ABSTRACT
OBJECTIVE: To retrospectively assess the effect of CT slice thickness on the reproducibility of radiomic features (RFs) of lung cancer, and to investigate whether convolutional neural network (CNN)-based super-resolution (SR) algorithms can improve the reproducibility of RFs obtained from images with different slice thicknesses. MATERIALS AND METHODS: CT images with 1-, 3-, and 5-mm slice thicknesses obtained from 100 pathologically proven lung cancers between July 2017 and December 2017 were evaluated. CNN-based SR algorithms using residual learning were developed to convert thick-slice images into 1-mm slices. Lung cancers were semi-automatically segmented and a total of 702 RFs (tumor intensity, texture, and wavelet features) were extracted from 1-, 3-, and 5-mm slices, as well as the 1-mm slices generated from the 3- and 5-mm images. The stabilities of the RFs were evaluated using concordance correlation coefficients (CCCs). RESULTS: The mean CCCs for the comparisons of original 1 mm vs. 3 mm, 1 mm vs. 5 mm, and 3 mm vs. 5 mm images were 0.41, 0.27, and 0.65, respectively (p < 0.001 for all comparisons). Tumor intensity features showed the best reproducibility while wavelets showed the lowest reproducibility. The majority of RFs failed to achieve reproducibility (CCC ≥ 0.85; 3.6%, 1.0%, and 21.5%, respectively). After applying the CNN-based SR algorithms, the reproducibility significantly improved in all three pairings (mean CCCs: 0.58, 0.45, and 0.72; p < 0.001 for all comparisons). The reproducible RFs also increased (36.3%, 17.4%, and 36.9%, respectively). CONCLUSION: The reproducibility of RFs in lung cancer is significantly influenced by CT slice thickness, which can be improved by the CNN-based SR algorithms.
Subject(s)
Learning , Lung Neoplasms , Lung , Retrospective StudiesABSTRACT
Objective@#The study is to assess the accuracy and reliability of three-dimensional simulated magnetic resonance imaging with silicone-excitation SPACE (sampling perfection with application optimized contrast using different flip angle evolutions) sequence for estimating implant volume.@*Methods@#(1) MRI examinations of 10 silicone implants (Wuhan Tongji Hospital from October 2018 to December 2018) were performed with T2, H2O-excitation SPACE sequence (T2-spc-H2O) and silicone-excitation SPACE sequence (T2-spc-Silicone) to find the most accurate method to estimate implant volume by ITK-SNAP. The effect of implant deformation and slice thickness of T2-spc-Silicone on volume measurement were investigated. (2) 13 normal patients and 6 patients with implant complications (Wuhan Tongji Hospital from March 2017 to May 2019) were enrolled for testing the accuracy and reliability of T2-spc-Silicone for volume measurement in vivo. The data were analyzed using Prism 8.0 software. The paired student t-test was used to compare the difference of two groups. One-way ANOVA was used to compare the difference of multiple groups. P<0.05 was considered statistically significant.@*Results@#The absolute volume differences of T2, T2-spc-H2O, T2-spc-Silicone were (42.19±2.31) ml, (23.27±1.55) ml and (6.28±1.22) ml. The absolute volume differences of T2-spc-Silicone group was significantly less than T2-spc-H2O and T2 group in vitro(F=195.3, P<0.001). No significant difference(F=1.36, P=0.22)was shown between the normality group and the deformation group for estimating the volume of implants with the slice thickness of SPACE increased from 0.5 mm×0.5 mm×0.5 mm to 5.0 mm×5.0 mm×5.0 mm. Besides, the slices thickness of SPACE from 0.5 mm×0.5 mm×0.5 mm to 5.0 mm×5.0 mm×5.0 mm did not significantly affect the accuracy of volume measurement of the implants in deformation state(F=1.22, P=0.29). The measurement error of SPACE was (8.82±0.99) ml in normal patients. Moreover, there was no significant difference between measured volume[(226.4±12.76)ml] and actual volume of implants[(225.9±11.94) ml](t=0.31, P=0.76)in patients with implant complications. The result showed excellent intraobserver reliability (ICC=0.997) and internal consistency ranged from 0.986 to 0.997 (P<0.001).@*Conclusions@#The method to measure implant volume by silicone-excitation SPACE sequence had desirable accuracy and reliability. The deformation of the implant and the slice thickness of the SPACE sequence did not exhibit a significant effect on the accuracy of volume measurement.
