this study was to develop a software program, called Landmarker, which would aid studies of complex anatomical morphometry by simplifying the manual identification of landmarks in 3D images. We also tested its precision on routine magnetic resonance imaging (MRI) scans. To understand human biological variation, there is a need to identify morphological characteristics from the exterior and the interior of human anatomy. MRI, as opposed to other radiographic methods (mainly based on X-ray techniques), supplies good soft tissue contrast, which allows for more complex assessments than what bony landmarks can provide. Because automation of this assessment is highly demanding, one of the primary goals for the new software was to enable more rapid identification of landmark sets in 3D image data. Repeat acquisition of head MRIs having a resolution of 0.94 x 0.94 x 1.20 mm3 were performed on 10 volunteers. Intra- and interoperator, as well as interacquisition variations of manual identification of exterior, craniofacial interior, and brain landmarks were studied. The average distances between landmarks were <1.8 mm, <2.3 mm, and <2.0 mm in the intra- and interoperator, and interacquisition evaluations, respectively. This study presents new software for time efficient identification of complex craniofacial landmarks in 3D MRI. To the best of our knowledge, no evaluation of software for rapid landmark-based analysis of complex anatomies from 3D MR data has yet been presented. This software may also be useful for studies in other anatomical regions and for other types of image data.

Manual landmark identification and tracking during the medial rotation test of the shoulder: An accuracy study using three-dimensional ultrasound and motion analysis measures

Palpation of movement is a common clinical tool for assessment of movement in patients with musculoskeletal symptoms. The purpose of this study was to measure the accuracy of palpation of shoulder girdle translation during the medial rotation test (MRT) of the shoulder. The translation of the gleno-humeral and scapulo-thoracic joints was measured using both three-dimensional ultrasound and palpation in order to determine the accuracy of translation tracking during the MRT of the shoulder. Two movements of 11 normal subjects (mean age 24 (SD=4), range 19–47 years) were measured. The agreement between measures was good for scapulo-thoracic translation ( r=0.83). Gleno-humeral translation was systematically under estimated ( p=0.03) although moderate correlation was found ( r=0.65). These results indicate that translation of the measured joints can be tracked by palpation and further tests of the efficacy of palpation tracking during musculoskeletal assessment may be warranted.

Generation of a statistical shape model with probabilistic point correspondences and the expectation maximization- iterative closest point algorithm

Objective: Identification of point correspondences between shapes is required for statistical analysis of organ shapes differences. Since manual identification of landmarks is not a feasible option in 3D, several methods were developed to automatically find one-to-one correspondences on shape surfaces. For unstructured point sets, however, one-to-one correspondences do not exist but correspondence probabilities can be determined.

Voxel-based modeling and quantification of the proximal femur using inter-subject registration of quantitative CT images

We have developed a general framework which employs quantitative computed tomography (QCT) imaging and inter-subject image registration to model the three-dimensional structure of the hip, with the goal of quantifying changes in the spatial distribution of bone as it is affected by aging, drug treatment or mechanical unloading. We have adapted rigid and non-rigid inter-subject registration techniques to transform groups of hip QCT scans into a common reference space and to construct composite proximal femoral models. We have applied this technique to a longitudinal study of 16 astronauts who on average, incurred high losses of hip bone density during spaceflights of 4–6 months on the International Space Station (ISS). We compared the pre-flight and post-flight composite hip models, and observed the gradients of the bone loss distribution. We performed paired t-tests, on a voxel by voxel basis, corrected for multiple comparisons using false discovery rate (FDR), and observed regions inside the proximal femur that showed the most significant bone loss. To validate our registration algorithm, we selected the 16 pre-flight scans and manually marked 4 landmarks for each scan. After registration, the average distance between the mapped landmarks and the corresponding landmarks in the target scan was 2.56 mm. The average error due to manual landmark identification was 1.70 mm.

Automated image analysis technique for measurement of femoral component subsidence in total hip joint replacement

A new technique for the measurement of subsidence of the femoral components is proposed. The method relies on the implantation of two ball markers around the femoral stem. A single radiographic image of the hip is analysed using image-processing techniques to minimize subjectivity related to manual identification of landmarks. Dimensions of the stem are used to correct for magnification and out-of-plane rotations resulting from radiographic positioning. This technique has been applied to a specific design of implant (Exeter™). A study of the effect of radiographic positioning has been conducted using a cadaveric bone phantom. Results for the variation in the measurement of axial migration compared to the neutral position (in millimetres) were: 0.942 (10° extension); 0.347 (20° flexion); −0.435 (40° internal rotation); 0.187 (30° external rotation) for distances measured between the bone marker and the implant. Results for distances measured between the implant and the cement centralizer were: 0.107 (10° extension); −0.277 (20° flexion); 0.085 (40° internal rotation); 0.280 (30° external rotation). The variations from within a more realistic range of positions demonstrate that axial migration measurements of ca. 0.5 mm between the bone and implant, and less than 0.3 mm between the implant and the bone cement, may be expected.