Groupwise 2012 9509 Please Enter Your Password Again
one.
Introduction
Traumatic encephalon injury (TBI) is a leading cause of morbidity and bloodshed with an estimated 1.half-dozen million hospitalized and 66,000 deceased patients in Europe every year.1 Computed tomography (CT) is mostly performed as a main standard imaging modality for acute TBI.2 The advantages of CT include large-scale availability and brusk preparation and study time.three , 4 CT reliably detects intracranial hemorrhages and contusions, bone pathology, cerebral edema, and incipient herniation.4 , v Limitations of noncontrast CT include a poor sensitivity for small-scale hemorrhages, diffuse axonal injury (DAI), arterial dissection, and vascular damage.half-dozen , 7 In item, in TBI patients with secondary progression of initially undetected small cognitive hemorrhages might cause a decline in the functional status of the patient, which motivates the need of early detection of these traumatic lesions.8
Several approaches for automatic detection to assist in visual analysis have been proposed, ranging from car learning on whole imagesix , 10 to figurer-assisted segmentations and morphological operations.11 , 12 In contempo years, machine learning gained broad attention in the field of medical imaging by detecting several abnormalities. Kumar et al.13 introduced a robust method by expanding multivariate generalized Gaussian distribution to a reproducing kernel Hilbert space for mixture modeling. According to the authors, this statistical learning method is robust compared with the previous studies that were highly sensitive to the outliers acquired by several errors, including motion and imaging artifacts. The method was examined using retinopathy and gastric and esophageal cancer datasets. After comparing the results with seven other methods, the statistical learning method showed a college accurateness.13 Another report published by Schlegl et al.xiv used generative adversarial networks to find the markers that provide information regarding the progression and treatment of a certain affliction. The method was examined on 8792 ii-dimensional retina images for the detection of retinal fluid and other retinal lesions. This technique was able to identify several abnormalities.
An culling technique is comparing findings with an average head CT that was starting time performed by Rorden et al.xv A template of subjects with a hateful age of 65 years was intended for the detection of encephalon impairment in stroke patients. Gillebert et al.16 used this template in comparison CT scans of stroke patients and showed detection of lesion boundaries with a sensitivity of 75%.
The aim of our study was to design and realize a simple and robust automated detection method for intracranial hemorrhages smaller than 10 mm in diameter in patients with TBI, as these are nearly hands missed on the CT browse images in an emergency OR situation. In some cases, secondary injury occurs if these lesions are non detected promptly. Our primary goal was to develop a method with the power to detect the presence of any type and size of intracranial hemorrhage and so that the emergency room physicians can telephone call in a neuroradiologist, a resource non likely available 24/seven in smaller regional hospitals.
2.
Materials and Methods
ii.1.
Data Acquisition
Nosotros retrospectively collected noncontrast CT scans from the radiology database of the Medisch Spectrum Twente, The netherlands (for details, see Appendix Table iii). Equally controls, we used CT scans from 30 patients (mean age 20.9, age range eighteen to 24 years, 13 males) with normal findings. 2 CT scans were made for reasons of persisting headache, whereas the other 28 patients were neurotrauma patients. We further collected CT images from ix TBI patients (mean age 49, age range 18 to 77 years, seven males) with hemorrhages smaller than ten mm in diameter on the original CT. Images were obtained from TOSHIBA (Toshiba Medical Systems Corporation, Tokyo, Nippon) and SIEMENS (Siemens Healthcare GmbH, Erlangen, Deutschland) CT scanners. CT scans with artifacts, a large corporeality of caput asymmetry, or patients with excessive head rotation were excluded. The Medical Ethics Committee Twente waived the demand for informed consent equally data were obtained as part of standard intendance.
two.2.
Preprocessing
Axial CT images of both the written report group and control group were reconstructed to obtain a directly position of the head using IntelliSpace software (IntelliSpace Portal vii.0, Philips, Eindhoven, The Netherlands). Irrelevant slices, divers equally slices to a higher place the skull and beneath the foramen magnum, were removed. Additionally, we reconstructed new slices with a slice thickness of ane mm (range: 1.00 to 1.06 mm) using the thin submillimeter slices of the original scans (for details, see Appendix Table 3). Finally, the surrounding structures, such as pilus, pillows, and sheets, around the encephalon visible on the original CT scans of the control grouping were removed by thresholding.