ABSTRACT
PURPOSE: To compare the orbital volume calculated from various slice thickness facial computed tomography scans using a semi-automated computer program. METHODS: Axial and coronal scans of 2, 2.5, 3 mm slice thickness facial computed tomography scans were used to measure the orbital volume. The cross-sectional area was determined from each slice using a semi-automated computer program (MATLAB 2009a®, MathWorks, Inc., Natick, MA, USA), and then the volume was calculated from serial reconstruction of the cross sections. RESULTS: The measured value in the 2 mm images was 33.14 ± 2.37 cm³ in the right orbit and 34.32 ± 2.60 cm³ in the left orbit for the axial scans, and 35.54 ± 3.58 cm³ in the right orbit and 34.96 ± 4.05 cm³ in the left orbit for the coronal scans. In the 2.5 mm images, the values were 33.28 ± 3.35 cm³ in the right orbit and 33.73 ± 4.10 cm³ in the left orbit for the axial scans, and 35.24 ± 3.98 cm³ in the right orbit and 35.10 ± 3.93 cm³ in the left orbit for the coronal scans. In the 3 mm images, the values were 33.23 ± 2.70 cm³ in the right orbit and 33.39 ± 2.69 cm³ in the left orbit for the axial scans, and 33.20 ± 3.64 cm³ in the right orbit and 32.95 ± 3.45 cm³ in the left orbit for the coronal scans. In the 3 mm image, there was not a significant difference in the calculated volume between the axial and coronal scans (p(3mm) = 0.62). CONCLUSIONS: Because there is no difference in the results of the orbital volumetric measurements between three other slice thicknesses in the axial scan, using axial scan images with a computer program that semi-automatically calculates orbital volume is useful. In addition, the volume measured by thick slice images has more reproducibility than the volume measured by thin slice images.
Subject(s)
OrbitABSTRACT
The accuracy of technique adopted for Forensic age diagnostics of young adults and adolescents especially in case of livings lies in the standardization of the technical parameters used. The emerging radiological techniques, when used in standardized way may minimize the possibilities of misinterpretation, as it has been practically shown in present study. CT scans of 100 live subjects were performed on 16-slice (Siemen’s Sensation 16) CT scan machine and the volumetric data acquired was reconstructed into five separate sets of slice thickness for each one of the subjects included in the study and the ossification status for each set of slice thickness was determined for all the subjects separately. The results are almost identical while evaluating ossification stages from 1 and 2 mm thick slice data but the differences are found in the ossification stages when evaluated using 3 mm, 5 mm and 7 mm slice thickness as compared those found in 1 and 2 mm slice thickness. It was concluded that by increasing slice thickness the rate of error-nous interpretation are also increasing. Thus, the minimum reliable thickness to produce high resolution scans in order to get maximum accuracy is 2 mm for staging medial clavicular ossification from CT scan and the reconstruction should be done using kernel (filter) B60F at window setting osteo (1500/450HU).