2.3.
Registration Framework
In medical imaging, paradigm registration is performed past spatially mapping different datasets. This procedure involves alignment of the moving image to the reference image, the fixed epitome.17 The goal of image registration is to detect a coordinate transformation that ensures the spatial alignment betwixt the fixed and moving paradigm. The transformation, which affects the registration between the fixed and moving image, is calculated in an iterative process, in which an optimizer minimalizes the value of the price function.18 In this study, registration techniques from Elastix software (Elastix Prototype Registration, The Netherlands) were applied.17 , 19 Using Elastix, both rigid torso registration (rigid and affine) and nonrigid torso registration (B-spline) were performed.19 The registration components are shown in Table 1.
Table ane
Registration steps of Elastix software with the specific components.nineteen
| Registration steps | Components used |
|---|---|
| Calibration infinite to reduce data complexity | Four levels, Gaussian smoothing with standard departure values of 4, ii, ane, and 0.v20 , 21 |
| Image sampler | Random coordinate sampler:18 , 19 , 22 3000 random coordinates |
| Interpolator | B-spline interpolator:xviii , 23 , 24 |
| • Linear interpolation: first iii levels of scale space | |
| • Cubic interpolation: terminal level of scale space | |
| Cost function | Mutual information:23 , 25 – 27 |
| • Number of histogram bins: 32 | |
| • Minimum required coordinate alignments to trigger the epitome registration: 150 | |
| Transformation | Rigid:22 , 24 translation and rotation |
| Affine: translation, rotation, scaling, and shearing | |
| B-spline:22 nonrigid transformation | |
| • Command point spacing: 20 mm | |
| Optimization | Adaptive stochastic slope descent:xviii , 28 |
| • Rigid and affine: 1500 iterations | |
| • B-spline transformation: 2500 iterations |
2.iv.
Creating the Average Image
The CT of one of the patients in the command group was selected every bit the fixed image. Subsequently, all 30 CT scans (including the stock-still image) were defined equally moving images and were aligned to the fixed image. After image registration, all images had equal size and number of slices. Following registration, an average paradigm from the 30 resulted images was generated. The boilerplate prototype was used for the detection of hemorrhages in the brains of TBI patients.
ii.5.
Detection of Small Hemorrhages
Since the skull shape, piece number, and size of the CT image of the TBI patient differed from the boilerplate paradigm, deformation of one of the images was required. Using epitome registration, the average prototype was deformed to obtain the same features as the CT images in the TBI group. This footstep ensured an average image and a CT epitome of each TBI patient with the same dimensions but different intracranial information. The concluding stride of this method was subtraction of the average image from the CT images of TBI patients. On the resulted images, simply the tissue that is not present in the average image volition remain. Since subtraction of the CT images is sensitive for the deviating densities, pocket-size hemorrhages can exist detected. The resulted images were visualized using a await-upwardly table of ITK Snap software.29
2.6.
Examining the Hemorrhages
A neuroradiologist and a radiologist in training were independently asked to examine the intensity differences in the resulted images and make up one's mind the regions of hemorrhage. The clinical reports of the original CT scans were used every bit the golden standard. The findings of the neuroradiologist and radiologist in grooming were compared with the clinical notes. False positives and negatives were noted, and the sensitivity was calculated.
3.
Results
As shown in Fig. i(a), the skull of each patient in the control group was altered to fit the skull shape of the stock-still image (white pointer), while the intracranial details differed. From these three-dimensional images, an average image was generated where each voxel in the average paradigm represented the boilerplate of 30 voxels at that indicate [Fig. 1(b)].
Fig. one
(a) The fixed prototype (white arrow) with 29 resulted images after image registration of the CTs of patients in the control group. All images are obtained from the aforementioned slice in the scan. The heads of the patients on the CT images have similar skull shapes with different detailed information in the brain. From these images, (b) an average image is generated.