Subject(s)
Adolescent , Clavicle/anatomy & histology , Clavicle/diagnostic imaging , Epiphyses/anatomy & histology , Epiphyses/diagnostic imaging , Forensic Medicine , Humans , Organ Size , Osteogenesis , Radiation Dosage , Tomography, X-Ray Computed , Young AdultABSTRACT
Purpose: To find the optimal slice thickness (Δτ) setting for patient registration with kilovoltage cone-beam CT (kVCBCT) on the Varian On Board Imager (OBI) system by investigating the relationship of slice thickness to automatic registration accuracy and contrast-to-noise ratio. Materials and method: Automatic registration was performed on kVCBCT studies of the head and pelvis of a RANDO anthropomorphic phantom. Images were reconstructed with 1.0 ≤ Δτ (mm) ≤ 5.0 at 1.0 mm increments. The phantoms were offset by a known amount, and the suggested shifts were compared to the known shifts by calculating the residual error. A uniform cylindrical phantom with cylindrical inserts of various known CT numbers was scanned with kVCBCT at 1.0 ≤ Δτ (mm) ≤ 5.0 at increments of 0.5 mm. The contrast-to-noise ratios for the inserts were measured at each Δτ. Results: For the planning CT slice thickness used in this study, there was no significant difference in residual error below a threshold equal to the planning CT slice thickness. For Δτ > 3.0 mm, residual error increased for both the head and pelvis phantom studies. The contrast-to-noise ratio is proportional to slice thickness until Δτ = 2.5 mm. Beyond this point, the contrast-to-noise ratio was not affected by Δτ. Conclusion: Automatic registration accuracy is greatest when 1.0 ≤ Δτ (mm) ≤ 3.0 is used. Contrast-to-noise ratio is optimal for the 2.5 ≤ Δτ (mm) ≤ 5.0 range. Therefore 2.5 ≤ Δτ (mm) ≤ 3.0 is recommended for kVCBCT patient registration where the planning CT is 3.0 mm
ABSTRACT
PURPOSE: This study was to search the optimal slice thickness of computed tomography (CT) in an intensity modulated radiation therapy plan through changing the slice thickness and comparing the change of the calculated absorbed dose with measured absorbed dose. MATERIALS AND METHODS: An intensity modulated radiation therapy plan for a head and neck cancer patient was done, first of all. Then CT with various ranges of slice thickness (0.125~1.0 cm) for a head and neck anthropomorphic phantom was done and the images were reconstructed. The plan parameters obtained from the plan of the head and neck cancer patient was applied into the reconstructed images of the phantom and then absorbed doses were calculated. Films were inserted into the phantom, and irradiated with 6 MV X-ray with the same beam data obtained from the head and neck cancer patient. Films were then scanned and isodoses were measured with the use of film measurement software and were compared with the calculated isodeses. RESULTS: As the slice thickness of CT decreased, the volume of the phantom and the maximum absorbed dose increased. As the slice thickness of CT changed from 0.125 to 1.0 cm, the maximum absorbed dose changed ~5%. The difference between the measured and calculated volume of the phantom was small (3.7~3.8%) when the slice thickness of CT was 0.25 cm or less. The difference between the measured and calculated dose was small (0.35~1.40%) when the slice thickness of CT was 0.25 cm or less. CONCLUSION: Because the difference between the measured and calculated dose in a head and neck phantom was small and the difference between the measured and calculated volume was small when the slice thickness of CT was 0.25 cm or less, we suggest that the slice thickness of CT should be 0.25 cm or less for an optimal intensity modulated radiation therapy plan.
Subject(s)
Humans , Head , Head and Neck Neoplasms , NeckABSTRACT
PURPOSE: To evaluate clinical usefulness of facial soft tissue thickness measurement using 3D computed tomographic images. MATERIALS AND METHODS: One cadaver that had sound facial soft tissues was chosen for the study. The cadaver was scanned with a Helical CT under following scanning protocols about slice thickness and table speed; 3 mm and 3 mm/sec, 5 mm and 5 mm/sec, 7 mm and 7 mm/sec. The acquired data were reconstructed 1.5, 2.5, 3.5 mm reconstruction interval respectively and the images were transferred to a personal computer. Using a program developed to measure facial soft tissue thickness in 3D image, the facial soft tissue thickness was measured. After the ten-time repeation of the measurement for ten times, repeated measure analysis of variance (ANOVA) was adopted to compare and analyze the measurements using the three scanning protocols. Comparison according to the areas was analyzed by Mann-Whitney test. RESULTS: There were no statistically significant intraobserver differences in the measurements of the facial soft tissue thickness using the three scanning protocols (p>0.05). There were no statistically significant differences between measurements in the 3 mm slice thickness and those in the 5 mm, 7 mm slice thickness (p>0.05). There were statistical differences in the 14 of the total 30 measured points in the 5 mm slice thickness and 22 in the 7mm slice thickness. CONCLUSION: The facial soft tissue thickness measurement using 3D images of 7 mm slice thickness is acceptable clinically, but those of 5 mm slice thickness is recommended for the more accurate measurement.