The information of the boilerplate image of patients in the control grouping were compared with the CT images of TBI patients, and the hemorrhages were automatically visualized. According to the clinical notes, 67 lesions were detected where the type and the region of hemorrhage differed per patient (Table 2). The neuroradiologist and the radiologist in grooming missed the aforementioned two hemorrhages (sensitivity of 97%) that were present in the temporal lobe about the os. The neuroradiologist found 3 false positives, and the radiologist in preparation detected two of these simulated positives. Figure 2 shows one of the true positive slices of the original CT of a TBI patient and the resulting image afterwards automatic detection. The original CT image shows ii small hemorrhages at the edge of the lateral ventricles. After applying the automatic detection technique, the hemorrhages are highlighted in green/yellow. Besides, the edges of the skull were visualized as a high-intensity region. Since our primary goal was detection of small intracranial hemorrhages, the skull was ignored in the visual assessment. Withal, to maintain the anatomical orientation in the resulted image, skull stripping was not performed. For one patient, our method showed a cerebral hemorrhagic contusion that was originally missed. In the resulted images of this TBI patient, the radiologists detected a minor asymmetry in the temporal lobe, which was not noted in the clinical data of the patient. After re-examining the original CT image of the patient, a cerebral contusion in the temporal lobe was detected. Figure 3(a) shows the original CT image of the patient with bilateral minor subdural hemorrhages forth Meckel's cave (green arrows) and small initially undetected hemorrhagic contusion in the left temporal expanse (scarlet arrow).
Tabular array 2
Type and location of hemorrhages in the brain of TBI patients.
| Patients | Number of hemorrhages | Type and location of the hemorrhages |
|---|---|---|
| Patient i | 12 | Subarachnoid: left temporal |
| Cerebral contusions: left parietal and bilateral frontal and temporal | ||
| DAI: bilateral randomly nowadays | ||
| Patient 2 | two | Cognitive contusions: left temporal |
| Patient 3 | xi | DAI: corpus callosum |
| Subarachnoid: right temporal | ||
| Intraventricular: bilateral | ||
| Patient 4 | 16 | DAI: left parietal, temporal, and basal ganglia |
| Subarachnoid: bilateral frontal | ||
| Cerebral contusions: left temporal | ||
| Patient v | 3 | DAI: near the right ventricle |
| Subarachnoid: right temporal | ||
| Patient 6 | 14 | Cerebral contusions: bilateral frontal |
| DAI: bilateral randomly present | ||
| Subdural: left frontoparietal | ||
| Patient 7 | 2 | Subdural: bilateral temporal |
| Patient 8 | 1 | Subarachnoid: correct frontotemporal |
| Patient nine | 6 | Cerebral contusions: left parietal |
| Subarachnoid: bilateral occipital | ||
| DAI: bilateral randomly nowadays |
Fig. 2
(a) A slice of the original CT image of a TBI patient with two small regions of hemorrhage (greenish arrows). (b) Later applying the automated detection method on the original paradigm, the two hemorrhages were easily distinguishable from the healthy tissue. However, high intensities were also detected at the edges of the skull.
Fig. iii
(a) A slice of the original CT image of a TBI patient with bilateral small subdural hemorrhages along Meckel's cave (green arrows) and an originally missed cognitive hemorrhagic contusion (red arrow). (b) Afterward applying the automatic detection method on the original image, the ii subdural hemorrhages were detected, while the hemorrhagic contusion was partially detected. All the same, the hemorrhagic contusion was non noted in the clinical notes of the patient.
The final images of TBI patients besides showed high-intensity regions along the cerebellar tentorium, pons, and around the lateral ventricles. Furthermore, the radiologists had no difficulties in discerning dense vessels and calcifications from hemorrhages in the obtained images.
four.
Discussion
In this pilot report, we mimicked the visual analysis of the neuroradiologists with an automated detection method past easily identifying small cerebral hemorrhages in TBI patients using a computer-generated "average brain" from 30 command patients. From 67 lesions, the automated detection method missed only two lesions that were present in the temporal fossa about the bone. The middle temporal fossa is highly variable in shape and size per individual. These variations were observed on the original CT scans of 30 patients in the control group. Since only 30 young patients were included in the control group, the average image will be affected by these individual differences. The shape of the fossa media is partially race dependent; therefore, development of several templates based on race may solve a part of the trouble.
The automated method detected a hemorrhage that was initially missed by a resident. After detecting the hemorrhage by our automatic method, 2 neuroradiologists and a radiologist in training critically re-examined the original CT paradigm. The three physicians were able to ostend the presence of hemorrhage on the original CT image.
In most hospitals, the test of the (neuro)radiologist and the resulting clinical notes are used equally a gilt standard for TBI patients. Since autopsy is usually not an option and our work is a retrospective pilot written report, we used the visual analysis of our neuroradiologists as the gold standard.
Al-Ayyoub et al.9 examined CT images of 74 patients classified into normal, epidural, subdural, and intraparenchymal hemorrhages. They practical several techniques, such every bit image preprocessing, segmentation, region growing, characteristic extraction, and classification, resulting in an accuracy of 100% for the detection of hemorrhages. Shahangian and Pourghassem30 performed automatic brain hemorrhage division and nomenclature algorithm based on weighted grayscale histogram feature in a hierarchical nomenclature structure. Detection accuracies of epidural, subdural, and intracerebral hemorrhages were 96.xv%, 94.87%, and 95.96%, respectively. Liu et al.10 used auto learning for the detection of cerebral hemorrhages in neurotrauma patients with a detection accuracy of 80%. The authors included mostly large hemorrhages that were not easily overlooked on the original CT images. In dissimilarity to these studies, we focused on hemorrhages smaller than 10 mm in bore. Comparison our airplane pilot results to the results of these studies, our method generally falls in the aforementioned accurateness range. Withal, it is noteworthy that we accept only included 9 CT scans in the TBI group and 30 CT scans in the control group. Additionally, our method did not require whatever segmentation of the skull, ventricles, or cerebrospinal fluid (CSF).
The original CT images of the brain are ofttimes initially read by a radiology resident or sometimes by a nonradiologist. Our automatic method may assist the radiology resident and other physicians in the detection of small traumatic hemorrhagic lesions. Since our method visualizes normal tissue with a blue color and other findings with other colors, the physicians may quickly know where to focus on the image. Automatic detection tin save fourth dimension by straight distinguishing lesions from healthy brain areas. Moreover, it may reduce the interobserver variation between radiologists.
In this study, nosotros compared the CT scans of a relative young command grouping with TBI patients of different age groups. Since aging changes the brain, sometimes reducing it in size,31 information technology is more precise to compare patients with age-matched control groups. This may reduce misdetection and may crusade less deformation of the average paradigm to correctly align to the TBI images.
To initiate the epitome registration, a reference prototype was required to align the images. In this work, a CT image of a patient in the control group was selected as the fixed image. Since in that location was one fixed prototype and 30 moving images, the fixed image had a college influence on the resulted images. An approach for eliminating the bias is a groupwise epitome registration where all images are aligned to a common space. Applying a groupwise registration, the information of all images will have an equal influence on the boilerplate image.32 – 34 Recently, Elastix introduced the method of Huizinga et al.32 in its database for groupwise registration of MRI images. To our knowledge, using Elastix software, this arroyo has not been applied on CT images.
Several registration steps were applied to obtain a good merchandise-off between quality and speed. The goal was to create an average brain CT from several CT images to include the normal variability in brain tissue. Subsequently applying rigid and affine transformation, the details of the encephalon tissue in the resulted images looked nearly the same as the fixed image. Since the images were obtained from different patients, rigid transformation was not able to correctly align the images and to include the details of the moving images in the results. To perform the registration more precisely, nonrigid B-spline transformation was applied. The B-spline method uses a grid with automatically placed control points on the vertices of the grid squares that are overlaid on the image. The spacing of the control points determines the deformation. Large spacing describes a global deformation, whereas small spacing defines a local deformation. To match the small structures in the images, different spacing values were tested. As larger values of xxx and 40 mm resulted in a mismatched registration prototype, a spacing of 20 mm between two control points was preferred for the CT registration. The results of nonrigid transformation showed a good brain overlap that was confirmed by visual inspection. Using nonrigid transformation, the tissue density may change locally, especially in the transition regions from white to grey matter, CSF, and parenchyma. However, these density differences are nonetheless lower than the density of a hemorrhage. The densities of CSF and brain tissue are between 3 and 40 HU, and the density of blood is generally higher than 60 HU.35 , 36 The automatic method is developed for detecting intracranial hemorrhages and will not be afflicted past other focal lesions, such every bit white matter lesions.
The resulted images of TBI patients showed higher intensity regions around the lateral ventricles. The reason for these high-intensity regions may be the large differences in contrast between the ventricles and the brain parenchyma. Since the shape and position of the ventricles differ per private, voxel resampling to some other brain may cause this misdetection. Furthermore, some TBI patients develop a midline shift that causes a displacement of the ventricles. This displacement can have an influence during the comparison of TBI scans to the control group and may crusade the false positives in the regions around the ventricles. The neuroradiologist and radiologist in training did not place the high-intensity regions around the ventricles as pathologic. It is, nonetheless, possible that other physicians may translate these areas every bit simulated positives. A possible solution for this problem could be the sectionalization of the ventricles earlier paradigm registration.
Similar problematic areas were regions around the pons and/or cerebellar tentorium. A reason for these areas being highlighted could exist artifacts on the original CT images that are difficult to distinguish from hemorrhage. Other reasons for the high intensities in the cerebellum could be subarachnoid blood or a effect of our method. Subarachnoid blood is usually chop-chop diluted by the CSF, and then it may not be present on a follow-up report. For these types of findings, prospective research with a follow-up MRI could provide additional information. Using this data, we can examine if these high-intensity regions are caused by our method or by other factors.
v.
Conclusion
Nosotros have introduced an automatic detection method for the detection of minor traumatic encephalon hemorrhages in TBI patients using a computer-generated boilerplate CT. Our automated detection method showed encouraging pilot results and a good correlation with the visual analysis of the neuroradiologists. The automatic comparison of individual CT scans with the computed average may assist the physicians in early detection of small-scale hemorrhages.
Appendices
Appendix
The patients obtained a CT scan with varied settings in some cases. A overview of these settings are listed in Table 3. Since the piece thickness differed in some scans, the images were reconstructed to equalise the thicknesses.
Tabular array 3
Size and setting information of the CT images of both control and TBI group.
| CT features | Control group | TBI group |
|---|---|---|
| Corporeality of slices | 133 to 171 | 138 to 154 |
| Peak kilovoltage (kVp) | 120 kV | 120 kV |
| 10-ray tube current | 124 to 452 mA | 187 to 518 mA |
| Exposure | 124 to 410 mAs | 187 to 470 mAs |
| Pitch | 0.55 to 0.66 | 0.55 to 0.66 |
| Total collimation width | 6 to 38.4 mm | 6 to 16 mm |
| Pixel spacing | [0.3899;0.3899] to [0.4814;0.4814] | [0.4187;0.4187] to [0.5859;0.5859] |
| Matrix size | [512 512] | [512 512] |
| Piece thicknesses before reconstruction | 0.5 to 1 mm | 0.5 to ane mm |
| Slice thicknesses afterwards reconstruction | 1 to 1.03 mm | ane to 1.06 mm |
Disclosures
All authors declare that they have no financial or conflicts of interest that may affect the research reported in the enclosed paper.
Acknowledgments
We are grateful to Dr. Jurrit Hof and Dr. Farhood Mojtahedi for the cess of the images and to Dr. Harold Hom for helping us in finding suitable TBI patients. We would also similar to thank Dr. Ir. Stefan Klein for giving u.s.a. information concerning Elastix software.
References
1.
A. H. P. Willemse-van Son et al., "Prognostic factors of long-term functioning and productivity after traumatic brain injury: a systematic review of prospective cohort studies," Clin. Rehabil., 21 (11), 1024 –1037 (2007). https://doi.org/10.1177/0269215507077603 Google Scholar
6.
H. L. Fred, "Drawbacks and limitations of computed tomography: views from a medical educator," Tex. Eye Inst. J., 31 (4), 345 –348 (2004). THIJDO 0730-2347 Google Scholar
ix.
Grand. Al-Ayyoub et al., "Automatic detection and classification of brain hemorrhages," WSEAS Trans. Comput., 12 (x), 395 –405 (2013). Google Scholar
10.
R. Liu et al., "Hemorrhage slices detection in brain CT images," in 19th Int. Conf. on Pattern Recognition, 2008 (ICPR 2008), one –4 (2008). Google Scholar
12.
1000. Matesin, South. Loncaric and D. Petravic, "A rule-based approach to stroke lesion analysis from CT brain images," in Proc. 2nd Int. Symp. on Paradigm Signal Processing and Analysis 2001 (ISPA 2001), 219 –223 (2001). https://doi.org/10.1109/ISPA.2001.938631 Google Scholar
13.
N. Kumar et al., "Kernel generalized-Gaussian mixture model for robust aberration detection," in Int. Conf. on Medical Image Computing and Figurer-Assisted Intervention (MICCAI 2017), 21 –29 (2017). Google Scholar
18.
Southward. Klein, One thousand. Staring and J. P. West. Pluim, "Evaluation of optimization methods for nonrigid medical paradigm registration using common data and B-splines," IEEE Trans. Image Process., 16 (12), 2879 –2890 (2007). https://doi.org/10.1109/TIP.2007.909412 IIPRE4 1057-7149 Google Scholar
24.
J. Ashburner and K. J. Friston, "Spatial transformation of images," Hum. Encephalon Funct., 43 –58 (1997). Google Scholar
25.
P. Thévenaz, M. Unser and P. Thevenaz, "Optimization of mutual information for multiresolution epitome registration," IEEE Trans. Image Process., 9 (12), 2083 –2099 (2000). https://doi.org/10.1109/83.887976 IIPRE4 1057-7149 Google Scholar
thirty.
B. Shahangian and H. Pourghassem, "Automatic encephalon hemorrhage sectionalization and classification algorithm based on weighted grayscale histogram feature in a hierarchical classification structure," Biocybern. Biomed. Eng., 36 (1), 217 –232 (2016). https://doi.org/10.1016/j.bbe.2015.12.001 Google Scholar
35.
C. Claussen et al., Computed Tomography and Magnetic Resonance Tomography of Intracranial Tumors: A Clinical Perspective, second ed.Springer-Verlag, Berlin, Heidelberg (1989). Google Scholar
36.
L. A. Cala et al., "Encephalon density and cerebrospinal fluid space size: CT of normal volunteers," Am. J. Neuroradiol., 2 (one), 41 –47 (1981). Google Scholar
Biography
Liza Afzali-Hashemi received her MSc degree in technical medicine from the University of Twente, The netherlands, in 2017. Currently, she is working as a technical physician in the Neurophysiology Department, the Medisch Spectrum Twente, Kingdom of the netherlands. Her research interests include noncontrast CT, perfusion CT, transcranial Doppler, and intercranial pressure.
Marieke Hazewinkel studied medicine in Rotterdam, The Netherlands. She completed her radiology grooming at Noordwest Ziekenhuisgroep, Alkmaar, and the VU Medical Centre, Amsterdam, The Netherlands. Later on a fellowship on neuro-/head and neck radiology at the VUMC, she went on as a neuro/head and neck radiologist at Medisch Spectrum Twente, Holland, where she has been working since 2012.
Marleen C. Tjepkema-Cloostermans obtained her PhD from the Clinical Neurophysiology Group, the University of Twente, Enschede, Holland. Currently, she is working as a technical physician in the Neurology and Clinical Neurology Section, the Medisch Spectrum Twente, The Netherlands. Her enquiry mainly focuses on the comeback of neuromonitoring of patients in the intensive care unit, ranging from comatose patients after cardiac abort to patients with traumatic brain injury.
Michel J. A. Thousand. van Putten studied medicine in Leiden and applied physics in Delft, The Netherlands. In 2000, he received his PhD in applied physics from Delft University of Technology and became a lath-certified neurologist in the same year. He is a professor of clinical neurophysiology at the MIRA Institute for Biomedical Engineering and Technical Medicine, Academy of Twente, Kingdom of the netherlands. His main research interests include neuromonitoring, epilepsy, and ischemia.
Cornelis H. Slump received his MSc degree in electric engineering from Delft Academy of engineering science, The netherlands, in 1979. In 1984, he obtained his PhD in physics from the Academy of Groningen, The Netherlands. In 1989, he joined the Department of Electrical Engineering science, Academy of Twente. In 1999, he was appointed as a full professor of signal processing. His research interest is in paradigm analysis every bit part of medical imaging.
olivareswashe1945.blogspot.com
Source: https://www.spiedigitallibrary.org/journals/journal-of-medical-imaging/volume-5/issue-02/024004/Detection-of-small-traumatic-hemorrhages-using-a-computer-generated-average/10.1117/1.JMI.5.2.024004.full
0 Response to "Groupwise 2012 9509 Please Enter Your Password Again"
Postar um comentário